# Signals - Full Content
> Complete public markdown content for returnsignals.com.
> For a summary, see https://www.returnsignals.com/llms.txt
---
# Signals
> The AI store associate for every customer, over iMessage. Signals starts a personal, one-to-one conversation with every customer after delivery, helping them find what to buy next while giving them the same high-quality support they’d expect from your best store associate.
Signals creates iMessages that convert users and keep them coming back. The AI agent uses a brand's catalog and policies to answer like the best associate, helping users choose with confidence before checkout and following up after delivery to bring customers back.
The main value is additional revenue for DTC brands: helping new customers buy by answering the questions that block checkout, then increasing LTV from existing customers through cross-sell, reorder, subscription-save, and repeat-purchase conversations.
Signals is backed by Y Combinator and is live with multiple apparel, cosmetics, and pet food brands.
## What Signals Does
Signals is a lifecycle conversation platform for ecommerce brands. Shoppers can text before they buy, customers can reply after delivery, and the same two-way iMessage/SMS/RCS thread turns uncertainty into checkout confidence, exchanges, repeat purchases, product feedback, and customer intelligence.
Signals is strongest for brands where the customer receives a physical good, customer LTV is above $100, and there is a natural repeat-purchase, replenishment, or cross-sell opportunity.
## Product Promise
- persistent channel
- 51% more repeat purchases
- Integrates with your stack
Signals supports common customer conversations before purchase, after delivery, during support moments, and before the next order:
1. **Pre-purchase fit** - A shopper is between sizes. Signals uses product and policy context to recommend the right choice, answer exchange and return confidence questions, and keep the thread open after delivery.
2. **Contextual cross-sell** - A post-delivery care question becomes a confident next purchase. Signals recommends the right related product, answers variant and size questions, and can assist checkout with saved customer context.
3. **Subscription save** - A customer is unsure whether a pet food product is working. Signals gives transition guidance, detects churn risk or buying intent, and moves the customer toward a better plan or replenishment path.
## Integrations
Signals integrates with iMessage, RCS/SMS, WhatsApp, Shopify, Gorgias, Listrak, EasyPost, Slack, and 50+ other commerce and support systems.
Signals can work alongside the systems brands already use:
- SMS marketing tools keep running broadcasts, flows, and list growth.
- Helpdesks can remain systems of record for escalations and complex cases.
- Return portals can remain operational backends for logistics and policy workflows.
- Signals can also own the customer-facing support and returns experience directly in the thread.
- Signals becomes the relationship layer that starts and continues the one-to-one customer thread from first question to repeat order.
## From First Question To Repeat Order
Signals keeps one iMessage conversation between shopper and brand open across the moments that decide sales, support, and repeat revenue.
1. **Question** - Shoppers text Signals product questions that help them make a buying decision.
2. **Order** - New orders can be made through Signals threads, and existing orders get support.
3. **Delivery** - Signals checks in with customers after delivery to make sure they have an excellent experience.
4. **Next purchase** - Signals continues building relationships with customers, cross-selling, and driving repeat revenue.
## Eight Campaigns
Signals keeps the before-purchase and after-delivery moments connected, so sales, support, returns, and retention do not become separate workflows.
### Before Purchase: Inbound Text Us
Answer the questions that decide whether a shopper buys now or leaves.
- **Fit and size concierge** - Trigger: a shopper is between sizes. Outcome: use product, order, and policy context to recommend the confident choice.
- **Product matching** - Trigger: a shopper asks what goes with what. Outcome: turn styling, bundle, and comparison questions into assisted revenue.
- **Availability and restock** - Trigger: a desired variant is sold out. Outcome: suggest in-stock alternatives or capture restock intent in the same thread.
- **Shipping and policy confidence** - Trigger: the blocker is operational. Outcome: answer delivery, exchange, return, and gift questions before checkout.
### After Delivery: Outbound Concierge
Follow up when the customer has the product and the next signal is strongest.
- **Delivery check-in** - Trigger: the box arrives. Outcome: ask how the product is working, collect feedback, and keep the thread alive.
- **Exchange rescue** - Trigger: fit, color, or expectation is off. Outcome: recommend available alternatives before the customer disappears into a return portal.
- **Contextual cross-sell** - Trigger: a customer says what they need next. Outcome: suggest the right product from the live catalog with purchase context in hand.
- **Replenishment and save** - Trigger: usage, cadence, or churn risk appears. Outcome: turn product experience into reorder timing, subscription changes, or a better plan.
### System Principles
- **Shared context** - Pre-purchase and post-delivery playbooks use the same customer context.
- **Human handoff** - Unusual conversations escalate without losing thread history.
- **Measurement** - Revenue, support outcomes, and customer requests stay tied to the conversations that created them.
## Case Study
Quaker Marine used Signals to turn post-delivery check-ins into 51% more repeat purchases among customers who replied.
In a randomized experiment across 1,910 customers:
- Treatment customers repurchased 16% more often within 3 weeks.
- Customers reached through iMessage and RCS replied at about 58%.
- Customers who engaged in conversation had 51% higher repeat purchase rates than the control group.
Kevin McLaughlin, owner of Quaker Marine Supply Co., said: "My philosophy has always been to make a customer, not a sale. Signals gives us a way to do online what the best retailers have always done in person."
Read case studies: /blog/tag/case-study
## One Thread For Sales, Support, And Returns
Signals is built for the one-to-one buyer conversation: answering questions, resolving support issues, managing returns and exchanges, and handing off to existing systems when that is what the team wants.
| Dimension | Your stack today | With Signals |
| ----------------------- | --------------------------------------------------------------------------- | -------------------------------------------------------------------------------- |
| How it starts | Campaign calendar, support ticket, or portal visit | A customer texts, or Signals checks in when the moment matters |
| When it shows up | Before purchase as a promotion, or after the customer has already escalated | Before checkout, after delivery, during an exchange, and before the next order |
| What the customer feels | A broadcast, ticket queue, or self-serve workflow | A personal thread that works whether they reply now or six hours later |
| What your team learns | Clicks, tickets, return reasons, and campaign attribution | Fit blockers, product confusion, sales intent, issue context, and reorder timing |
| Where it resolves | Across disconnected tools | In conversation first, with handoff to the right system when needed |
## Operating Surface
Signals lets teams manage every buyer question in one operating surface. Most inbound and post-delivery conversations resolve fully with AI. When something unusual appears, it escalates to a human in the same thread, and the team can step in without losing context.
Conversation states include automated, escalated, and blocked.
## Frequently Asked Questions
### How often do customers text Signals?
Signals sees about 2.5 texts per 1,000 impressions, with roughly a 40/60 support-to-sales mix. About 20% of sales conversations result directly in a sale.
### How does Signals drive repeat purchases?
Signals creates additional revenue for DTC brands in two ways: it helps new customers buy by answering the questions that block checkout, then it increases LTV from existing customers through reorders, cross-sells, upsells, and subscription saves. Brands see 50%+ more repeat purchases from engaged customers.
### What kind of revenue impact should we expect?
Results vary by brand, but typical outcomes include more new-customer purchases from answered questions, 50%+ repeat purchase lift among engaged customers, 4-6x higher conversion than other text campaigns, and measurable cross-sell and upsell revenue. The goal is to prove about 10x ROI through a refundable 4-week pilot that runs as a randomized control trial.
### Who is Signals built for?
Signals is built for DTC brands and ecommerce companies that want to turn one-time buyers into repeat customers. It is best for brands selling physical products with natural reorder or cross-sell opportunities. It is not a fit for teams looking for a bulk SMS marketing tool.
### How is this different from email or SMS marketing?
Marketing campaigns broadcast offers to a list. Signals starts one-to-one conversations with customers at the moment they are most engaged. Customers reply, share feedback, and buy again inside that conversation. That is why engagement rates reach 61%, versus single-digit rates for email and traditional SMS.
### Why iMessage instead of email or app notifications?
iMessage is where customers already have real conversations. Messages are read in minutes, customers can reply instantly, share photos, and pick up the thread later. There is no app to install and no inbox to compete with.
### Does proactive outreach feel spammy to customers?
The opposite. A well-timed check-in after delivery feels like genuine care, not marketing. Customers respond because the outreach is personal and helpful. About 15% of customers voluntarily share future buying intents during the conversation without being asked.
### How does Signals handle subscription retention?
Signals detects early churn signals like product buildup, skipped orders, or dissatisfaction during post-purchase conversations. Instead of waiting for a cancellation, it can adjust delivery frequency, offer alternatives, or resolve issues before the subscription is lost.
### How long does it take to see results?
Setup can take 10-20 minutes total when access is ready, because Signals lives on top of the customer stack instead of replacing it. The refundable pilot then runs as a 4-week randomized control trial to tune the first playbooks and measure lift against a holdout. The goal is to prove roughly 10x ROI by the end of the pilot.
### What integrations do you support?
Signals integrates with Shopify, Gorgias, Listrak, Slack, and 50+ other platforms. It lives on top of the stack the brand already uses, so helpdesks, return portals, SMS marketing tools, and commerce systems can stay in place. Brands provide access to the commerce platform and an escalation path for support. Signals handles the rest.
### How do you measure business impact?
Every pilot is refundable and includes a randomized control test with a holdout group. Signals measures incremental new-customer conversion, revenue per customer, cross-sell conversion, repeat purchase rate, and subscription retention by comparing Signals-engaged customers against the control.
Signals typically charges per engaged conversation per month, so ROI is measured against actual conversation volume and cost.
### How is pricing structured?
Signals typically charges per engaged conversation per month. Exact pricing depends on expected volume and refundable pilot scope, which is why pricing and ROI are usually scoped together in the demo.
### How do you handle security and compliance?
Signals separates support and customer-initiated conversations from optional unsolicited marketing. When marketing is enabled, Signals gates brand-initiated sends by channel, consent state, and restricted-state signals, honors STOP-style and natural-language opt-outs, and maintains an active SOC 2 program. See /compliance and /security.
## From Our Blog
- [What apparel retailers learn when customers can text back after delivery](/blog/what-apparel-retailers-learn-post-delivery-texts.md)
- [Memory turns a post-delivery text thread into a revenue channel](/blog/memory-post-delivery-revenue-channel.md)
- [How Quaker Marine turned post-delivery check-ins into 51% more repeat purchases with Signals](/blog/quaker-marine-case-study.md)
- [Post-Purchase Concierge vs SMS Marketing, Helpdesks, and Return Portals](/blog/ecomm-cx-marketing-return-platforms.md)
- [Best Texting Services for Ecommerce Apparel (2026)](/blog/best-apparel-texting-services-2026.md)
- [Best Customer Support Platforms for Ecommerce Apparel (2026)](/blog/best-apparel-customer-support-platforms-2026.md)
- [Best Return Portals for Ecommerce Apparel (2026)](/blog/best-apparel-return-portals-2026.md)
- [The End of Reactive Support](/blog/return-starts-before-return.md)
- [SMS CX Shouldn't Be NoReply](/blog/sms-cx-shouldnt-be-noreply.md)
- [View all posts](https://www.returnsignals.com/blog)
## Ready To Engage Your Customers?
Find out what customers say when you actually ask.
Scope a measured lifecycle pilot for your store. Most teams can connect Text Us, post-delivery outreach, review workflows, and measurement within the first week.
- [Book a Demo](https://cal.com/alejandro-zaniolo/30min?overlayCalendar=true)
## Key Links
- Website: https://www.returnsignals.com
- Blog: https://www.returnsignals.com/blog
- Book a Demo: https://cal.com/alejandro-zaniolo/30min?overlayCalendar=true
- Compliance: https://www.returnsignals.com/compliance
- Security: https://www.returnsignals.com/security
- Email: hello@returnsignals.com
## Legal
- [Messaging Compliance](/compliance.md)
- [Privacy Policy](/privacy.md)
- [Security](/security.md)
- [Terms of Service](/terms.md)
---
# Best Customer Support Platforms for Ecommerce Apparel (2026)
> A 2026 comparison of Gorgias, Zendesk, Intercom, Gladly, Re:amaze, and Signals for apparel brands. Helpdesks, chat suites, and personal customer relationships through concierge texting.
{
question: "What's the best helpdesk for Shopify apparel brands?",
answerHtml: 'If you want a helpdesk purpose-built for ecommerce and Shopify context, Gorgias is often the first place to look. ([Gorgias Helpdesk](https://www.gorgias.com/products/helpdesk))',
answerText: "If you want a helpdesk purpose-built for ecommerce and Shopify context, Gorgias is often the first place to look.",
},
{
question: "Is Intercom good for ecommerce?",
answerHtml: 'It can be a strong fit for chat-heavy support teams that want an integrated chat + helpdesk suite, especially when you\'re building structured workflows and automation. ([Intercom Suite](https://www.intercom.com/suite))',
answerText: "It can be a strong fit for chat-heavy support teams that want an integrated chat + helpdesk suite, especially when you're building structured workflows and automation.",
},
{
question: "Does Signals replace my helpdesk?",
answerHtml: "No. Signals builds a personal relationship between your brand and every customer through concierge texting. It works alongside your helpdesk. Many issues get resolved through that relationship before they ever become tickets, but the helpdesk stays the system of record for escalations and complex cases.",
answerText: "No. Signals builds a personal relationship between your brand and every customer through concierge texting. It works alongside your helpdesk. Many issues get resolved through that relationship before they ever become tickets, but the helpdesk stays the system of record for escalations and complex cases.",
},
{
question: "Can I do SMS support inside my helpdesk?",
answerHtml: 'Some helpdesks support SMS as a channel (for example, Gorgias offers SMS support). ([Gorgias SMS](https://www.gorgias.com/product/sms)) But adding SMS as a channel isn\'t the same as proactive engagement. The workflow design matters.',
answerText: "Some helpdesks support SMS as a channel (for example, Gorgias offers SMS support). But adding SMS as a channel isn't the same as proactive engagement. The workflow design matters.",
},
{
question: "What metrics should an apparel support team track in 2026?",
answerHtml: "A practical set: time-to-first-response, resolution time, CSAT, ticket deflection, return reasons and exchange conversion, and repeat purchase after support interaction.",
answerText: "A practical set: time-to-first-response, resolution time, CSAT, ticket deflection, return reasons and exchange conversion, and repeat purchase after support interaction.",
},
];
Apparel customer support in 2026 is a weird mix of high volume and high emotion.
Customers aren't just asking "where is my order." They're asking:
- "Does this shrink?"
- "Is this final sale?"
- "Can I swap sizes without paying shipping?"
- "This looked different in photos, what are my options?"
Meanwhile, almost every platform is pushing "AI agents," while many customers still want a human. Gartner found 64% of customers would prefer companies didn't use AI for customer service. ([Gartner press release](https://www.gartner.com/en/newsroom/press-releases/2024-07-09-gartner-survey-finds-64-percent-of-customers-would-prefer-that-companies-didnt-use-ai-for-customer-service))
So this guide compares the tools that actually make an apparel support org work, and a tactic most comparison guides skip: proactive post‑purchase texting.
## TL;DR
For most apparel brands in 2026, the best "customer support platform" is actually two systems: a helpdesk to manage inbound tickets across email/chat/social, and a relationship layer that gives your brand a personal connection with every customer. Signals is our pick for that layer. It starts post-purchase and grows into something customers actually talk to: asking for styling advice, volunteering what they want to buy next, resolving issues through conversation instead of tickets. For the helpdesk: Gorgias is the strongest for Shopify commerce context, Zendesk for enterprise scale, Intercom for chat + AI agent workflows, Gladly for customer‑first omnichannel, and Re:amaze for smaller Shopify teams.
Table of contents
[Top customer support platforms for apparel in 2026](#top-tools)
[Comparison table](#comparison-table)
[How to choose](#how-to-choose)
[Where Signals fits](#where-return-signals-fits)
[Common matchups](#common-matchups)
[FAQ](#faq)
## Top customer support platforms for apparel in 2026
### 1. Signals: best for building a personal relationship with every customer
Signals texts every customer after delivery to check in on fit and quality. Customers reply because the message is about their specific order, in a channel they already use. In a [randomized experiment with Quaker Marine](/blog/quaker-marine-case-study), 58% of customers replied on iMessage and RCS, and the ones who engaged repurchased 51% more often.
Here's what we've measured:
- 58% reply rate on iMessage and RCS ([Quaker Marine experiment](/blog/quaker-marine-case-study))
- +51% repeat purchase lift among customers who engaged in conversation
- 70% of return-intent conversations choose an exchange
- 13% mentioned reorder plans; 9% asked about products they hadn't bought yet
Ticket prevention, exchange conversion, product feedback: those are all use cases of the relationship. Signals typically works alongside your helpdesk (Gorgias/Zendesk/etc.).
### 2. Gorgias: best helpdesk for Shopify‑centric apparel brands
Gorgias is a helpdesk built for ecommerce that unifies support channels and pulls in Shopify order context. ([Gorgias Helpdesk](https://www.gorgias.com/products/helpdesk))
If you live in Shopify every day, Gorgias is usually the default shortlist.
They also offer SMS as a support channel, which can be useful for "send a photo of the damage" workflows. ([Gorgias SMS](https://www.gorgias.com/product/sms))
### 3. Zendesk: best for enterprise-scale ticketing and omnichannel operations
Zendesk consolidates conversations across channels with AI tooling and enterprise controls. ([Zendesk Service](https://www.zendesk.com/service/))
For larger teams with complex routing, SLAs, and reporting needs, Zendesk stays competitive (the learning curve is real though).
### 4. Intercom: best for chat-first support teams and AI-agent workflows
Intercom combines an AI agent ("Fin") and a helpdesk in one platform. ([Intercom Customer Service Suite](https://www.intercom.com/suite))
If your CX motion is chat‑heavy and you want an AI agent + a modern inbox, Intercom is the most opinionated option.
### 5. Gladly: best for customer-first (not ticket-first) omnichannel service
Gladly is built around treating each customer as a single conversation across time and channels (instead of 4 separate tickets from 4 different channels about the same pair of jeans). ([What is Gladly?](https://help.gladly.com/docs/what-is-gladly))
### 6. Re:amaze: best for smaller Shopify teams that want an all-in-one inbox
Re:amaze is a Shopify‑available helpdesk/live chat tool that's frequently chosen by smaller ecommerce teams. ([Re:amaze on the Shopify App Store](https://apps.shopify.com/reamaze))
It's not the most "enterprise," but it's fast to ship.
## Comparison table
| Tool | Best for | Key strength | Limitation |
| --- | --- | --- | --- |
| **Signals** | Personal relationship with every customer | Concierge texting; 58% reply rate and +51% repeat purchase lift among engaged customers ([Quaker Marine](/blog/quaker-marine-case-study)) | Not a ticketing system |
| **Gorgias** | Shopify apparel brands | Shopify order context; automation; unified inbox | Not built for proactive outreach |
| **Zendesk** | Enterprise scale, SLAs, reporting | Omnichannel consolidation; enterprise controls | Heavier setup and overhead |
| **Intercom** | Chat‑first teams | Strong chat UX; integrated AI agent | Less Shopify‑native than Gorgias |
| **Gladly** | Relationship‑driven service | Customer‑not‑ticket model; omnichannel thread | Often a bigger platform change |
| **Re:amaze** | Smaller Shopify teams | Fast to adopt; all‑in‑one inbox | Less enterprise depth |
## How to choose (apparel-specific)
Use these questions to narrow quickly:
1. Are you Shopify-first? Start with Gorgias (and evaluate Re:amaze if you're smaller).
2. Is chat your main support surface? Consider Intercom.
3. Do you have SLAs, complex routing, or multiple brands? Zendesk becomes more attractive.
4. Do you want long-lived customer threads across channels? Gladly is built around that model.
5. Do you want your brand to have a personal relationship with every customer (and fewer tickets as a side effect)? Add Signals.
## Where Signals fits
Most support tools wait for the customer to have a problem. The entire relationship is reactive: something goes wrong, a ticket opens, you respond. For apparel, the customer has already decided to return before you get a chance to help.
Signals builds a different kind of connection. We text after delivery, but the goal isn't "checking in." It's making the brand someone the customer trusts enough to talk to honestly. That trust is the product.
What that trust makes possible:
- Customers tell you about fit issues before they file a return (and 70% choose an exchange when they do)
- Happy customers tell you what they want to buy next, without being asked
- You get real product signal: which SKUs have sizing problems, which items look different in photos, which packaging disappoints
- Fewer tickets, because the customer already has someone to talk to
The relationship starts post-purchase, but it doesn't stay there. It becomes the channel where customers ask about new drops, request styling help, and tell you what they need next season.
## Common matchups (quick answers)
### Gorgias vs Zendesk
- Choose Gorgias if you're ecommerce/Shopify-first and want commerce-native context fast.
- Choose Zendesk if you're optimizing for enterprise controls, routing, and reporting across a large org.
### Intercom vs Zendesk
- Choose Intercom if chat is your primary channel and you want an integrated AI-agent + inbox suite.
- Choose Zendesk if you're standardizing broad omnichannel ticketing and enterprise workflows.
### Does Signals replace a helpdesk?
No. Signals is a relationship layer. It gives your brand a personal connection with every customer through concierge texting. That relationship prevents a lot of tickets (because the customer already has someone to talk to), but your helpdesk stays the system of record for escalations and complex cases.
"@context": "https://schema.org",
"@type": "ItemList",
"name": "Best customer support platforms for e-commerce apparel (2026)",
"itemListElement": [
{ "@type": "ListItem", "position": 1, "name": "Signals", "url": "https://www.returnsignals.co" },
{ "@type": "ListItem", "position": 2, "name": "Gorgias", "url": "https://www.gorgias.com" },
{ "@type": "ListItem", "position": 3, "name": "Zendesk", "url": "https://www.zendesk.com" },
{ "@type": "ListItem", "position": 4, "name": "Intercom", "url": "https://www.intercom.com" },
{ "@type": "ListItem", "position": 5, "name": "Gladly", "url": "https://www.gladly.ai" },
{ "@type": "ListItem", "position": 6, "name": "Re:amaze", "url": "https://www.reamaze.com" }
]
};
---
# Best Return Portals for Ecommerce Apparel (2026)
> A 2026 comparison of Loop, Narvar, AfterShip, Happy Returns, ReturnGO, and Signals for apparel brands. What each tool does best and where the personal customer relationship fits.
{
question: "What is a return portal?",
answerHtml: "A return portal is a self-serve workflow that lets customers initiate a return or exchange, select a resolution (refund, credit, exchange), and generate the return method (label, QR code, or drop-off) within the brand's policies.",
answerText: "A return portal is a self-serve workflow that lets customers initiate a return or exchange, select a resolution (refund, credit, exchange), and generate the return method (label, QR code, or drop-off) within the brand's policies.",
},
{
question: "Which return portal is best for Shopify apparel brands?",
answerHtml: 'If Shopify is your system of record and exchange flow quality matters, Loop is commonly shortlisted because it focuses on returns and exchanges within the Shopify ecosystem. ([Loop listing](https://apps.shopify.com/loop-returns))',
answerText: "If Shopify is your system of record and exchange flow quality matters, Loop is commonly shortlisted because it focuses on returns and exchanges within the Shopify ecosystem.",
},
{
question: "Can a return portal reduce returns?",
answerHtml: "Portals can reduce friction and increase exchanges vs refunds with good flows. But actually reducing return rate requires reaching the customer before they commit to returning, and that means having a relationship where they'll tell you what's wrong instead of just requesting a label.",
answerText: "Portals can reduce friction and increase exchanges vs refunds with good flows. But actually reducing return rate requires reaching the customer before they commit to returning, and that means having a relationship where they'll tell you what's wrong instead of just requesting a label.",
},
{
question: "What's the best way to increase exchanges instead of refunds?",
answerHtml: "Two levers work together: strong exchange UX inside the portal (variant swaps, store credit options) and a pre-return intervention that helps customers land on the right outcome before they commit to a return.",
answerText: "Two levers work together: strong exchange UX inside the portal (variant swaps, store credit options) and a pre-return intervention that helps customers land on the right outcome before they commit to a return.",
},
{
question: "Where does Signals fit if I already use Loop / Narvar / AfterShip?",
answerHtml: "Signals layers on top. It gives your brand a personal, two-way relationship with every customer starting at delivery. That relationship handles returns, exchanges, and buying intent in conversation, and routes customers into your existing portal when self-serve is the best path.",
answerText: "Signals layers on top. It gives your brand a personal, two-way relationship with every customer starting at delivery. That relationship handles returns, exchanges, and buying intent in conversation, and routes customers into your existing portal when self-serve is the best path.",
},
{
question: "Is boxless return drop-off worth it?",
answerHtml: 'For many apparel brands it can be, because convenience impacts customer behavior. Happy Returns is one of the best-known options for box-free drop-offs via its Return Bar network. ([Happy Returns](https://happyreturns.com/))',
answerText: "For many apparel brands it can be, because convenience impacts customer behavior. Happy Returns is one of the best-known options for box-free drop-offs via its Return Bar network.",
},
{
question: "What should I measure when switching return software?",
answerHtml: "Beyond portal completion rate, measure: % exchanges vs refunds, cost per return (including labor), time-to-refund and CSAT, and return reasons by SKU (and what you changed upstream).",
answerText: "Beyond portal completion rate, measure: % exchanges vs refunds, cost per return (including labor), time-to-refund and CSAT, and return reasons by SKU (and what you changed upstream).",
},
];
Returns software is crowded in 2026.
And if you're an apparel brand, it's not optional.
- The NRF projects $849.9B in U.S. retail returns in 2025, and estimates ~19.3% of online sales will be returned. ([NRF: 2025 Retail Returns Landscape](https://nrf.com/research/2025-retail-returns-landscape))
- For apparel specifically, ICSC reports consumers returned ~22% of products bought online (vs 6.2% in-store). ([ICSC Consumer Returns Survey write‑up](https://www.icsc.com/news-and-views/icsc-exchange/brick-and-mortar-shopping-drives-lower-return-rate-than-online-shopping))
- Returns are also expensive: Radial cites an estimate of $27 to process a return for a $100 eCommerce order. ([Radial: Returns Management in 2024](https://www.radial.com/insights/returns-management-2024))
So brands buy a return portal. Portals handle logistics well. But a portal only reaches the customer after they've already decided to return the item.
The brands with the strongest repeat purchase rates are doing something different: building a personal relationship with every customer, starting right after delivery. [Quaker Marine](/blog/quaker-marine-case-study) saw 51% more repeat purchases among engaged customers from a single post-delivery check-in. That relationship is what turns a refund into an exchange, a one-time buyer into a repeat customer, and a support ticket into a conversation. This guide compares return portals *and* the relationship layer that sits before them.
## TL;DR
For most apparel brands in 2026, you need two layers: a self-serve portal for customers who already decided to return, plus a relationship layer that gives your brand a personal connection with every customer starting at delivery. Signals is our pick for the relationship layer (two-way concierge texting that starts post-purchase and grows from there). For the portal itself: Loop for Shopify‑native exchanges, Narvar for enterprise, AfterShip Returns for multi‑carrier value, Happy Returns if box‑free drop‑offs matter.
Table of contents
[Top return portals for apparel in 2026](#top-tools)
[Comparison table](#comparison-table)
[How to choose](#how-to-choose)
[Where Signals fits](#where-return-signals-fits)
[Common matchups](#common-matchups)
[FAQ](#faq)
## Top return portals (and tools) for apparel in 2026
This is a ranked list in the spirit of a buyer's guide. For apparel, the question that matters most is: how much revenue can you retain when a customer is about to return?
### 1. Signals: best for building a personal relationship with every customer
Signals gives your brand a direct, two-way connection with every customer starting right after delivery. That connection handles returns and exchanges, yes, but it also surfaces buying intent, collects real feedback, and turns one-time buyers into people who actually talk to your brand.
In a [randomized experiment with Quaker Marine](/blog/quaker-marine-case-study), here's what we measured:
- 58% reply rate on iMessage and RCS
- +51% repeat purchase lift among customers who engaged in conversation
- 70% of return-intent conversations choosing an exchange instead of a return
- 13% mentioned reorder plans; 9% asked about products they hadn't bought yet
Customers reply because the message is about their specific order and arrives in a channel they already use. If you already use Loop/Narvar/AfterShip, Signals layers on top.
### 2. Loop: best return portal for Shopify‑heavy apparel brands
Loop is a widely adopted returns & exchanges platform for Shopify brands, with a strong focus on exchange flows and post‑purchase experience. ([Loop on the Shopify App Store](https://apps.shopify.com/loop-returns))
Loop also supports exchanges through Shopify's native exchange infrastructure, which matters for apparel brands doing heavy variant swaps. ([Loop: Shopify Native Exchanges](https://help.loopreturns.com/en/articles/2771905))
### 3. Narvar: best for enterprise returns experiences and post‑purchase ops
Narvar is an enterprise platform that spans post‑purchase experiences (tracking, returns, etc.). On the returns side, it handles policy enforcement, exchanges, and analytics. ([Narvar Return overview](https://support.narvar.com/hc/en-us/articles/11307520101651-Narvar-Return-Overview))
### 4. AfterShip Returns: best value for multi‑carrier / global returns operations
AfterShip's returns product automates the exchange and refund workflow, with a focus on operational scale. ([AfterShip Returns](https://www.aftership.com/returns))
For brands with lots of SKUs, regions, and carriers, AfterShip competes on automation and policy rules.
### 5. Happy Returns: best if box‑free drop‑offs (and fraud checks) matter
Happy Returns (a UPS company) is best known for its drop‑off network (Return Bars) and reverse logistics capabilities, plus a return & exchange portal. ([Happy Returns](https://happyreturns.com/))
If you're battling fraud and want stronger item verification at drop‑off, Happy Returns also markets fraud prevention capabilities. ([Happy Returns Return Bar network](https://happyreturns.com/box-free-return-bar-network))
### 6. ReturnGO: best for Shopify brands that want a configurable self‑serve flow
ReturnGO is a Shopify‑friendly returns/exchanges product focused on self‑serve flows, labels, and automation. ([ReturnGO on the Shopify App Store](https://apps.shopify.com/returngo))
If you want a portal with lots of rules without jumping to an enterprise contract, ReturnGO is commonly shortlisted.
## Comparison table
This table focuses on what each tool is actually optimized to do and where it fits in an apparel stack.
| Tool | Best for | Key strength | Limitation |
| --- | --- | --- | --- |
| **Signals** | Personal relationship with every customer | Two-way thread after delivery; 58% reply rate and +51% repeat purchase lift among engaged customers ([Quaker Marine](/blog/quaker-marine-case-study)) | Not a label/RMA system |
| **Loop** | Shopify brands optimizing exchanges | Strong exchange UX; Shopify native exchanges | Portal starts after decision |
| **Narvar** | Enterprise post‑purchase ops | Enterprise controls, integrations, analytics | Implementation complexity |
| **AfterShip Returns** | Multi‑carrier / global ops | Automation rules; global orientation | Portal starts after decision |
| **Happy Returns** | Boxless drop‑offs, fraud prevention | Return Bar network; physical verification | Best value with their network |
| **ReturnGO** | Shopify brands, configurable flows | Flexible self‑serve portal; Shopify integration | Less suite breadth |
## How to choose a return portal for an apparel brand
A simple decision checklist:
1. If you want more exchanges (and fewer refunds): prioritize exchange flows *and* add a pre‑return layer.
2. If you're Shopify‑centric: start with a Shopify‑native ecosystem tool (Loop / ReturnGO) and evaluate exchange mechanics.
3. If you're enterprise/multi‑system: Narvar (or an enterprise returns suite) may be worth the integration overhead.
4. If drop‑off behavior matters: Happy Returns is unique because it pairs software with a physical network.
5. If you ship globally / multi‑carrier: operational automation matters as much as portal UX (AfterShip often competes here).
One more question that usually matters most:
> Does your brand have a relationship with the customer, or just a transaction history?
- If the customer's first interaction after buying is a return portal, you're optimizing logistics.
- If the customer's first interaction is a real conversation with someone who knows them, you're building a relationship. The retention follows from that.
## Where Signals fits
Most returns stacks have a gap: nobody actually talks to the customer.
- The shipping notification is transactional.
- The return portal is an exit.
- Support tickets are high-effort and reactive.
Signals fills the gap by giving your brand a personal relationship with every customer, starting with a simple text after delivery.
That relationship is what makes everything else work. In the [Quaker Marine experiment](/blog/quaker-marine-case-study), 58% of customers replied because the message was about their specific order and arrived in a channel they already use. Once that connection exists, the use cases follow naturally:
1. Returns and exchanges get handled in conversation, while customers still feel helped instead of blocked. 70% of return-intent threads end in an exchange.
2. Fit questions, care questions, "this looks different than the photos" get caught early, before a return decision forms.
3. 13% of customers mentioned reorder plans; 9% asked about products they hadn't bought yet.
4. Real product feedback flows upstream (what SKUs confuse people, what sizing guidance is missing).
It starts post-purchase. It grows into something much bigger. And it layers on top of whatever portal you already use.
## Common matchups (quick answers)
### Loop vs Narvar
- Choose Loop if you're Shopify‑first and want a strong portal/exchange flow without an enterprise suite.
- Choose Narvar if returns are part of a broader enterprise post‑purchase transformation (tracking + comms + ops).
### AfterShip Returns vs ReturnGO
- Choose AfterShip if multi‑carrier/global operational automation is the core problem.
- Choose ReturnGO if you want a Shopify‑friendly, configurable portal experience with simpler implementation.
### Happy Returns + a portal
Many brands use Happy Returns for the drop‑off network and still run a separate portal workflow.
The best architecture is often:
- Portal = self-checkout (policy, labels, RMA)
- Conversation = a person at the brand who knows what the customer bought and can help with sizing, exchanges, or what to buy next
- Network = the drop-off convenience
"@context": "https://schema.org",
"@type": "ItemList",
"name": "Best return portals for e-commerce apparel (2026)",
"itemListElement": [
{ "@type": "ListItem", "position": 1, "name": "Signals", "url": "https://www.returnsignals.co" },
{ "@type": "ListItem", "position": 2, "name": "Loop Returns", "url": "https://www.loopreturns.com" },
{ "@type": "ListItem", "position": 3, "name": "Narvar", "url": "https://corp.narvar.com" },
{ "@type": "ListItem", "position": 4, "name": "AfterShip Returns", "url": "https://www.aftership.com/returns" },
{ "@type": "ListItem", "position": 5, "name": "Happy Returns", "url": "https://happyreturns.com" },
{ "@type": "ListItem", "position": 6, "name": "ReturnGO", "url": "https://returngo.ai" }
]
};
---
# Best Texting Services for Ecommerce Apparel (2026)
> A 2026 comparison of Attentive, Postscript, Klaviyo, Yotpo SMSBump, Emotive, and Signals for apparel brands. Marketing broadcasts vs two-way conversations.
{
question: "What's the difference between SMS marketing and conversational texting?",
answerHtml:
'SMS marketing is mostly one-to-many messaging (campaigns and automated flows). Conversational texting is one-to-one, where customers can reply and the brand resolves an issue or helps them purchase.',
answerText:
'SMS marketing is mostly one-to-many messaging (campaigns and automated flows). Conversational texting is one-to-one, where customers can reply and the brand resolves an issue or helps them purchase.',
},
{
question: 'Attentive vs Postscript vs Klaviyo, which should I choose?',
answerHtml:
'A simple rule of thumb: Attentive is a strong enterprise option, Postscript is a Shopify-native favorite, and Klaviyo SMS fits best when Klaviyo already runs your lifecycle marketing and segmentation.',
answerText:
'A simple rule of thumb: Attentive is a strong enterprise option, Postscript is a Shopify-native favorite, and Klaviyo SMS fits best when Klaviyo already runs your lifecycle marketing and segmentation.',
},
{
question: 'Does Signals replace my SMS marketing platform?',
answerHtml:
"No. Signals builds a personal relationship between the brand and each customer across the lifecycle: before checkout, after delivery, during exchanges, and before the next order. In the Quaker Marine experiment, 13% of customers mentioned reorder plans and 9% asked about products they hadn't bought yet. Most brands keep their SMS marketing platform for broadcasts and flows, and use Signals for the ongoing 1:1 relationship.",
answerText:
"No. Signals builds a personal relationship between the brand and each customer across the lifecycle: before checkout, after delivery, during exchanges, and before the next order. In the Quaker Marine experiment, 13% of customers mentioned reorder plans and 9% asked about products they hadn't bought yet. Most brands keep their SMS marketing platform for broadcasts and flows, and use Signals for the ongoing 1:1 relationship.",
},
{
question: 'How do I keep SMS from feeling spammy?',
answerHtml:
"Two levers matter most: relevance (send fewer, more contextual messages) and consent with easy opt-out. Signals doesn't feel spammy because the customer is talking to someone they trust, about something they actually care about. The conversation starts from delivery context, not promotion calendars.",
answerText:
"Two levers matter most: relevance (send fewer, more contextual messages) and consent with easy opt-out. Signals doesn't feel spammy because the customer is talking to someone they trust, about something they actually care about. The conversation starts from delivery context, not promotion calendars.",
},
{
question: "What's the fastest way to reduce apparel returns using texting?",
answerHtml:
'Use texting to resolve fit and care questions quickly, guide customers to the right variant, and make exchanges easier than refunds. That generally requires a true two-way workflow, not just a campaign platform.',
answerText:
'Use texting to resolve fit and care questions quickly, guide customers to the right variant, and make exchanges easier than refunds. That generally requires a true two-way workflow, not just a campaign platform.',
},
]
"SMS for ecommerce" is a messy phrase in 2026.
Because it describes two different things:
1. SMS marketing platforms (broadcast campaigns + automated flows)
2. Conversational texting (two‑way support, sales, concierge)
If you're an apparel brand, you usually need both, but you shouldn't expect one tool to do both well.
Compliance note: This article is not legal advice. In the U.S., SMS marketing
generally requires prior express written consent, and you must honor opt‑outs. (See{' '}
[Compliance basics](#compliance) below.)
## TL;DR
Texting software for apparel splits into two jobs: broadcast marketing (campaigns, automations) and two‑way conversational texting (support, sales, concierge). For marketing SMS, Attentive works well at enterprise scale, Postscript is the Shopify favorite, Klaviyo SMS makes sense if Klaviyo already runs your lifecycle, and Yotpo SMSBump is solid on value. For building a real personal relationship with customers after delivery (one that prevents returns, drives exchanges, and surfaces buying intent because the customer actually trusts the brand), Signals is what we'd pick. Emotive also plays in the two-way space with a staffed sales model.
Table of contents
[Top texting services for apparel in 2026](#top-tools)
[Comparison table](#comparison-table)
[How to choose](#how-to-choose)
[Signals vs SMS marketing tools](#return-signals-vs-sms-marketing)
[Compliance basics](#compliance)
[FAQ](#faq)
## Top texting services for apparel in 2026
### 1. Signals: best for building a personal relationship with every customer
Signals turns your brand into someone the customer actually texts with. It starts after delivery, but the relationship it builds goes beyond that moment.
The connection starts simple: "Hey, how's the fit on that jacket?" The customer replies, you help them with a size question or a care tip, and now the customer has a person at the brand they can text. That trust is what makes everything else possible:
- They ask about fit or care, and you resolve it before it becomes a return
- They want to exchange instead of refund, and you handle it in the thread
- They tell you what they're shopping for next ("I need something lighter for summer")
- They ask for purchase suggestions, and you help them buy right there
In the [Quaker Marine experiment](/blog/quaker-marine-case-study), 13% of customers mentioned reorder plans and 9% asked about products they hadn't bought yet. When someone says "I love this, do you have it in green?", you can help them buy it right there in the thread.
You'd keep your existing broadcast tool for campaigns, list growth, and A/B testing.
### 2. Attentive: strong enterprise SMS marketing platform
Attentive handles SMS marketing at scale: personalization, list growth, messaging across SMS/RCS/email. ([Attentive](https://www.attentive.com/))
They also have deep Shopify integrations. ([Attentive + Shopify](https://info.attentive.com/shopify/))
### 3. Postscript: Shopify-native favorite for SMS marketing
Postscript is the go-to for Shopify brands running SMS marketing programs (campaigns, segments, automations). ([Postscript on the Shopify App Store](https://apps.shopify.com/postscript-sms-marketing))
### 4. Klaviyo SMS: best when Klaviyo is your lifecycle system of record
If Klaviyo runs your lifecycle (email + segmentation + flows), Klaviyo's SMS can be the cleanest "single brain" setup. ([Klaviyo on the Shopify App Store](https://apps.shopify.com/klaviyo-email-marketing))
### 5. Yotpo SMSBump: solid value SMS marketing + automation
Yotpo's SMSBump covers SMS marketing and automation for Shopify brands. ([Yotpo SMSBump (Shopify)](https://smsbump.com/pages/shopify))
### 6. Emotive: staffed two‑way conversational texting for sales/marketing
Emotive runs conversational texting as two‑way workflows supported by humans + automation. ([Emotive conversational texting](https://emotive.io/conversational-texting))
Emotive focuses on sales-oriented two-way texting with a staffed model.
## Comparison table
| Tool | Best for | 1‑way blasts | 2‑way conversations |
| ----------------- | ---------------------------------------------------------------------------------- | ------------- | -------------------------------------------- |
| **Signals** | Personal customer relationships that drive retention, exchanges, and buying intent | Not core | **Core**: single ongoing thread per customer |
| **Attentive** | Enterprise marketing programs | **Core** | Possible, but marketing‑oriented |
| **Postscript** | Shopify brands scaling SMS revenue | **Core** | Limited |
| **Klaviyo SMS** | Teams standardized on Klaviyo | **Core** | Limited |
| **Yotpo SMSBump** | Value / Shopify‑friendly programs | **Core** | Limited |
| **Emotive** | Sales‑oriented two‑way texting | Not the focus | **Core**: staffed "humans + AI" model |
## Engagement benchmarks by campaign type
Most SMS tools report click-through rates (CTR) because they're optimized for traffic and attributable revenue. Signals tracks reply rate, because the reply is the start of a conversation. Reply rates and CTRs measure different things: a reply means the customer engaged in a conversation; a click means they tapped a link. We include both below so you can see channel-level engagement, but they aren't directly comparable. In the [Quaker Marine experiment](/blog/quaker-marine-case-study), Signals saw a 58% reply rate on iMessage and RCS. Customers who replied repurchased 51% more often than control.
Use this table to calibrate expectations across common campaign types.
| Platform | Campaign type | Rate |
| ----------- | ------------------------------------- | --------------------------------------------------------------------- |
| **Signals** | Post‑delivery check‑in · iMessage/RCS | **Reply rate:** 58% ([Quaker Marine](/blog/quaker-marine-case-study)) |
| Klaviyo | Broadcast campaign · SMS | CTR: 5.76% avg, 14.89% top 10% |
| Klaviyo | Broadcast campaign · Email | CTR: 1.29% avg, 4.74% top 10% |
| Klaviyo | Welcome flow · Email | CTR: 4.92% avg, 15.69% top 10% |
| Klaviyo | Abandoned cart · Email | CTR: 5.21% avg, 12.35% top 10% |
| Klaviyo | Browse abandonment · Email | CTR: 4.74% avg, 11.08% top 10% |
| Klaviyo | Post‑purchase flow · Email | CTR: 3.48% avg, 12.19% top 10% |
| Omnisend | Campaign SMS | CTR: 7.6% |
| Omnisend | Automated SMS | CTR: 9.4% |
| Omnisend | Campaign email | CTR: 1.22% |
| Omnisend | Automated email | CTR: 5.4% |
| Postscript | Campaign (broadcast) · SMS | CTR: 3.64%–9.20% (p25–p75) |
| Postscript | Abandoned cart · SMS | CTR: 9.35%–17.98% |
| Postscript | Browse abandonment · SMS | CTR: 7.37%–13.59% |
| Postscript | Welcome series · SMS | CTR: 4.93%–12.31% |
| Postscript | Post purchase · SMS | CTR: 4.47%–14.65% |
| Postscript | Back in stock · SMS | CTR: 36.24%–52.74% |
| Postscript | Win back · SMS | CTR: 2.81%–6.85% |
**Notes:** Signals reply rate is from a randomized experiment with [Quaker Marine](/blog/quaker-marine-case-study). Klaviyo rates from the [Klaviyo Benchmark Report](https://www.klaviyo.com/resources/benchmark-report) (Americas). Omnisend rates from [Omnisend benchmarks](https://www.omnisend.com/blog/digital-marketing-statistics/). Postscript ranges are 25th–75th percentile for Fashion & Apparel from [Postscript SMS Benchmarks](https://postscript.io/sms-benchmarks).
## How to choose a texting service (apparel-specific)
Ask these in order:
1. Are you trying to drive revenue with campaigns and flows? Start with Attentive / Postscript / Klaviyo / Yotpo.
2. Do you want customers to see your brand as someone they can text, not just a store they bought from? Signals builds that relationship starting at delivery, and it's what drives exchanges, prevents returns, and surfaces future buying intent.
3. Do you want to staff two-way texting internally? If yes, choose a tool that supports agent workflows. If no, consider a staffed model like Emotive.
4. Where does your lifecycle segmentation live today? If it's already in Klaviyo, adding Klaviyo SMS can reduce complexity.
## Signals vs SMS marketing tools
It's tempting to evaluate Signals against Attentive/Postscript/Klaviyo, but they're built for different jobs.
Marketing SMS tools optimize for reach and revenue attribution. Signals optimizes for something different: giving the customer a person at the brand they can actually text. Exchanges, fit resolution, and buying intent all flow from that relationship. [Quaker Marine](/blog/quaker-marine-case-study) saw a 58% reply rate on iMessage and RCS, and the customers who replied repurchased 51% more often than control.
A few practical differences:
- One continuous relationship: Signals keeps a single thread per customer. It's the same thread where they asked about sizing, got help with an exchange, and mentioned they're shopping for a wedding. Marketing tools create disconnected interactions.
- Starts at delivery, grows from there: The first message is about their order. But the relationship it builds is why they come back and text the brand months later about something new.
- Fewer messages, real relevance: Each message is about something the customer actually cares about. That's what happens when you're talking to a person, not blasting a list.
Signals isn't a broadcast tool, but purchase intent surfaces in these conversations without prompting. In the [Quaker Marine experiment](/blog/quaker-marine-case-study), 13% of customers who replied mentioned plans to reorder. They'll say things like "I'm also looking for a lighter jacket for summer" or "do you have this in a different color?" When that happens, we can help them buy right there in the thread. That signal also feeds your existing marketing tools for smarter targeting.
If you want a deeper head-to-head, see our broader comparison: [Post-Purchase Concierge vs SMS Marketing, Helpdesks, and Return Portals](/blog/ecomm-cx-marketing-return-platforms).
## Compliance basics (what teams forget)
A few things that trip up ecommerce SMS programs:
- Consent: in the U.S., you generally need prior express written consent for marketing texts (TCPA). ([TCPA/CTIA overview](https://www.bloomreach.com/en/blog/understanding-tcpa-and-ctia-compliance-for-sms-marketing-in-the-us))
- Opt-out: industry best practices emphasize clear opt‑out instructions (e.g., "Reply STOP to opt out") and honoring variations. ([CTIA Messaging Principles (PDF)](https://api.ctia.org/wp-content/uploads/2023/05/230523-CTIA-Messaging-Principles-and-Best-Practices-FINAL.pdf))
- Support vs marketing: support texting can have different consent expectations than marketing promotions. Don't assume "we have their number" is enough.
If you're building a program, involve legal early.
'@context': 'https://schema.org',
'@type': 'ItemList',
name: 'Best texting services for e-commerce apparel (2026)',
itemListElement: [
{ '@type': 'ListItem', position: 1, name: 'Signals', url: 'https://www.returnsignals.co' },
{ '@type': 'ListItem', position: 2, name: 'Attentive', url: 'https://www.attentive.com' },
{ '@type': 'ListItem', position: 3, name: 'Postscript', url: 'https://postscript.io' },
{ '@type': 'ListItem', position: 4, name: 'Klaviyo SMS', url: 'https://www.klaviyo.com' },
{ '@type': 'ListItem', position: 5, name: 'Yotpo SMSBump', url: 'https://smsbump.com' },
{ '@type': 'ListItem', position: 6, name: 'Emotive', url: 'https://emotive.io' },
],
}
---
# Post-Purchase Concierge vs SMS Marketing, Helpdesks, and Return Portals
> How building a one-to-one relationship with every customer differs from broadcast SMS, support ticketing, and self-serve return portals.
{
question: 'Is Signals an SMS marketing platform?',
answerHtml:
"Marketing happens naturally, but it's a side effect. In the Quaker Marine experiment, 13% of customers mentioned reorder plans and 9% asked about products they hadn't bought yet. The product is a personal relationship between the brand and every customer: shoppers can text before they buy, and the same thread continues after delivery. Most brands keep their marketing SMS tool for campaigns and flows.",
answerText:
"Marketing happens naturally, but it's a side effect. In the Quaker Marine experiment, 13% of customers mentioned reorder plans and 9% asked about products they hadn't bought yet. The product is a personal relationship between the brand and every customer: shoppers can text before they buy, and the same thread continues after delivery. Most brands keep their marketing SMS tool for campaigns and flows.",
},
{
question: 'Will customers think this is spam?',
answerHtml:
"No, if done right. The key is timing and relevance: start after delivery, ask about the item, be helpful, and don't turn every message into an ask. With existing customers, less than 0.1% of customers asked to stop texting.",
answerText:
"No, if done right. The key is timing and relevance: start after delivery, ask about the item, be helpful, and don't turn every message into an ask. With existing customers, less than 0.1% of customers asked to stop texting.",
},
{
question: 'Does Signals replace Klaviyo / Attentive / Postscript?',
answerHtml:
'Usually no. Signals builds a personal relationship with every customer across the lifecycle: before checkout, after delivery, during exchanges, and before the next order. It sits on top of your existing marketing SMS, helpdesk, and returns portal. Most brands keep all their existing tools and add Signals as the relationship layer.',
answerText:
'Usually no. Signals builds a personal relationship with every customer across the lifecycle: before checkout, after delivery, during exchanges, and before the next order. It sits on top of your existing marketing SMS, helpdesk, and returns portal. Most brands keep all their existing tools and add Signals as the relationship layer.',
},
{
question: 'Does Signals replace Gorgias?',
answerHtml:
"No. Signals isn't a ticketing system. It reduces pressure on your helpdesk by resolving many post-purchase issues before they become tickets, but your helpdesk remains the system of record for escalations.",
answerText:
"No. Signals isn't a ticketing system. It reduces pressure on your helpdesk by resolving many post-purchase issues before they become tickets, but your helpdesk remains the system of record for escalations.",
},
{
question: 'How hard is the integration?',
answerHtml:
'We designed Signals to be lightweight. In many cases (especially on Shopify), brands can get onboarded quickly because we sit on top of your existing stack rather than ripping anything out. We feed the conversation data back into your systems of record.',
answerText:
'We designed Signals to be lightweight. In many cases (especially on Shopify), brands can get onboarded quickly because we sit on top of your existing stack rather than ripping anything out. We feed the conversation data back into your systems of record.',
},
{
question: 'Is this a returns tool?',
answerHtml:
"It touches returns, but the goal is bigger. The product is a personal relationship with every customer. Returns happen to be one of the most measurable use cases of that relationship (when someone you trust says 'want to try a different size?', most people say yes). Customers also use the thread for fit advice, product questions, and telling us what they want to buy next.",
answerText:
"It touches returns, but the goal is bigger. The product is a personal relationship with every customer. Returns happen to be one of the most measurable use cases of that relationship (when someone you trust says 'want to try a different size?', most people say yes). Customers also use the thread for fit advice, product questions, and telling us what they want to buy next.",
},
{
question: "What's the biggest reason teams adopt Signals?",
answerHtml:
'The response. Most channels are designed for announcements. Signals is designed so the customer has someone at the brand they can actually text. When customers trust the thread, they reply. And when they reply, you can help before they give up.',
answerText:
'The response. Most channels are designed for announcements. Signals is designed so the customer has someone at the brand they can actually text. When customers trust the thread, they reply. And when they reply, you can help before they give up.',
},
]
Three kinds of software touch the customer after an e‑commerce purchase: SMS marketing (Attentive, Postscript, Klaviyo, etc.), customer support (Gorgias, Zendesk, Intercom, etc.), and returns and exchanges (Loop, Narvar, Happy Returns, etc.). Most brands own all three. And yet the customer still feels alone the moment the box arrives.
That's the moment the online experience diverges from a good store. In a store, uncertainty triggers help. _"How's the fit? Want to try a size up?"_ Online, uncertainty triggers silence until the customer opens a return portal, creates a ticket, or decides they're done.
Signals exists for the post-delivery window: a personal relationship with every customer, starting right after delivery. If you're evaluating Signals, you're probably asking some version of "Is this like Attentive?", "Do I replace Gorgias?", or "Is this a returns product like Loop?" This post breaks down exactly how Signals differs from each category.
Table of contents
[Signals vs Attentive](#return-signals-vs-attentive)
[Signals vs Postscript](#return-signals-vs-postscript)
[Signals vs Klaviyo (SMS)](#return-signals-vs-klaviyo)
[Signals vs Gorgias](#return-signals-vs-gorgias)
[Signals vs Zendesk](#return-signals-vs-zendesk)
[Signals vs Intercom](#return-signals-vs-intercom)
[Signals vs Loop](#return-signals-vs-loop)
[Signals vs Narvar](#return-signals-vs-narvar)
[Signals vs Happy Returns](#return-signals-vs-happy-returns)
## TL;DR
Signals builds a personal relationship between your brand and every customer. It starts with a text after delivery, but the relationship quickly becomes more than post-purchase: customers share fit preferences, ask for product recommendations, and treat the brand like a trusted advisor. Returns, exchanges, and repeat purchases are all use cases of that relationship.
In a [randomized experiment with Quaker Marine](/blog/quaker-marine-case-study), customers who engaged in conversation repurchased 51% more often. Here's what the data looked like:
- 58% reply rate on iMessage and RCS
- +51% repeat purchase lift among customers who engaged in conversation
- 13% mentioned reorder plans; 9% asked about products they hadn't bought yet
- In return-intent conversations, ~70% choosing an exchange instead of a return
- <0.1% of customers asking us to stop texting
Customers reply because the message is about their specific order, arrives after delivery, in a channel they already use. The exchange rate is high because when someone you trust says "want to try a different size?", it doesn't feel like a save attempt. For a lot of brands, the exchange conversion alone is worth roughly 5% margin expansion. And because customers volunteer buying intent without being asked, the follow-up targeting gets surprisingly effective (many brands model this as ~5-10% LTV upside over time).
- If you need one-to-many promotions, you want Attentive / Postscript / Klaviyo.
- If you need ticketing + inbox management, you want Gorgias / Zendesk / Intercom.
- If you need self‑serve returns logistics, you want Loop / Narvar / Happy Returns.
- If you want a personal relationship with every customer (that starts post-purchase and grows from there), you want Signals.
Most brands don't replace those tools with us. They layer Signals on top.
## Comparison table
What each category of tool is actually optimized to do:
| | **Signals** | **SMS Marketing** | **Helpdesks** | **Return Portals** |
| ---------------------------- | -------------------------------------------------------------------------------------- | -------------------------------------------------- | ------------------------------------- | ---------------------------------------------------- |
| **Primary job** | Create a personal relationship with every customer | Drive revenue via campaigns + flows | Manage inbound issues efficiently | Make returns/exchanges easy + policy‑compliant |
| **Moment** | Right after delivery (before return intent hardens) | Anytime (often pre‑purchase + promo windows) | When the customer reaches frustration | After the customer has decided to return |
| **Interaction** | 1:1 conversation | 1:many broadcast | Ticket queue / inbox | Form / portal workflow |
| **Who initiates** | Brand initiates with a check‑in | Brand initiates promos | Customer initiates | Customer initiates |
| **What customers feel** | "Someone is paying attention to me." | "I'm being marketed to." | "I need help; I'm opening a case." | "I'm leaving; make it fast." |
| **Best outcome** | Customer trusts the brand; exchanges, kept items, and purchase intent follow naturally | Clicks + conversions | Resolution + deflection | Completed return |
| **Replaces existing tools?** | No, sits on top and feeds data back | No | Sometimes | Sometimes |
If you take nothing else away: Signals builds a relationship. The behavioral moment is where it starts. The relationship is what compounds.
## Signals vs Attentive
If you found this post via "Signals vs Postscript" or "Signals vs Klaviyo," this section applies. These tools are strong broadcast SMS marketing platforms. Signals builds a personal relationship with every customer, starting post-purchase.
### What SMS marketing platforms are great at
These platforms are built for revenue operations:
- Subscriber capture (popups, opt‑ins, list growth)
- Segmentation and targeting
- Campaign calendars
- Automation flows (welcome series, cart abandonment, winback)
- Attribution and reporting
If your main objective is _"send the right promo to the right cohort,"_ these tools are excellent.
### Where SMS marketing breaks down
When SMS becomes a broadcast channel, a subtle problem appears. A text thread is the closest thing to a real conversation most brands have with customers. If you train customers that your SMS thread is only a place where the brand asks for money, they'll treat it like one. They'll mute you, opt out, or simply stop believing you're there to help.
One veteran retailer put it bluntly:
> "They're all standing there with their hand out… trying to get into your wallet."
That's a critique of an incentive structure, not any specific vendor: when the primary KPI is attributable revenue, the default move is "more sends."
Signals is designed around the opposite constraint:
- One durable thread where the customer has a person at the brand they can text
- Fewer messages, each one about something the customer cares about right now
- A relationship anchored to _what the customer just bought_ and _what they're experiencing_, that grows into something bigger over time
### The easiest way to tell which category you need
Ask: what is the "unit of work"? In broadcast SMS, the unit of work is a _campaign_. In Signals, the unit of work is a _conversation_. That's why our core metric is response rate, not CTR. When a customer replies, whether it's "I love it" or "this doesn't fit," you're in a conversation. And conversations lead to exchanges, repurchases, and product feedback.
### When to use Signals _with_ Attentive / Postscript / Klaviyo
This is the common "best stack" outcome:
- Keep your marketing SMS tool for what it does best: promotional messaging and lifecycle automation.
- Use Signals for what _we_ do best: building a personal relationship with every customer after delivery.
The handoff looks like this: your marketing tool drives the purchase, Signals builds the relationship (delivery → try-on → uncertainty → resolution → ongoing trust), and the data from those conversations (fit notes, preferences, sentiment, future buying intent) feeds back into your CRM/ESP for smarter targeting. The goal is a deeper relationship, not more texts.
### Signals vs Postscript
Postscript is a popular choice for lifecycle SMS, especially for Shopify brands. If your primary question is "can Signals replace my SMS marketing platform?", the answer is usually no. Campaign tooling and concierge conversation are different jobs.
### Signals vs Klaviyo SMS
Klaviyo is often the system of record for lifecycle messaging (especially email). If you're evaluating Klaviyo SMS specifically: the distinction is the same. Klaviyo excels at segmentation, flows, and attribution. Signals excels at building a personal relationship with every customer after delivery.
## Signals vs Gorgias
Everything in this section also applies if you're comparing Signals vs Zendesk or Signals vs Intercom. These are helpdesks: they manage inbound support. Signals builds a proactive relationship with every customer, starting right after delivery.
### What helpdesks are great at
Helpdesks exist because inbound support is chaotic. They give you an inbox and SLA workflows, macros, routing, tagging, multi‑channel support coordination, agent productivity and QA, and reporting and staffing visibility. If you're managing volume, they're indispensable.
### Where helpdesks can't go
Helpdesks are reactive by design. Even if you respond quickly, the customer has to notice a problem, decide it's worth the effort, find the support entry point, describe the issue, and wait. That's fine for many issues, but it misses the most important moment in e‑commerce: the moment the customer is uncertain but not angry. That's where returns are born, and where the relationship is either created or lost.
Signals flips the sequence:
- Brand speaks first
- The customer replies in the lowest‑effort way possible (a text)
- The issue is resolved in the same thread, with photos when helpful
### Why the channel matters for AI + human handoffs
One of the underrated advantages of SMS is that it is both synchronous and asynchronous. If a bot needs to hand off to a human in web chat, the customer is often literally waiting. In SMS, a handoff can happen naturally. Two minutes, two hours, still normal. That makes the AI/human hybrid actually workable.
### Do you replace your helpdesk?
Usually no. Signals isn't a ticketing system. We sit on top of your stack: we handle the post‑purchase thread, and when something needs escalation, we route it into your existing support workflows. The result is fewer tickets in the first place, because customers already got helped in-thread.
### Signals vs Zendesk
Zendesk is a standard helpdesk for ticketing across many industries. If your support org lives in Zendesk, you don't replace it with Signals. You add Signals to capture and resolve the post‑purchase moments _before_ they become tickets.
### Signals vs Intercom
Intercom is often used for in‑app chat and proactive messaging, particularly in software. In e‑commerce contexts, the same logic applies: chat sessions are great when customers are on‑site; SMS threads are better when customers are living their lives away from your site.
## Signals vs Loop
If you found this post via "Signals vs Narvar" or "Signals vs Happy Returns," this section applies. These products are return portals / returns infrastructure. Signals is the personal relationship that often prevents the return in the first place.
### What return portals are great at
Return portals win on operational clarity:
- Policy enforcement
- Label creation and tracking
- Exchange flows
- Store credit options
- Warehouse and reverse logistics integration
They make returning easy.
### The limitation is timing
Return portals begin when the customer has already decided _"This isn't working for me."_ At that point, "help" can feel like "friction." Signals starts earlier, after delivery, before the customer has turned uncertainty into a decision. With existing customers, in conversations where customers expressed a desire to return, ~70% chose an exchange instead. That outcome is hard to achieve once the customer is already inside a portal flow.
### The right architecture is often "portal + concierge"
The cleanest mental model: the portal is the _lane_, and Signals is the _relationship_. Sometimes customers want the lane. They already know what they want, so make it fast. But for everything else, having someone who knows you and cares matters more than a faster workflow.
Signals can coexist with a portal: if the customer wants self‑serve, they can still use it. If they text back with uncertainty, we keep it human and help them land on the right outcome.
### Signals vs Narvar
Narvar is widely used for post‑purchase tracking and returns experiences. If you already have Narvar, great. We're not competing to show the same tracking page. We're trying to make sure the customer doesn't silently drift into a return without ever feeling helped.
### Signals vs Happy Returns
Happy Returns (and similar networks) improve the _physical_ logistics of returning. That matters. But logistics isn't the same as a relationship. Signals aims to reduce the number of returns that happen simply because a customer was uncertain and alone.
## What we're actually competing with
It's tempting to turn this into a feature checklist, but the real competitor is the absence of a relationship. Most brands have transactional shipping updates, a dead SMS thread, a fragmented support stack, and a return portal that works beautifully as an exit. The customer has tools. What they don't have is a person.
We're betting on a different principle. One retailer told us:
> "People prefer to feel special rather than ordinary."
That's what happened with [Quaker Marine](/blog/quaker-marine-case-study). A simple check-in after delivery led to 58% of customers replying. The ones who replied repurchased 51% more often. They shared what they liked, what didn't work, what they wanted next. The relationship became the source of every other metric: fewer returns, more exchanges, higher LTV, better product data.
## What to measure
If you're used to evaluating marketing tools, you'll look for CTR, conversion rate, and attributable revenue. Those matter, but they measure transactions. The metrics that matter here are proxies for _how much the customer trusts the brand_.
The early leading indicators are:
- Response rate (do customers actually reply?)
- Sentiment (does it feel like care or annoyance?)
- Opt-out rate (are we earning the right to be in the thread?)
- Exchange conversion (when there's return intent, do we save the relationship?)
- Signal capture (how often do customers volunteer fit notes, preferences, future buying intent?)
In the [Quaker Marine experiment](/blog/quaker-marine-case-study), 13% of customers who replied mentioned plans to reorder and 9% asked about products they hadn't bought yet (e.g. "I'll buy this in green if it comes back"). We typically run a 4‑week pilot with a proper control group so you can see real impact.
## The simplest way to choose
If you're trying to decide between Signals and "the usual suspects," here's the decision tree:
- If you need promotional SMS, buy an SMS marketing platform.
- If you need ticketing, buy a helpdesk.
- If you need returns logistics, buy a returns portal.
- If you want a personal relationship with every customer, add Signals.
Every other tool on this list manages a transaction. Signals builds the relationship that makes every transaction better.
---
# Gemini 3.5 Flash-Lite lost to 3.1 Flash-Lite, and more thinking didn't help
> We benchmarked Gemini 3.5 Flash-Lite against Gemini 3.1 Flash-Lite and Gemini 3 Flash Preview on an e-commerce catalog agent. The newer model cost more and passed fewer tasks end to end.
When [Gemini 3.5 Flash Lite came out](https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-6-flash-3-5-flash-lite-3-5-flash-cyber/), we got super excited. A big part of what makes our agent work well is being able to understand a customer's catalog, which may include thousands of SKUs. For this we delegate to a subagent, historically run on Gemini Flash and Flash-Lite models, that crunches through search results and surfaces the most appropriate items. This often takes multiple search-tool calls, so being cheap per token doesn't necessarily mean being cheap per task as some models tend to waste a lot of tokens and run extraneous tool calls. Being well-grounded is also very important, as seeing potentially hundreds of related results can cause models to hallucinate combinations that are not part of the catalog.
However, this was the first time a newer model failed to beat an older one. We ran the comparison on the catalog-browsing agent inside our customer support automation for Shopify merchants. We compared `gemini-3-flash-preview`, [`gemini-3.1-flash-lite`](https://ai.google.dev/gemini-api/docs/models/gemini-3.1-flash-lite), and [`gemini-3.5-flash-lite`](https://ai.google.dev/gemini-api/docs/models/gemini-3.5-flash-lite) as the model driving that agent, running both Lite models at `minimal`, `medium`, and `high` [thinking](https://ai.google.dev/gemini-api/docs/generate-content/thinking).
The result: `gemini-3.5-flash-lite` lost. No 3.5 configuration beat `gemini-3.1-flash-lite` on the downstream benchmark at any thinking level. The best profile overall was 3.1 at medium thinking, which also beat its own high setting on pass rate, cost, and tokens.
## The task
When a shopper texts something like "i am looking for a puffer jacket to match my [recently bought] pants" the customer-facing agent delegates the catalog part of the turn to the catalog subagent, passing a natural-language goal plus whatever structured constraints it extracted: products and variants already in the conversation, order IDs, size, color, category, budget.
On top of that goal, the catalog agent gets a context packet built for the conversation: recent messages, recently referenced products, the customer's order history, a style profile computed from past purchases (likely sizes, preferred categories and vendors), and a store profile describing the merchant's catalog (merchandise domain, search rewrite rules, vocabulary, grounding guidance).
It then runs a bounded tool loop over seven domain-specific tools, plus the ability to spawn additional subagents. The tools cover search over a mirrored copy of the merchant's Shopify catalog, product detail lookups, live price and availability validation, and on-demand context like recent orders, style profile, merchant guidance, etc.
Each turn ends in a structured result: short recommendation text, grounded options and alternatives (product cards carrying price, availability, image, and purchase URL), and full product details for follow-up questions. A deterministic validator checks every product, variant, price, availability, and link claim in that result against the tool transcript. Anything unsupported is flagged as a grounding error.
Downstream, the customer-facing agent treats this result as evidence and merges the grounded cards and recommendation text with the non-catalog parts of the turn and writes the reply the shopper actually receives. The models below were swapped in as the catalog agent only; everything around it stayed fixed.
## How we measured
We collected a number of examples that have historically caused our agent to make mistakes, across two test types:
- 13 isolated tasks that call the catalog agent directly.
- 42 production-shaped conversations in which a separate customer-facing agent receives the catalog result and writes the reply.
Beyond objective measures, we also asked the LLM to subjectively judge the quality of the response again a rubric. In every downstream run, the customer-facing agent, simulated customer, and judge were pinned to `gemini-3.5-flash`.
Costs below cover the catalog agent only, at standard [Vertex AI provider rates](https://cloud.google.com/vertex-ai/generative-ai/pricing). Every per-pass metric keeps failed work in the numerator: total catalog spend, tokens, tool calls, or wall time divided by successful tasks.
## Isolated catalog tasks
### Isolated benchmark results
Detailed results across 13 direct catalog-agent tasks, grouped by model family.
| Model family | Thinking | Pass rate | Avg. judge | Cost / pass | Tokens / pass | Tool calls / pass | Seconds / pass | Grounding errors |
| --- | --- | ---: | ---: | ---: | ---: | ---: | ---: | ---: |
| 3 Flash Preview | Preview | 92.3% | 4.692 | $0.0233 | 41.3k | 5.42 | 4.85 | 4 |
| 3.1 Flash-Lite | Minimal | 84.6% | 4.692 | $0.0104 | 36.2k | 5.00 | 6.77 | 2 |
| 3.1 Flash-Lite | Medium | 100.0% | 4.923 | $0.0087 | 29.8k | 4.31 | 2.70 | 0 |
| 3.1 Flash-Lite | High | 92.3% | 4.769 | $0.1143 | 116.3k | 5.83 | 37.07 | 6 |
| 3.5 Flash-Lite | Minimal | 84.6% | 4.462 | $0.0147 | 37.9k | 5.55 | 6.62 | 0 |
| 3.5 Flash-Lite | Medium | 100.0% | 5.000 | $0.0179 | 37.5k | 5.23 | 5.47 | 1 |
| 3.5 Flash-Lite | High | 100.0% | 4.923 | $0.0230 | 40.3k | 5.46 | 13.89 | 1 |
Both Lite models solved all 13 tasks at medium thinking, but 3.1 Flash-Lite got there with the lowest cost, fewest tokens, fewest tool calls, and shortest wall time per pass.
High thinking moved the two generations in opposite directions. It kept 3.5 Flash-Lite at 13/13, at more time and spend. On 3.1 Flash-Lite it slipped to 12/13 while burning 1.40 million catalog tokens (more than 3.5 times its medium run) and throwing 6 grounding errors, the most of any profile.
## Downstream agentic tasks
The direct benchmark doesn't test the handoff. A catalog result can appear all right in isolation but the main agent might still choose to make subsequent requests with alternative plain-English queries before replying to the customer. Because of this we also evaluate catalog-browsing in the context of full agent turn.
### Downstream benchmark results
End-to-end pass rate after the catalog output flowed through the customer-facing agent, grouped by model family.
| Model family | Thinking | Pass rate | Avg. judge | Cost / pass | Tokens / pass | Tool calls / pass |
| --- | --- | ---: | ---: | ---: | ---: | ---: |
| 3 Flash Preview | Preview | 91.4% | 4.800 | $0.0547 | 112.1k | 6.81 |
| 3.1 Flash-Lite | Minimal | 94.3% | 4.857 | $0.0221 | 91.2k | 5.73 |
| 3.1 Flash-Lite | Medium | 97.1% | 4.943 | $0.0219 | 88.4k | 5.44 |
| 3.1 Flash-Lite | High | 88.6% | 4.686 | $0.0937 | 152.0k | 6.42 |
| 3.5 Flash-Lite | Minimal | 91.4% | 4.771 | $0.0343 | 104.5k | 6.56 |
| 3.5 Flash-Lite | Medium | 88.6% | 4.686 | $0.0422 | 115.7k | 6.74 |
| 3.5 Flash-Lite | High | 91.4% | 4.714 | $0.0522 | 121.8k | 6.94 |
While in isolated test 3.5 Flash Lite medium and high reached the same 100% pass rate as 3.1 Flash Lite, in end-to-end testing 3.1 Flash Lite was a clear winner.
3.1 Flash-Lite medium passed 97.1% downstream tasks, the best pass rate in the table. It beat every 3.5 Flash-Lite setting and its own high setting on pass rate, cost per pass, tokens, and tool calls, all at once. Against 3 Flash Preview it ran roughly 60% cheaper per passing task, on 21% fewer tokens and 20% fewer tool calls.
## What we took away
The newer model lost. 3.5 Flash-Lite scored slightly higher on isolated judge scores and stayed at 13/13 even at high thinking, but downstream, every 3.5 configuration passed fewer tasks than 3.1 at minimal or medium and cost at least 50% more per passing task than 3.1 medium. Overall, it's an interesting reminder: if we had upgraded based on the release notes, this agent would have gotten worse and more expensive.
---
# How good is Gemini 3.7 Flash at commerce?
> Gemini 3.7 Flash High matched our production Gemini 3.5 Flash configuration on 35 customer-agent tasks while cutting Standard-tier cost per run by 52%.
We use Gemini models to bring a store associate over iMessage to every customer after they receive their item. When [Gemini 3.7 Flash launched](https://cloud.google.com/vertex-ai/generative-ai/docs/models/gemini/3-7-flash), we wanted to know if we should switch. Newer doesn't always mean better in the context of tasks that are not being hill-climbed by the labs, and for example [last month Gemini 3.5 Flash-Lite lost to 3.1 Flash-Lite](/blog/gemini-3-5-flash-lite-agent-eval) on our catalog agent. If we just upgraded because "it is newer" we would have gotten worse and more expensive. Our earlier tests on 3.6 Flash were also disappointing, which is why we stayed on 3.5 Flash for our main agent.
For this blog we'll show our bake-off between Gemini 3.5, 3.6, and 3.7 Flash, each at Low, Medium, and High thinking. 9 configurations, 315 customer-agent executions (full [task description and methods](#task-description-and-methods) at the end of the post). Thinking more is not always better, sometimes performance peaks at less than high (our current agent runs 3.5-medium because 3.5-high didn't outperform), sometimes higher thinking leads to overthinking and actually degrades performance. What was also interesting is that higher thinking was also not always more expensive, this is because higher thinking often meant better use of tools, and fewer tool turns often compensated more tokens per turn.
In our current test three configurations tied at the top with successfully completing 26 of 35 tasks: 3.5 Medium, 3.5 High, and 3.7 High. Among these, 3.7 High tied for the top judge score, and both cost about half as much and had half the latency per run.
## Results
Each configuration ran the same 35 production-shaped tasks against our main customer-facing agent, with everything except the primary model and its thinking level held fixed.
### Primary customer-agent thinking matrix
Matched results across 35 tasks per configuration. Lower cost and latency are better.
| Model family | Thinking | Pass rate | Avg. judge | Standard cost / run | Median latency / run |
| --- | --- | ---: | ---: | ---: | ---: |
| 3.5 Flash | Low | 65.7% | 4.143 | $0.2012 | 42.40s |
| 3.5 Flash | Medium | 74.3% | 4.371 | $0.2534 | 52.40s |
| 3.5 Flash | High | 74.3% | 4.400 | $0.2534 | 55.53s |
| 3.6 Flash | Low | 71.4% | 4.286 | $0.0962 | 28.64s |
| 3.6 Flash | Medium | 68.6% | 4.229 | $0.1164 | 33.42s |
| 3.6 Flash | High | 60.0% | 3.829 | $0.0874 | 30.82s |
| 3.7 Flash | Low | 68.6% | 4.171 | $0.1125 | 25.29s |
| 3.7 Flash | Medium | 68.6% | 4.143 | $0.1188 | 28.75s |
| 3.7 Flash | High | 74.3% | 4.400 | $0.1216 | 28.41s |
Thinking didn't improve quality monotonically. High was best for 3.7, Low was best for 3.6, and Medium and High tied on pass count for 3.5. Another interesting observation is that 3.6-high was actually cheaper and faster than 3.6-medium because despite higher per-turn token usage, it used much fewer tools.
Gemini 3.6 is a good example of how more thinking isn't always better. High thinking passed 21 of 35 tasks, six fewer than Low, and had higher median latency per run. Its lower total spend reflects shorter and fewer agent trajectories, not better task performance.
## Task description and methods
Our primary agent writes the reply a shopper actually receives based on multiple tools and subagent replies. When someone texts "can i return this jacket? also looking for something warmer for fall," that single message pulls on order history, return eligibility, the return action itself, and catalog search (via catalog sub-agent).
The agent has to pick out the right order line, progress the return safely, delegate the product search, hold context across turns, and fold everything into one useful reply. Add customer memory, review requests, follow-up scheduling, and escalation to a person, and a conversation can wander through eight subsystems and five third-party integrations before it's done.
The 35 tasks come from our previous real-customer interaction failures and so represent a real-world "challenge" dataset:
- 11 core single-turn tasks covering returns, exchanges, product questions, mixed return-and-browse requests, and review suppression.
- 12 core multi-turn tasks covering context retention, return execution, follow-on purchases, price deltas, existing labels, and review requests.
- 12 customer-quality regressions covering delivery status, escalation timing, feedback handling, soft fit complaints, and settled conversations.
The 12 multi-turn tasks use an LLM simulator that plays the customer. The simulator gets a persona, an opening text, and a private goal the agent never sees (say, "return the jacket, then get a purchase link for a replacement"). It reacts to whatever the agent actually says, deciding each turn whether to keep going.
Grading has two layers. First, deterministic expectations check the trace for hard facts: were the right tools called, is every product claim grounded in tool evidence, did required artifacts like a purchase link survive into the customer-visible reply, did the agent escalate when it shouldn't have. Second, an LLM judge reads the scenario definition, the simulator's private goal, the full transcript, and compact tool evidence, then scores the conversation 1 to 5 against a fixed rubric. A task passes only if no deterministic expectation fails and the judge scores 4 or 5. A 3 (partially succeeds but misses an important requirement) is counted as a fail.
Costs cover the customer agent and its fixed helper routes, excluding the eval simulator and judge. We priced every usage row at Google's public Standard tier as of August 13, 2026, then divided each configuration's total agent cost by all 35 evaluated runs, whether they passed or failed. Median latency per run gives each run one observation: we sum its customer-agent turn durations, including fixed helper work, then take the median across the 35 runs.
## Takeaway
Gemini 3.7 Flash High matched the quality leaders across the complete 35-task set while running at 52% lower cost per run and with 46% lower median latency per run than our production 3.5 Medium configuration. After two Gemini releases in a row that couldn't beat their predecessors on our benchmarks, this one earned the upgrade.
---
# Jordan Craig started iMessaging customers after delivery. Two weeks in, each conversation was worth an extra $30
> In a completed A/B test across thousands of customers, each Signals conversation with a Jordan Craig customer added an estimated $30 in repeat revenue within two weeks of the post-delivery check-in, measured against a randomized holdout.
Jordan Craig × Signals
In streetwear, the next sale starts the moment the box opens. [Jordan Craig](https://jordancraig.com) used Signals to check in right after delivery, suggest the piece that completes the look, and measure the result against a holdout group that got nothing. Each conversation added an estimated $30 in repeat revenue within two weeks of delivery, $21 of it in the first week alone, and the earliest customers are tracking toward roughly $56 by week four. The full-pilot gap is statistically significant (p = 0.03).
Per conversation
+$30
Estimated extra repeat revenue per conversation within 14 days, measured against the holdout.
Repliers
+40%
Customers who replied repurchased 40% more often than the holdout.
Reply rate
36%
Customers texted who wrote back, with replies still arriving 10 days after the check-in.
“What stands out is that the growth is both incremental and profitable. In an A/B test we saw that every Signals conversation made us an extra $30 after 2 weeks.”
Rob Varon
Vice President, Digital, Jordan Craig
## At a glance
Jordan Craig is a premium streetwear brand built on denim, cargos, fleece, and outerwear. The catalog is made for sets: denim with the matching trucker jacket, shorts with the work shirt, the same silhouette coming back across washes and drops. That makes the day the box opens the single best moment to suggest the next piece (the customer is literally holding half the outfit). Most brands go quiet right then: after the delivery notification, nothing.
Signals gave Jordan Craig a way to show up in that window. The check-in starts as a simple question about how the order landed and, when the answer is positive, turns into a cross-sell: the matching jacket for the wash they just bought, a photo, a direct link, a restock alert when the piece is sold out. No survey, no discount.
“I appreciate the follow up. That’s damn good customer service.”
— Wayland, Jordan Craig customer
## A small change to the customer journey
From May 11 to June 11, 2026, every new Jordan Craig customer was randomly assigned 50/50: half got the post-delivery check-in from Signals, and half were held out, continuing through the normal post-purchase experience. Every result below is the check-in group measured against that holdout.
50 / 50
treatment / control split
31 days
live, May 11 to June 11, 2026
36%
of customers texted replied
## The result: customers came back and bought again
The dollars below come from Shopify order data, and every figure is an increment against the holdout. In the first 7 days, each conversation added an estimated $21.16 in repeat purchases. By week two that grew to $30.34. And the earliest customers, the ones who've had a full four weeks since their check-in, are tracking at $56.40 extra per conversation, though that cohort is still small (about 100 customers a side).
**Extra repeat revenue vs. holdout**
| Basis | 7 days | 14 days | 28 days |
| :------------------ | ------: | ------: | ------: |
| Per conversation | +$21.16 | +$30.34 | +$56.40 |
| Per customer texted | +$7.97 | +$11.51 | +$21.61 |
Customers who engaged in a conversation repurchased 40% more often than the holdout over the pilot (p = 0.012).
“Thanks a milly James, as soon as they restock, I’ll most def cop them on the 1st‼️”
— Daniel, Jordan Craig customer
## More than a third of customers replied
36% of the customers Signals texted wrote back. Marketing texts usually land between 1% and 5%, so a message that gets over a third of customers to respond operates on a different kind of channel. And the replies kept coming: 31% answered within 6 hours, and the rate was still climbing 10 days after the check-in, because people respond when they've actually lived with the product.
These weren't throwaway replies, and the threads rarely stopped at "love it." Once a customer confirmed the order landed well, the agent suggested the piece that completes the look, asked a size, sent a photo, dropped a direct link. Sold out in their size? It set a restock alert, and customers answered with exactly what they'd buy on restock day.
One real thread, lightly edited
Signals: saw your order arrived. how's it looking so far?
Customer: these pants are so fire and exactly what im looking for. I love em
Signals: love that fit! the hudson denim trucker jacket in mesa wash is a great match if you want to complete the look. what's your usual shirt size?
Customer: 3xl
Signals: that one's sold out in 3xl, but the local dealer trucker jacket in cement wash is available. want to see a photo?
Customer: Sure 👍
Customers drove it themselves just as often: asking for the same pair in black, for the jacket that matches the wash, for restock dates on three styles at once. That's the kind of buying signal brands usually spend real marketing dollars to surface, and here it's worth real money: an extra $30 per conversation inside two weeks, before counting anything that compounds later.
## The same thread did the support work
A conversation that starts as a check-in doesn't stay in its lane. About 36% of the conversations Signals opened during the pilot ended up working a real issue: a size that didn't fit, a package that went sideways, a question about the return policy. The agent handled those in the same thread, and roughly a quarter of them ended in a return or exchange completed right there, no email queue, no hold music.
That protects the revenue story too, because the fastest way to lose a second purchase is to fumble the first one. When a customer said the fit was off, the agent checked stock in their size, suggested the right swap, and linked the exchange portal in one message. One customer's mixed-up order got corrected before it even shipped; another's "where's my package" was settled with a tracking link in minutes.
The obvious worry is that putting an exchange link one text away invites more returns. It didn't. The check-in group returned about 13% fewer of their delivered items than the holdout. The returns that were going to happen anyway just became a better experience: caught early, in a thread the customer already had open, and steered toward the right size instead of a refund.
## Why this mattered for Jordan Craig
Jordan Craig didn't change its brand, its product, or its drop calendar. It added one well-timed conversation, and the post-delivery window stopped being silence: it sells the next piece, books the exchange, and hears about it when something's off.
The read after a full month is simple. When Jordan Craig showed up after delivery, customers answered. And more of them bought again.
“Thank you. That’s some awesome customer service. I love it.”
— Javon, Jordan Craig customer
Want results like these for your brand?
Signals helps apparel and lifestyle brands turn post-delivery check-ins into real customer conversations and measurable repeat purchase lift.
[Book a Demo](https://cal.com/alejandro-zaniolo/30min?overlayCalendar=true)
---
# Memory turns a post-delivery text thread into a revenue channel
> A post-delivery text thread that remembers the exchange, the gift recipient, and the product a customer was already eyeing stops being a delivery check-in and starts earning revenue, conversation after conversation.
> “Hi Olivia! I LOVE the striped tee but it’s a little tight; how do I exchange for a large?”
> “Hi!! I love it. I haven’t given it to the person I purchased it for yet, so I’m hoping it will fit him!”
> “Btw, do you have favorite items?”
Those are three replies to three separate Signals check-ins. One customer needs an exchange, another is thinking through a gift they haven't handed over yet, and the third wants to know what the brand thinks is worth buying next. In an ordinary ecommerce stack, every one of those requests would get routed to a different team's inbox and never meet the other two. The customer doesn't see any of that happening behind the scenes. They texted one number and expect whoever is on the other end to help.
What jumped out most was how much work the thread ended up doing once people replied. A conversation that started as a simple "how did the delivery go?" could shift into a return request by the second message and a cross-sell by the third, with a piece of product feedback tucked somewhere in the middle. And because Signals kept remembering each piece, the next check-in a month later didn't start from scratch. It picked up where the last one ended.
## First, the thread has to feel alive
> “Thank you Olivia! I just got home and tried on the shirt. Love the fit and will keep it.”
> “I love that you texted to ck on tops!! They are perfect and will definitely be ordering more colors!!”
More than half of the first replies said “thanks.” 24% said “love.” and 18% mentioned "Olivia" (our bot persona) by name. Throughout the vast majority of our conversation, it was clear that customers treated us as a person, not just an outbound campaign.
Even customers who suspected automation still answered like they were texting a person. “Hey sorry, figured this was automated. Got the hat yesterday, it’s great, thanks for checking in!” Once the thread feels alive, the customer uses it.
A thread that feels alive is the preamble to every reorder, upsell, and exchange that follows. And this is how Signals starts building a memory of the customer, one warm reply at a time. They mention a spouse they've been shopping for; Signals remembers. They mention a trip they're packing for; Signals remembers that too. By the time the next check-in lands, the conversation has a history, and the longer that history runs, the less it feels like a brand reaching out and the more it feels like an old acquaintance keeping in touch.
## One message often does 2 jobs
Half the replies that flagged an issue still included a "thanks" or a "love it" in the same message. Customers write what's on their mind, and in a real conversation, it turns out people don't separate the complaint from the compliment. They arrive together, in the same sentence, because the customer is answering a friend who just asked how the tee fit.
Patricia opens the thread with an exchange request. She loves the tee, but the size is wrong. 9 days later, Patricia's exchanged tee gets delivered. Signals already knows she swapped for the large, so when it checks back in, it doesn't ask her to re-explain anything; it just asks if the large fits better. Patricia answers by asking for a tee for her husband, and when she confirms the size Signals provides the link to buy the second item. The thread remembered the first conversation and picked up where it left off, so a customer who could easily have become a refund turned into a retained sale plus a cross-sell, all inside the same warm thread that started with a fit complaint.
## The customer's living a life
Because this is in an iMessage thread rather than a chatbot in an app or on a website, customers don't feel pressured to reply immediately; instead, Signals fits into their schedule. They reply when the product is physically in their hands, and they've had a minute to think about it. In fact, about half of our eventual replies come from a follow up ping (about 24 hours later), rather than our initial outreach.
Gift and spouse threads are the clearest example of where the memory aspect is most valuable. About 1 in 10 replied threads involved somebody other than the purchaser as the actual wearer. Someone's buying for a husband, or a teenage son, or a friend whose birthday is still three weeks out. In practice, Signals ends up managing purchases for two or three people inside a single thread, keeping each person straight. Once the context is stored, Signals can come back weeks later to ask how the gift landed and continue the relationship.
This is what ecommerce systems usually miss. The customer's using one thread to bring the brand into their life as it's actually happening, on their schedule and around whatever else they're doing that day. Every other channel makes the customer come to it. The thread comes to the customer, and it shows up already knowing who they're shopping for and when the gift has to be ready.
## The thread has a memory
Think about what this thread is quietly collecting. One week the customer mentions their size, the next they volunteer that they're buying for a birthday at the end of the month, and a few weeks later they admit the last jacket ran tight in the shoulders. None of it feels like data at the moment it's shared, but all of it gets remembered. By the time a new order or a new check-in lands, there's already a running record of who the customer is and what they care about, and the next message picks up from exactly where the last one ended.
When Signals follows up with Patricia, it already knows she exchanged for a large and that she bought a second tee for her husband. When it checks in with Linda, it already knows there's a gift recipient who hasn't received the cap yet. When a new order comes in from Roy, it already knows which size was the mistake and which one fit his wife. The customer never has to re-explain anything. The brand never has to guess.
That memory is what turns individual check-ins into a relationship. It's also what makes upsell and cross-sell land the right way. "Hey, you mentioned you wanted something thinner for summer, the linen version just dropped" reads like a friend who was paying attention, and that's because the thread actually was.
## The thread turns into a revenue channel
Over 20% of replied threads end up discussing another product, a restock, or something the customer is planning to buy next. Those are the chattiest threads in the whole dataset, because excited customers keep talking, and what they're excited about is usually the next thing they want to own.
A lot of that intent arrives in the first reply. One customer will ask whether a piece comes in a darker wash. Another will mention a jacket she's been eyeing on the site but hasn't pulled the trigger on. The important part is that these are products the brand actually sells today, in a size the thread already knows the customer wears. Signals doesn't have to invent a recommendation; it just has to match what the customer said to what's in stock, and come back a week later when the moment is right.
By that point, the thread's doing direct revenue work. A customer will casually mention another product one week, and place a reorder the next. Because the thread remembers both conversations, Signals can come back days later and close the upsell it first heard about unprompted. Across replied conversations, 13% mentioned reorder plans and 9% asked about products the customer hadn't bought yet. That's real buying intent showing up on its own, from customers already holding the product, and most brands pay Klaviyo and Attentive real money trying to surface the same signal.
## The revenue impact
The thread generates real revenue, and it's measurable.
In a [randomized experiment with Quaker Marine](/blog/quaker-marine-case-study/), customers reached on iMessage and RCS replied 58% of the time. Treatment customers repurchased 16% more often than control within 3 weeks. Among customers who actually engaged in conversation, repeat purchase lift was 51%.
Those numbers make sense once you read the conversations. The lift comes from a thread that picks up where it left off, leaning on what it remembers from three weeks ago. A sizing problem that would have hardened into a refund turns into an exchange instead, because the conversation catches it before it escalates. A customer mentions they're eyeing a new product today, and a week later Signals comes back with it in hand. None of that is a single well-timed message; it's a relationship unfolding inside the same window where the shipment notifications used to go quiet.
What's worth holding onto is that the customer kept using the same thread as the work changed. Every exchange, every gift question, every reorder, every stray piece of product feedback went into the same running record, and the brand came back to each one knowing everything it had already been told. That's the difference between a delivery check-in and a revenue channel.
Laurie never had to re-explain herself in a portal, a ticket, and a marketing system. The same thread that opened with a check-in became the place where the return could actually start.
## Memory and a long-lasting thread enables a new revenue channel for DTC brands
DTC brands already own the rarest asset in ecommerce: a direct line to every customer who's ever bought from them. The trouble is that the line usually goes quiet the moment the package lands. Most brands spend the week before delivery reaching out over Shopify notifications and the week after delivery quietly hoping the customer comes back on their own.
A post-delivery text thread changes that, but only when the thread is allowed to live. The same thread that handled Patricia's exchange last month is still holding everything it learned along the way, so the next time Signals has a reason to reach out, it already knows she bought for herself and for her husband and doesn't have to start over. That kind of continuity is what turns a one-off check-in into a relationship, and it runs inside the same channel the brand was already paying to send shipment updates into.
The revenue shows up in the numbers. In the Quaker Marine experiment, engaged customers repurchased 51% more often than control in the first three weeks alone, and that lift came from the same threads that earlier handled the exchanges and the gift questions. Every conversation the thread remembered made the next one easier to earn.
If the post-delivery thread stays one-way, it's just logistics. If it remembers, it becomes the richest revenue channel a DTC brand already owns.
If you want to see this kind of thread running on your own post-delivery window, [book a demo](https://cal.com/alejandro-zaniolo/30min?overlayCalendar=true) and we'll walk through what it would look like for your brand.
---
# How Quaker Marine turned post-delivery check-ins into 51% more repeat purchases with Signals
> In a randomized experiment across 1,910 customers, Signals lifted repeat purchases by 16% at 3 weeks. Among customers who engaged in post-delivery conversations on iMessage and RCS, repeat purchase rates were 51% higher.
Quaker Marine × Signals
Most brands disappear after delivery. [Quaker Marine](https://quakermarine.com) used Signals to turn that moment into a conversation. In a randomized experiment across 1,910 customers, treatment customers repurchased 16% more often within 3 weeks. On iMessage and RCS, about 58% replied. And among customers who engaged in conversation, repeat purchase rates were 51% higher than the control group.
Overall lift
+16%
3-week repeat purchase lift for treatment customers vs. control.
Reply rate
58%
Customers reached on iMessage and RCS who replied to the check-in.
Engaged cohort lift
+51%
3-week repeat purchase lift among customers who replied.
“My philosophy has always been to make a customer, not a sale. Signals gives us a way to do online what the best retailers have always done in person.”
Kevin McLaughlin
Owner, Quaker Marine Supply Co.
## At a glance
Quaker Marine is the kind of brand people want to hear from. The products are distinctive. The brand feels personal. The service has character.
But like most ecommerce brands, the customer experience still risked going quiet right after delivery. Customers got the standard order emails, then the relationship depended on whether they decided to come back on their own.
Signals gave Quaker Marine a better way to show up in that window. Instead of letting the moment fade, the brand sent a simple post-delivery check-in over iMessage and RCS. Customers could ask about sizing, care, restocks, returns, or what to buy next. It felt less like a campaign and more like a real conversation.
That shift turned post-delivery into something more valuable than a courtesy touch. It became a retention channel.
“QMS is the only company where someone has reached out to follow up on the clothing I’ve ordered. Please share with your team that this element of service is greatly appreciated.”
— Garry, Quaker Marine customer
## A small change to the customer journey
Starting on February 18, 2026, every new order was randomly assigned 50/50 to treatment or control.
Treatment customers received a post-delivery check-in from Signals. Control customers received Quaker Marine's normal Shopify transactional emails only.
The message itself was simple. It arrived after delivery, in a channel customers actually use, and opened the door to a helpful conversation. No survey. No discount. No hard sell. Just a timely reason to reply.
For the channel analysis below, we focus on customers reached through iMessage and RCS, where Signals delivers its richest conversational experience.
1,910
customers in the experiment
961 / 949
treatment / control split
773
customers reached via iMessage or RCS
454
customer conversations generated
“Glad to hear from you! I’ll be back because of this text. Means so much!”
— Mona, Quaker Marine customer
## The result: customers came back and bought again
The check-in drove measurable repeat purchase across every time window we measured.
**Repeat purchase rate: treatment vs. control**
| Cohort | 1 week | 2 weeks | 3 weeks |
|:-------|-------:|-------:|-------:|
| Control | 3.6% (29/796) | 6.6% (39/595) | 9.7% (45/464) |
| Treatment (iMessage + RCS) | 4.6% (31/670) | 8.3% (41/493) | 11.2% (41/366) |
| Replied customers | 5.0% (20/397) | 10.2% (29/285) | 14.7% (31/211) |
| **Lift (treatment vs. control)** | **+27%** | **+27%** | **+16%** |
| **Lift (replied vs. control)** | **+38%** | **+55%** | **+51%** |
Each column uses customers reached via iMessage or RCS with enough elapsed time for that measurement window.
Two things stand out.
First, the lift is real across the entire treatment group, not just among customers who replied. Even customers who saw the check-in and never responded repurchased at higher rates than control. Receiving the message alone moved the needle. Having a conversation moved it further.
Second, the replied cohort's advantage grew over time. At 1 week, responded customers repurchased 38% more than control. By 2 weeks, that gap widened to 55%, and at 3 weeks it held at 51%. That pattern is consistent with what you'd expect from relationship-building: the conversation creates a connection to the brand that compounds as the customer thinks about their next purchase.
“I love that you texted to check on the tops. They are perfect and I will definitely be ordering more colors.”
— Holly, Quaker Marine customer
## More than half of customers replied
Of the 773 customers reached via iMessage or RCS, 57.6% replied. On iMessage (717 customers), the reply rate was 56.9%. On RCS (56 customers), it was 66.1%. RCS is a smaller sample but consistently outperformed.
These are not typical marketing response rates. For context, SMS campaigns average 1-5% reply rates. Signals reached 57% because the message arrives in a personal channel, at the right moment, and asks a question the customer actually wants to answer.
**Engagement rate by channel**
| Channel | Customers | Reply rate |
|:--------|----------:|-----------:|
| iMessage | 717 | 56.9% |
| RCS | 56 | 66.1% |
| Combined | 773 | 57.6% |
“Impressive follow-up. This is my second purchase from Quaker Marine.”
— Thor, Quaker Marine customer
## These were real conversations, not throwaway replies
We categorized all 454 conversations where a customer replied.
**Conversation outcomes (454 replied customer conversations)**
| Outcome | % of conversations |
|:--------|-------------------:|
| Thankful response | 73% (333) |
| Support issue handled | 17% (79) |
| Mentioned plans to reorder | 13% (59) |
| Asked about new products | 9% (40) |
Categories overlap. A single conversation can appear in multiple rows.
Most conversations (73%) were straightforwardly positive: customers saying the product was great, thanking the team for checking in, or sharing how they were using the item. That matters because it means the outreach was welcomed, not tolerated. Customers did not treat the message as noise. They treated it as attention.
17% of conversations turned into support moments: exchanges for a different size, return labels, care instructions. These are issues that would normally require a customer to navigate a returns portal or open a helpdesk ticket. Instead, they got handled in the same thread.
The most commercially interesting finding: 13% of customers (59) volunteered plans to reorder, and another 9% (40) asked about products they hadn't bought yet. Restock dates, new colors, gifts for a spouse. These are buying signals that brands usually spend real marketing dollars to surface. Here they showed up organically, in a conversation the customer initiated, at zero incremental cost.
“This was truly the most pleasant return experience.”
— Emily, Quaker Marine customer
## What this means in dollars
The incremental 3-week repurchase rate for engaged customers was 5.0 percentage points (14.7% vs. 9.7%). That means for every conversation Signals starts that a customer replies to, there is a 5% chance it generates an additional order that would not have happened otherwise.
At scale, that adds up quickly.
**Incremental revenue per engaged conversation**
| Average order value | Revenue per conversation |
|:--------------------|-------------------------:|
| $100 | $5.00 |
| $200 | $10.00 |
| $500 | $25.00 |
And this only counts the first 3 weeks. The conversations also generated product feedback, support resolutions, and buying intent data that compounds over time.
This is revenue that required no discount, no ad spend, and no promotional campaign. It came from one well-timed message that customers wanted to answer.
## Why this mattered for Quaker Marine
Quaker Marine did not need to reinvent its brand voice. It already had a service ethos customers responded to. Signals gave the team a way to extend that feeling into the days after delivery, when most brands go quiet.
The result was stronger repeat purchase, more customer conversations, smoother support moments, and clearer buying intent, all from a message customers actually wanted to answer.
That is the bigger takeaway from this case study. Post-delivery does not have to be the end of the journey. For Quaker Marine, it became the start of the next one.
“Good customer service, plan on ordering more in future.”
— Tommy, Quaker Marine customer
Want results like these for your brand?
Signals helps apparel and lifestyle brands turn post-delivery check-ins into real customer conversations and measurable repeat purchase lift.
[Book a Demo](https://cal.com/alejandro-zaniolo/30min?overlayCalendar=true)
---
# The End of Reactive Support
> How proactive post-purchase engagement can prevent returns, convert refunds to exchanges, and build customer loyalty in e-commerce.
In 2010, a company providing customer support meant hiring people, setting up a call center, and dealing with high-churn, inconsistent reps. Especially in tech, many companies opted to forgo actual customer support (try getting a real reply from Google!) and just pushed people into figuring things out via FAQs and community forums. For e-commerce, this meant high costs, a poor customer experience, and a general trend towards "frictionless returns" with few clicks but little opportunity to actually improve the situation.
That scarcity shaped how customer experience was designed. Customer support became impersonal by necessity. The standard interaction was a support ticket or an email form, answered when someone got around to it. Customers learned to wait until they were genuinely frustrated before reaching out, because contacting support felt burdensome and unlikely to yield meaningful help. Brands, in turn, treated every incoming message as a cost to be minimized rather than a relationship to be maintained.
Now, AI is resetting the economics. Companies like [Sierra](https://sierra.ai/) and [Decagon](https://decagon.ai/) are building agents that can handle customer conversations and [take action](https://openai.com/index/decagon/) within real systems, the kind of work that used to require large teams and extensive training. Marginal interactions are getting cheaper and faster, and for the first time in e-commerce history, it is realistic to offer something that resembles "a helpful associate" to every customer, not just to those who complain the loudest.
And yet, most of the industry is rebuilding the same old thing with new machinery. Most implementations simply attach AI to the existing ticket queue. Brands celebrate faster response times and higher deflection rates, but the underlying model remains unchanged: customers must still initiate contact, describe their problem, and reach a threshold of frustration before anyone pays attention. The fundamental design of support, built around scarcity, persists even as the scarcity itself disappears.
The problem is that dissatisfaction rarely begins at the moment a ticket is created. The ticket is the final step, not the cause.
## Returns are a big problem.
The National Retail Federation (NRF) and Happy Returns projected that U.S. retail returns would total [$890 billion in 2024](https://nrf.com/media-center/press-releases/nrf-and-happy-returns-report-2024-retail-returns-total-890-billion), or 16.9% of annual sales. In the same report, 76% of consumers said free returns are a key factor in deciding where to shop. Many high-end apparel brands have returns rates higher than 30%.
These facts put retailers in a difficult position. Customers now expect returns to be easy and free, but each return directly erodes margin and consumes operational resources. When retailers respond by tightening policies, adding fees, shortening return windows, or pushing store credit instead of refunds, customers get justifiably upset. The brand pays for this later through lower repeat purchases and higher customer acquisition costs. NRF has also reported that [67% of consumers](https://nrf.com/media-center/press-releases/nrf-and-happy-returns-report-2024-retail-returns-total-890-billion) say a negative return experience would discourage them from shopping with that retailer again.
It is tempting to blame customers for high return rates, to talk about policy abuse and bracketing run amok. While this is sometimes accurate (many, many customers buy multiple sizes, generating unavoidable returns), a more useful framing for apparel is that most returns stem from incomplete information. The customer did not know how the garment would fit her body, how the color would look in her lighting, how the fabric would feel against her skin, or whether the small imperfection she noticed upon arrival was a normal variation or an actual defect. She discovers these answers at home, and in that moment, she forms her opinion of whether the brand understands her.
## In apparel, returns are often not driven by quality.
Apparel returns are usually not due to catastrophic problems. The leading cause is fit; the garment simply does not work on the customer's body. After that comes appearance, when the color or style looks different in person than it did online. [PowerReviews found](https://www.powerreviews.com/research/apparel-shopping-trends-2023/merchandise-returns/) that 39% of consumers return apparel because it doesn't fit, and 28% return it because it didn't look as expected. Some of the brands we talked to reported that fit was responsible for up to 60% of their returns.
This has a significant economic impact. [Radial estimates](https://www.radial.com/insights/returns-management-2024) that merchants pay an average of $27 to process a return for a $100 e-commerce order, and that only 30% of returned merchandise is resold, with the rest going to donation, liquidation, or disposal. The cost of a return extends far beyond the refund itself. Each returned item must be shipped back, received, inspected, and either repackaged for resale or written off entirely. The operational burden is significant, and much of the returned inventory never recovers its original value.
Apparel brands, therefore, face a difficult tradeoff. Generous, frictionless return policies reduce hesitation at checkout and support conversion. But if you wait until a customer has already decided to return, you have already lost. At that point, any outreach feels like resistance, and the customer walks away feeling that the brand cared more about its margins than about her.
That is why timing matters. The best moment to help is before the return decision crystallizes.
## What if the brand spoke first?
Signals exists because we believe the industry is focused on the wrong problem. The goal should not be to process tickets more cheaply. The goal should be to reach customers earlier, while their dissatisfaction is still forming, while they still want the item to work out, and while the brand can still offer help without sounding defensive. After all, the customer bought the item for a reason. Before she concludes it was a mistake, she is hoping it will work out.
Signals works by texting the customer after delivery, checking in like a real associate would, following up if they have not tried the item yet, and resolving issues in-thread with fit guidance, styling help, and tailored exchange recommendations. When customers share photos, we can quickly compare what they are experiencing against the product catalog to find solutions that actually work. This also helps with reverse logistics: we can recommend restocking when it makes sense, rework when the issue is fixable, and sometimes suggest the customer keep the item when shipping it back would cost more than anyone would gain.
### Consider the moment your existing systems miss.
A customer orders a dress for an upcoming event. It arrives a few days later, but life gets in the way, and she does not try it on right away. By the time she finally opens the package, her event is close, and she is already a little anxious. If the dress does not fit perfectly, her instinct will be to return it and find something else.
This is where proactive outreach changes the outcome. A simple check-in after delivery reminds her that someone is paying attention. If she mentions she has not tried it on yet, we follow up later. When she finally puts it on and notices that the neckline does not sit quite right, she does not have to navigate a returns portal or write a formal complaint. She can just reply to the text, describe what she is seeing, share a photo if it helps, and get a real answer. Maybe a small adjustment fixes the problem. Maybe the garment is designed to fit that way. Maybe she needs a different size or a different cut altogether. The point is that she gets help while she still wants the dress to work.
## Customers are skeptical of AI in customer service.
One reason proactive concierge matters is that customers are increasingly skeptical of automation in customer service, and for good reason. A lot of AI-first support has been deployed as a gatekeeper rather than a helper.
[Gartner reported](https://www.gartner.com/en/newsroom/press-releases/2024-07-09-gartner-survey-finds-64-percent-of-customers-would-prefer-that-companies-didnt-use-ai-for-customer-service) that 64% of customers would prefer companies not to use AI for customer service, and that 53% would consider switching to a competitor if they found out a company was going to use AI for customer service. But really, the concern is not that "AI exists"; rather, they fear that AI is just another obstacle between them and the real help they need.
Proactive concierge changes that emotional framing. The customer is not fighting their way into a system. The brand is showing up early and offering assistance in a channel that feels natural and low effort. If you do it well, it reads less like automation and more like attentiveness.
## The ROI story is not one story; it is three.
Brands often talk about returns as if the only win is "prevent the return." Prevention is valuable, but it is not the only lever that matters, and it is not always the right lever. At Signals, we think about three kinds of value: converting returns into exchanges, preventing returns by helping customers keep the item, and improving retention by making customers feel heard and helped.
First, many unhappy customers still want the item they bought; they just believe something is wrong with it. The fit is off, the color looks strange, the fabric feels different from what was expected. If you can find a version that actually works for them, you do more than convert a refund into an exchange and preserve revenue. You help them achieve what they wanted when they placed the order in the first place. They get an item they love, and they feel cared for by the brand.
Second, some returns are avoidable because the underlying issue is confusion, not dissatisfaction. The customer may need help adjusting the waistband, understanding how lighting changes the appearance of fabric, or recognizing what is normal variation for a material. When you provide that guidance early, the customer often keeps the item and feels like they have been helped. This is good for both the customer and for the brand, which protects margin by keeping the sale and eliminating return processing costs.
Third, there is retention, the outcome that many teams undercount because it does not appear on the day the return is created. [HBR cites Bain](https://hbr.org/2014/10/the-value-of-keeping-the-right-customers) suggesting that increasing customer retention rates by 5% increases profits by 25% to 95%, depending on the business. [Narvar reported](https://corp.narvar.com/blog/emotion-driven-ecommerce) that 70% of consumers say an easy return or exchange is likely to make them a repeat customer. Post-purchase resolution is one of the clearest moments when a brand can either earn another purchase or quietly lose it.
## The bet Signals is making.
E-commerce leaders have spent the last decade optimizing the front of the funnel, and it worked. Better PDPs, better merchandising, better checkout, faster delivery. But the customer's relationship with an apparel brand often begins after delivery, when they try something on and decide whether the brand understands them.
AI is making support cheaper. That part is obvious. What is still underappreciated is that when support becomes abundant, you can redesign the experience around timing rather than defense.
Signals is built on the belief that the next competitive advantage in apparel lies in proactive post-purchase engagement: reaching customers before they have firmly decided to return, when help can still feel like genuine care rather than an obstacle. The financial case rests on three outcomes: converting refunds into exchanges that preserve revenue, helping customers keep items they would have returned for minor or fixable reasons, and improving retention by turning a potentially negative experience into a positive one. The brand benefit is harder to measure but equally important. When customers feel that a brand is attentive and responsive, they are far more likely to come back. In apparel, where fit and personal preference introduce inherent uncertainty, that sense of being understood often determines whether a customer writes off the brand or gives it another chance.
The best time to help a customer is before they decide they need to return. That is what we do.
---
# SMS CX Shouldn't Be NoReply
> E-commerce made returns frictionless by eliminating the moment to help. Proactive SMS gives that moment back.
{ type: "incoming", text: "Hey, just checking in. Did everything arrive okay?" },
{ type: "outgoing", text: "Yep, got it. Haven't tried it on yet." },
{ type: "incoming", text: "Totally. Want me to follow up tomorrow?" },
{ type: "outgoing", text: "Yes please." },
{ type: "incoming", timestamp: "Next day", text: "Quick check, did you get a chance to try it?" },
{ type: "outgoing", text: "I did. The waist feels tight. Is that normal?" },
{ type: "incoming", text: "If you send a photo, I can help. Also, what size did you get?" },
{ type: "outgoing", content: "image", imageSrc: "/images/narrative/purple-leggings.jpg", imageAlt: "Customer photo of clothing fit" },
{ type: "incoming", text: "Got it! These run a bit snug. Want me to send the next size up, or would you prefer store credit?" },
{ type: "outgoing", text: "Next size please." },
{ type: "incoming", text: "Done. You'll get a shipping confirmation shortly." },
];
Here is a strange thing about modern commerce.
You buy something online. A text arrives: "Order confirmed." Another: "Shipped." Another: "Out for delivery." And then, at last: "Delivered."
Four texts. A thread on your phone. A relationship, of sorts.
And then? Silence. The thread goes dead. The number might as well be called NoReply, because that's what it is. Nobody is on the other side.
Why do we build a thread, and then abandon it?
Here's what I've come to believe: proactive SMS, done right, collapses all the channels into one thread. It solves problems we've been working around for decades.
## There's no moment to help.
Think about what e-commerce optimized for over the past decade. Frictionless returns. Prepaid labels, QR codes, drop-off locations, portals that ask minimal questions. The goal: make returning so easy that customers never hesitate to buy.
It worked. But by optimizing for frictionless returns, you've eliminated any moment to intervene.
A customer opens the box. She wants the item to succeed (that's why she bought it), but something feels off. In a store, this is the moment an associate walks over. "How's the fit? Want to try a size down?" That moment is where a return becomes an exchange, where a one-time buyer becomes a repeat customer.
Online, that moment doesn't exist. The customer is alone with her uncertainty, and the path of least resistance is the return portal you've made so beautifully frictionless. Returns flow smoothly, but you don't learn why. You don't get a chance to help. You just process the return and hope she comes back.
## Two things make proactive SMS different
The first is the channel itself.
SMS has properties that no other channel combines. It's synchronous when you need it fast, asynchronous when you need time. It handles photos and video natively. It has the best push notification and doesn't require an install. The support channel announces itself: the thread appears when the purchase is made.
The second thing is more important: proactive outreach at this moment is perceived as care.
Most people assume "intervention = friction" because they imagine a pop-up blocking the return flow. That would be friction. But texting someone after delivery to ask how things fit? That's a store associate walking over to help.
Here's what that looks like in a good store. A customer comes in with a return. The associate doesn't stay behind the desk. She comes around to the customer's side and says: *"Let's find something you're going to love."*
The customer feels it. Even if she still returns the item, she walks out thinking: *They're nice there.* That's the experience that disappeared online, not because brands don't want it, but because they couldn't scale it.
One retailer told us about the early days when they were smaller: *"When we saw an order come through for a size 6, 8, and 10, someone would reach out and say, 'Let me tell you how this runs.' But that doesn't happen anymore. There are too many emails. We're too big."*
Proactive SMS shrinks the distance back down. Every customer gets someone paying attention.
The timing is everything. We wrote about this in [The End of Reactive Support](/blog/return-starts-before-return): the customer bought the item for a reason. Before she concludes it was a mistake, she's hoping it works out. She's uncertain, open, persuadable. That's the moment to show up with help.
*"Let us know how it fits. We're here if you need anything."*
That's an invitation. When the customer replies, "The waist feels tight, is that normal?", she's using a thread that works. Frictionless returns remove barriers, but they don't create connections. Proactive SMS does both: it lets you help without being in the way.
And yet most brands throw this away. The delivery text arrives from a short code that never responds. The customer learns the thread is a billboard.
If you're going to build a thread, the thread has to work.
## The channel that's both synchronous and asynchronous
What channel can feel like live chat when you need it fast, and like email when you need time to think?
Email is asynchronous, sure, but it's not reliably *present*. Your message competes with newsletters, receipts, spam, and whatever the inbox algorithm decided to bury today. Web chat is present, but only when the customer is on your site. Close the tab? Conversation's gone.
SMS works both ways.
Reply right now? Feels like chat. Reply 6 hours later? Still feels normal. No guilt, no "sorry for the delay," no ticket to reopen. Just a thread. The same thread.
This sounds small. It's the whole point. SMS collapses the other channels into one because it adapts to the moment. No "stay on the line" pressure (unlike web chat). No "did they even see this?" anxiety (unlike email). The customer engages on their timeline. The brand responds when it matters.
Channel comparison:
| | **SMS** | Email | Web Chat | App |
| -----------: | ------: | ----: | -------: | -----: |
| Sync | **Yes** | No | Yes | Yes |
| Async | **Yes** | Yes | No | Partial |
| Photos | **Yes** | Clunky | Varies | Yes |
| No install | **Yes** | Yes | Yes | No |
| Push | **Lock screen** | Buried | Tab-based | 33% opt-in |
| Self-announcing | **Yes** | No | No | No |
## The best push notification is the one that doesn't require an app
Push notifications. Everyone wants them. Think about what it takes to get there.
For app push, the customer has to install your app, create an account, and grant notification permissions. OneSignal's benchmarks for Shopping apps: [33% opt-in on iOS, 36% on Android](https://onesignal.com/mobile-app-benchmarks-2024). If your post-purchase engagement strategy lives in your app, 2/3 of customers will never see it.
For email, you can reach more people, but can you *reach* them? Apple's Mail Privacy Protection has made [open tracking unreliable](https://www.twilio.com/en-us/blog/insights/apple-mail-privacy-protection). Validity has documented how Apple's prefetch behavior [shifted pixel firing patterns entirely](https://www.validity.com/blog/case-closed-the-mystery-of-declining-email-open-rates/). Even good emails get buried.
The lock screen is the most-viewed interface in modern life. Asurion research found Americans reach for their phones [352 times per day](https://www.asurion.com/connect/news/tech-usage/). SMS is the only way to land there reliably.
## “Show me” beats “describe it.”
Support is visual. Especially in commerce.
*Is this a defect or normal variation?*
*Does this fit right?*
*Is this color off, or is it my lighting?*
*Did I install this correctly?*
*Is this stain something I can fix?*
Force customers to describe these things in words, and you get worse information. Slower resolution. More back-and-forth. Let them show you? One photo. Done.
The messaging inbox handles this naturally. Tap the camera. Take the shot. Send. No upload portal. No "please attach files in PNG or JPEG format under 5MB." Just the way humans already communicate.
## What surprised us in the first pilot
I went in skeptical.
I assumed only people with problems would reply. I assumed most customers would ignore us. I assumed we'd get a trickle of complaints.
I was wrong.
40% of buyers replied within 24 hours. Among those who needed a return, the number was 100%.
Not a typo. Most customers replied, quickly. The vast majority of replies were positive, expressing satisfaction with the item. Every customer who needed a return used the SMS thread. Zero used the portal.
Here's the kind of thing we see. It matters precisely because it isn't dramatic:
That's a conversation. The kind that used to happen in stores, between a customer and an associate who actually cared. Except now it happens in the default inbox, after delivery, at the exact moment the customer is forming an opinion.
The key is the first message. We don't just text "delivered." We text with *intent*: "Let us know how it fits. We're here if you need anything." That turns a notification into an invitation.
## Support as a sensor network
Here's the thing about frictionless returns: they don't teach you anything.
A customer returns something. You get a dropdown: "Didn't fit." Maybe "Changed my mind." That's it. No nuance, no signal you can act on.
Traditional support data is the same. It only shows you the extremes: the angriest customers (the ones motivated enough to fight through to a human) and the happiest (the ones who leave reviews). The middle stays silent. And the middle is where most of your product truth lives.
Two-way SMS pulls the middle into the conversation.
The cost of saying something drops to almost nothing. A customer doesn't have to decide "Is this worth opening a ticket?" She just replies. And when she replies, you learn: which SKUs create confusion (not just which ones break), which product photos are misleading, which size guidance is missing, which packaging detail creates a bad first impression.
That's signal you can actually act on. The kind that lets you fix causes instead of processing returns more efficiently.
## Timing, not volume
Let me be clear about what this is *not*. It's not "blast more texts." People hate that, and they should opt out.
This is about one thread. One durable, trusted thread that starts at delivery (because that's when uncertainty begins) and stays alive because replying actually works.
Zendesk's data backs this up: 72% of customers want immediate service. Customers are [2.4x more likely](https://www.zendesk.com/blog/customer-experience-statistics/) to stick with a brand when problems are solved quickly. And [64% will spend more](https://www.zendesk.com/blog/customer-experience-statistics/) if you resolve issues *where they already are*.
"Where they already are." That phrase is doing a lot of work. The messaging inbox is the clearest example of "already there" that exists.
That's why NoReply is such a trap. If your first SMS interaction teaches customers the thread is dead, you don't get a second chance. You've trained them to go back to the maze.
## Why this wouldn't work with only humans
You can't do this with humans alone.
Reaching out to every customer after delivery. Following up the next day. Answering questions at scale. Handling photos. Remembering context. The math doesn't work. The labor cost would be prohibitive. So historically, brands just... didn't do it.
AI changes the economics. But AI *as gatekeeper* is the real problem: the bot that loops, the automation that deflects, the systems designed to avoid talking to you. Customers don't hate AI. They hate being trapped in automation when they need real help.
SMS cracks this open because it enables seamless transitions between AI and humans.
Web chat has a speed-of-response SLA. If a human takes over, they need to respond *now*. The customer is sitting there, waiting. That makes handoffs expensive and stressful.
SMS doesn't have that pressure. Someone can reply in 2 minutes or 2 hours, and it still feels normal. The AI handles the routine (check-ins, context gathering, photo interpretation) and hands off to humans when things get complex. The customer never feels the seam.
That's what makes it work: AI and humans together, in one channel, transitioning seamlessly because the channel's timing norms allow it.
## What I believe now
I used to think of SMS as "one more channel." I was wrong.
Two things make proactive SMS different. The channel itself: sync when you need speed, async when you need time, photos and video native, the best push notification without requiring an install, and seamless AI-to-human handoffs because there's no "stay on the line" pressure.
And the timing. Proactive outreach after delivery is perceived as care. The customer still wants the item to work. Checking in feels like a store associate. Help that arrives before you've given up feels like attention.
When customers feel that attention, they reply. Even when nothing is wrong. They talk. And when they talk, you learn: which products confuse, which photos mislead, which sizing guidance is missing. You fix causes instead of processing returns.
If you're texting customers today, you already have this surface. The thread already exists. The only question is whether you'll treat it like a billboard or make it a place where help actually lives.
---
# What apparel retailers learn when customers can text back after delivery
> Signals data from thousands of post-delivery conversations shows why ordinary customer replies can become service recovery, repeat purchases, and usable retail intelligence.
{
label: 'Positive confirmation',
value: 65,
detail: '"Love it," "fits great," "thanks"',
},
{
label: 'Repeat-buy signal',
value: 13,
detail: '"I plan to buy a shirt next"',
},
{
label: 'New purchase or catalog question',
value: 7,
detail: '"What jacket do you recommend?"',
},
{
label: 'Exchange request',
value: 7,
detail: '"Different size or color please"',
},
{
label: 'Fit or sizing complaint',
value: 6,
detail: '"The fit is off"',
},
{
label: 'Return or refund request',
value: 5,
detail: '"I would like to send it back"',
},
{
label: 'Other feedback',
value: 4,
detail: 'Product suggestions, brand love, feature requests',
},
{
label: 'Gift or recipient context',
value: 4,
detail: '"Bought it for my husband"',
},
{
label: 'Product question',
value: 3,
detail: 'Care, styling, shrinkage, materials',
},
{
label: 'Shipping or delivery issue',
value: 3,
detail: 'Lost, wrong, or damaged order',
},
{
label: 'Product quality defect',
value: 2,
detail: 'Button broke, arrived stained',
},
{
label: 'Pricing or billing',
value: 1,
detail: 'Price adjustment, invoice, discount',
},
]
{
label: 'In-thread color or repeat-item expansion',
value: 33,
detail: 'Customer loves the item and wants another version.',
},
{
label: 'Cross-sell from recommendation',
value: 14,
detail: 'Customer asks what else the brand recommends.',
},
{
label: 'Gifting expansion',
value: 12,
detail: 'Happy customer buys for a spouse, parent, sibling, or friend.',
},
{
label: 'Replacement after sizing miss',
value: 9,
detail: 'Wrong size is kept for someone else, correct size is ordered.',
},
{
label: 'Fulfillment error into incremental sale',
value: 9,
detail: 'Wrong item arrives, customer keeps it and pays for the right one.',
},
{
label: 'Restock or waitlist to same-thread buy',
value: 9,
detail: 'Back-in-stock note triggers an immediate order.',
},
{
label: 'Price-adjust to second-item purchase',
value: 7,
detail: 'Goodwill refund creates enough trust for another order.',
},
{
label: 'AI-assisted confident new purchase',
value: 7,
detail: 'Product expertise gives the customer confidence to buy.',
},
]
{ label: 'Exchange request', value: 39 },
{ label: 'Pricing or billing', value: 33 },
{ label: 'Shipping or delivery issue', value: 24 },
{ label: 'Fit or sizing complaint', value: 14 },
{ label: 'Product question', value: 13 },
{ label: 'Positive confirmation', value: 12 },
{ label: 'Return or refund request', value: 9 },
{ label: 'Gift or recipient context', value: 8 },
{ label: 'Product quality defect', value: 7 },
]
Signals runs post-delivery iMessage conversations for apparel brands. In randomized client trials,
those conversations produced a [51% average lift](/blog/quaker-marine-case-study/) in engaged-customer repeat purchase rates. In this blog we'll dig into how helpful intent engagement after item is delivered turns into additional purchases by the customer.
## Conversation mix
Post-delivery threads usually sort into 3 buckets: positive affinity, product-focused questions,
and experience issues. In Signals production data, most conversations land in the first bucket.
Positive confirmation alone is 65% of engaged replies. "Love it," "fits great," and "thanks for
checking in" usually vanish inside ecommerce systems because they don't open a ticket or trigger a
return. Signals captures them as structured signals instead. A customer who volunteers "I'm also
looking for a jacket for the summer" should be a CRM pre-lead!
Product-focused categories, including exchanges, fit complaints, returns, product questions, and
quality defects, account for about 23% of engaged conversations. Most resolve in 1 or 2 turns driven by the AI,
which lets the support team spend its attention on the long-tail cases that genuinely need a human.
Experience issues are a smaller share, with shipping at 3% and pricing or billing at 1%. But they
matter out of proportion to their volume. A brand doesn't need many botched deliveries before the
customer's memory of the order becomes a memory of the recovery.
Note: The chart can exceed 100% because 2 useful signals can appear in one conversation. Customers text
like people, with multiple ideas in one reply.
## How new sales start inside the thread
What's most interseting is that many post-delivery conversations become purchases inside the same thread.
The biggest mechanism is simple: the customer likes the item and wants more of it. In-thread color
or repeat-item expansion accounts for 33% of thread-originated sales, cross-sell from
recommendation for 14%, and gifting expansion for 12%. The gift-adjacent share grows once you fold
in sizing-miss cases, where the original item stays with someone in the household and the correct
size gets ordered separately.
None of this requires pushing the customer into a campaign flow. The check-in is about the
delivered order, and buying intent appears because the customer is happy, the thread is alive, and
the brand is right there while the customer is thinking about the product.
This is a real commercial moment: a known customer, with a fresh product experience, asking the
brand what to buy next. Retailers spend heavily to infer that from browsing behavior. In the thread,
the customer just says it.
The useful part is how concrete these moments are. The customer is holding the product, and the
thread already knows what they bought. A passing mention of another color or a size problem or a
restock wish lets the next step happen right inside the conversation.
## Repeat purchases after service moments
The quieter mechanism is what happens after conversations that never mention a next purchase. These
are customers who talk to Signals about an exchange, a delivery issue, a fit problem, or a product
question, then buy again within 30 days anyway.
Exchange requests converted at 39% in the Signals data, pricing and billing threads at 33%, and
shipping and delivery issues at 24%. Even fit and sizing complaints, which look like the unhappiest
threads on the list, converted at 14%. Service moments are also moments when the customer is still
reachable.
Fit issues matter more in apparel than anywhere else. A customer who says "the sleeves are short"
opens a support path and hands the brand a SKU-level fit signal in the same breath, which sharpens
size guidance and gives the team a real shot at saving the sale before the return portal becomes
the easiest path.
Product questions matter too. Customers who asked about care, shrinkage, styling, or materials
repurchased at 13% within 30 days, slightly ahead of positive confirmations. Product expertise
builds loyalty because it helps the customer enjoy the thing they already bought.
## Working on the customer's clock
Signals' median AI reply time is 26 seconds. If a customer asks a question at 11pm, the thread can
answer at 11:00:26.
That matters because 62.5% of customer replies arrive outside Monday through Friday, 9 to 5 local
time. Apparel gets evaluated after work, in a bedroom mirror, on a Sunday morning, or after someone
asks their spouse what they think. The customer experience runs on the customer's clock, not the
brand's.
When a conversation needs a human, the handoff stays inside the same thread, with the order context
and prior messages still visible. This is very unlike any other channel where the customer has to go to
a portal but once they close the tab the conversation disappears.
That continuity is where a lot of the value lives. The AI can cover the broad post-delivery layer,
and human teams can step in for the exceptions that deserve judgment, inventory work, or a more
careful service recovery.
## What this means for the brand
The post-delivery thread is not just a support channel. It is one of the few places where an
apparel brand actually gets to talk to its customers right after they have tried the product.
Customers tell us what they want to buy next. They tell us who they bought it for. They tell us
when the sleeves are too short, when a sold-out item is still on their mind, and what they would
need to feel good about buying again. All information that would not have reached the brand otherwise.
The thread gives them that place. The brand hears it, the AI handles most of it, and the human
team gets to focus on the conversations that really need them. The repeat purchases follow from
there.
---
# Messaging Compliance: Signals
> How Signals handles customer messaging compliance across iMessage, RCS, SMS, 10DLC SMS, and WhatsApp.
Last Updated: July 3, 2026
Material Model, Inc., doing business as Signals ("Signals", "we", "us", "our"), provides customer messaging infrastructure for ecommerce brands. This page explains how we separate customer support from marketing, how we handle consent, and how customers can stop messages.
This page is a public summary of our operating policy and general compliance posture. Brands that use Signals remain responsible for their own legal review, notices, customer-facing terms, and consent collection where required.
## 1. Channels We Support
Signals' core service is an iMessage customer messaging service. When iMessage is unavailable, the conversation may fall back to RCS or SMS. We also support 10DLC SMS and WhatsApp for brands and customers that prefer those channels.
The same customer thread may support order help, product questions, returns and exchanges, delivery issues, feedback, styling questions, and purchase assistance. The compliance treatment depends on why the message is sent and what channel is used.
## 2. Support and User-Initiated Messages
Our primary use case is customer support and responding to user-initiated queries. A customer may ask about an order, a delivery, an item, a service, a return, an exchange, a brand policy, or a future purchase. Signals responds to that request with the relevant context from the brand's catalog, policies, and connected systems.
When a customer asks for styling advice, product matching, restock help, or help buying an item, we treat that as customer-initiated purchase assistance. Catalog and recommendation tools stay available for those requests. When a customer message includes support intent, such as a return, exchange, refund, damaged item, missing package, fit problem, or other issue, the support need takes priority before any optional product suggestion.
Some brands use Signals solely for support and customer-initiated conversations. Unsolicited marketing is optional and can stay disabled for a brand's program.
## 3. How Messages Are Sent
Signals' main service is to send and reply to messages using our AI agent. We also support human-sent messages, such as CX escalations, and messages sent by a customer's own AI agent through Signals MCP.
When a brand sends, initiates, or causes a message through Signals, including through our dashboard, API, or MCP connector, that brand is responsible for making sure the message content, purpose, timing, consent basis, and recipient selection comply with applicable law, carrier rules, platform rules, and the brand's own customer-facing terms.
## 4. Brand-Initiated Marketing Sends
For certain brands, Signals can send brand-initiated messages to highly engaged customers. These can include campaign sequence steps, proactive review asks, unsolicited cross-sell suggestions, and contextual purchase follow-ups that are not a direct response to the customer's latest request.
Those sends are treated as marketing sends. They are gated separately from ordinary support replies and user-initiated purchase help.
## 5. Consent Rules by Channel
For SMS, 10DLC SMS, and RCS marketing sends, we require [explicit written consent](https://www.ecfr.gov/current/title-47/chapter-I/subchapter-B/part-64/subpart-L/section-64.1200) before sending brand-initiated marketing content.
For iMessage marketing sends, we follow [FCC 19-73](https://docs.fcc.gov/public/attachments/fcc-19-73a1.pdf) and the text-message definition in [47 U.S.C. § 227(e)(8)(C)(iii)](https://uscode.house.gov/view.xhtml?edition=prelim&num=0&req=granuleid%3AUSC-prelim-title47-section227), the federal statute commonly known as the Telephone Consumer Protection Act (TCPA). FCC 19-73 excludes non-SMS and non-MMS messages sent over IP-enabled messaging services such as iMessage, WhatsApp, and Skype to other users of the same service from the FCC rule definition discussed there. Section 227(e)(8)(C)(iii) similarly excludes messages sent over an IP-enabled messaging service to another user of that same service, except SMS and MMS messages. Signals still applies additional state-level protections before brand-initiated iMessage marketing is allowed.
For customers with a restricted-state signal in Connecticut, Washington, California, Virginia, Florida, Maryland, Oklahoma, or Texas, Signals requires full written consent before brand-initiated marketing over iMessage or WhatsApp too. We treat a restricted-state signal as present when we see it from a phone area code, order shipping state, or order billing state.
Brands can use Signals even when they do not collect broad marketing consent. In those cases, Signals can still provide support and respond to user-initiated conversations, but brand-initiated marketing without explicit written consent is limited to confirmed iMessage conversations with no restricted-state signal.
## 6. How the Marketing Gate Works
Signals evaluates marketing-send eligibility before non-user-initiated marketing flows. A marketing send is allowed when the customer has explicit SMS marketing consent, or, for iMessage only, when the latest provider evidence confirms iMessage service and the customer has no restricted-state signal.
If the channel is SMS, RCS, WhatsApp, or unknown, the send requires explicit marketing consent. If provider evidence is missing, we fail closed and treat the route as non-iMessage. If consent evidence says the customer is unsubscribed, the send is blocked.
Signals can read consent evidence from commerce and messaging consent systems, including Shopify and Listrak. We also maintain our own suppression records. Suppression and unsubscribe evidence wins over subscribed evidence.
## 7. WhatsApp
Signals uses WhatsApp through the Meta Business messaging ecosystem. When a customer messages a brand on WhatsApp, the platform supports [free-form service replies inside the 24-hour customer service window](https://developers.facebook.com/documentation/business-messaging/whatsapp/messages/send-messages). Outside that window, brand-initiated messages must use [approved templates](https://developers.facebook.com/documentation/business-messaging/whatsapp/templates/overview) where required by Meta's rules.
For WhatsApp marketing outreach, Signals applies consent controls and the restricted-state protections described above. Where no written marketing consent is available, Signals does not use WhatsApp for brand-initiated marketing outreach. We also follow platform requirements for approved templates and customer-service-window behavior.
## 8. Opt-Outs and Suppression
Customers can opt out at any time. Signals supports [standard STOP-style keywords](https://api.ctia.org/wp-content/uploads/2023/05/230523-CTIA-Messaging-Principles-and-Best-Practices-FINAL.pdf), including STOP, STOPALL, UNSUBSCRIBE, CANCEL, END, and QUIT. These are handled immediately, case-insensitively, and without sending the request through the normal AI response path.
Signals also uses an LLM-based admission hook to detect ordinary-language stop requests, such as "please no more messages", "stop texting me", "don't contact me again", or "remove me from these texts". When detected, we block the conversation, suppress the phone number, and cancel pending proactive follow-ups.
START, SUBSCRIBE, and UNSTOP can remove an active suppression where resubscription is allowed.
## 9. Customer Outreach Preferences
Some customer replies are preferences rather than legal STOP requests. For example, a customer may say they only want order updates, do not want campaign follow-ups, or prefer fewer reminders. Signals can store a contextual outreach preference note and re-check it before future proactive outreach.
When that stored preference conflicts with a planned proactive send, the send is skipped. Important operational or customer-requested support replies are not suppressed unless the preference clearly blocks them.
## 10. Auditability
Signals records message route evidence, consent evidence, suppression state, blocked marketing decisions, and skip reasons. Marketing workflows are checked before sending, and the SMS send path includes a defense-in-depth backstop so a marketing send that bypasses an upstream gate is still blocked before provider delivery.
Campaign step decisions can record the marketing eligibility reason. Review requests, proactive cross-sell, and contextual purchase follow-ups are also gated, and blocked attempts are logged for operational review.
## 11. Brand Responsibilities
Brands can use Signals for support and user-initiated conversations even if they do not collect marketing consent. Before asking Signals to send marketing content on routes that require consent, brands must collect and maintain that consent. This includes clear opt-in language, required disclosures, accurate consent records, and a working privacy policy and terms where required.
When written marketing consent is not available, Signals limits brand-initiated marketing to confirmed iMessage conversations with no restricted-state signal.
Brands may not upload or use rented, sold, or shared opt-in lists with Signals. Brands must also keep connected consent systems accurate and notify us if they need stricter rules for a particular program, geography, channel, or customer segment.
## 12. Contact
For compliance questions, contact [hello@returnsignals.com](mailto:hello@returnsignals.com).
For privacy requests, contact [privacy@returnsignals.com](mailto:privacy@returnsignals.com).
---
# Privacy Policy: Signals
> How Signals collects, uses, and protects personal information when providing customer messaging services for ecommerce brands.
Last Updated: July 3, 2026
Material Model, Inc., doing business as Signals ("Signals", "we", "us", "our"), is committed to protecting privacy and securing personal information. This Privacy Policy explains how we collect, use, disclose, and safeguard information when we provide customer messaging services for ecommerce brands.
Signals is a business-to-business service. We process information for our business customers and their end customers through iMessage, RCS, SMS, 10DLC SMS, WhatsApp, and related commerce, support, analytics, and messaging systems.
For how we handle messaging consent, marketing sends, opt-outs, and channel-specific compliance controls, see our [Messaging Compliance](/compliance) page.
## 1. Information We Collect
### 1.1 Information Business Customers Provide
- Account information: name, email address, company name, phone number, role, and authentication information
- Billing and contract information: plan, subscription, payment, invoicing, and procurement details
- Brand configuration: catalog, product, policy, return, exchange, shipping, review, support, automation, and escalation settings
- Integration data: connected commerce, helpdesk, messaging, consent, analytics, fulfillment, return, and warehouse systems
- Team communications: support tickets, feedback, emails, calls, and other correspondence
### 1.2 End Customer Information
We may process end customer information on behalf of a business customer, including:
- Contact information: name, phone number, email address, and messaging handle
- Order information: order IDs, order names, products, variants, delivery status, tracking, return or exchange status, shipping region, and billing region
- Conversation content: inbound and outbound messages, attachments, images, reactions, message metadata, and thread history
- Support and purchase context: customer questions, return or exchange requests, product preferences, styling questions, purchase intent, feedback, reviews, and escalation notes
- Consent and compliance data: marketing consent state, opt-in or opt-out evidence, suppression status, unsubscribe requests, outreach preferences, and message route evidence
### 1.3 Information Collected Automatically
- Usage data: features used, conversations processed, response outcomes, workflow outcomes, and automation events
- Device and log data: browser type, operating system, IP address, access time, pages viewed, and clickstream data
- Analytics data: performance metrics through tools such as Google Analytics 4, PostHog, and Datadog RUM
- Cookies and similar technologies: session management, preference storage, security, and analytics
## 2. How We Use Information
We use information to:
- Provide, operate, secure, and improve Signals
- Send and receive customer messages across supported channels
- Respond to customer support, order, delivery, return, exchange, product, and purchase-assistance requests
- Generate AI-assisted replies, recommendations, workflow steps, summaries, classifications, and escalations
- Route conversations to humans when needed
- Process opt-outs, consent evidence, outreach preferences, and other compliance controls
- Measure conversation outcomes, support resolution, revenue impact, and product feedback
- Maintain billing, fraud prevention, audit, logging, security, and account administration
- Send service, security, administrative, and transactional notices
- Send marketing communications where permitted and with required consent
- Comply with legal obligations and enforce our Terms of Service
- Improve our products, services, technologies, and algorithms, including with aggregated or de-identified data
## 3. AI and Messaging Providers
Signals uses AI and infrastructure providers to process conversation content, commerce context, and workflow state. These providers may include cloud infrastructure, language model providers, messaging providers, analytics tools, and commerce or support integrations.
Where available, we use strict data protection terms and controls, including zero data retention terms for supported AI processing. AI providers are not permitted to use customer data to train their general models unless a separate agreement expressly allows it.
## 4. Data Sharing and Disclosure
We do not sell personal information. We may share information:
- With service providers that help operate Signals, under confidentiality and data protection obligations
- With a business customer's connected systems, such as commerce, support, messaging, consent, warehouse, analytics, and return systems
- With human operators or authorized brand personnel when a conversation escalates or requires review
- When required by law, subpoena, court order, or other legal process
- To protect our rights, property, safety, users, customers, or the public
- With explicit consent or at the direction of the relevant business customer
- In connection with a merger, acquisition, financing, restructuring, or sale of assets
## 5. Messaging Compliance and Opt-Outs
Signals processes opt-outs and suppression requests. Standard STOP-style keywords are handled immediately, and ordinary-language requests to stop future text or customer messages may also block future outreach.
Signals may process marketing consent state, opt-in evidence, opt-out evidence, customer outreach preferences, message channel evidence, order geography, and related compliance metadata. For more detail, see [Messaging Compliance](/compliance).
## 6. Data Security
We implement security measures designed to protect information, including:
- TLS encryption for data in transit
- Encryption for stored data
- Private cloud infrastructure on Google Cloud Platform
- Role-based access controls and multi-factor authentication
- Audit logging and monitoring
- Vulnerability management, backups, and incident response procedures
For more detail, see our [Security](/security) page.
## 7. Data Retention
We retain information for as long as needed to provide Signals, comply with legal obligations, resolve disputes, enforce agreements, maintain security, and support legitimate business purposes.
Typical retention categories include:
- Account data: for the account term and a reasonable period afterward
- Conversation data: based on business customer settings, operational needs, and legal requirements
- Media attachments: based on configurable retention policies where available
- Consent, suppression, and audit records: as needed to honor opt-outs, prove compliance, and maintain system integrity
- Billing and financial records: as required by law
- Aggregated or de-identified data: may be retained indefinitely
You may request deletion of identifiable data by contacting [privacy@returnsignals.com](mailto:privacy@returnsignals.com). Business customers may submit end customer deletion requests on behalf of their end customers. Deletion requests do not require us to delete aggregated or de-identified data, or records we must retain for legal, security, billing, or compliance purposes.
## 8. Your Privacy Rights
Depending on your location, you may have rights to:
- **Access:** Request a copy of personal information we hold
- **Correction:** Update or correct inaccurate information
- **Deletion:** Request deletion of personal information
- **Portability:** Receive data in a structured, machine-readable format
- **Restriction:** Limit how we process information
- **Objection:** Object to certain processing
- **Withdrawal:** Revoke consent where processing is based on consent
To exercise these rights, contact [privacy@returnsignals.com](mailto:privacy@returnsignals.com). We will respond as required by applicable law.
## 9. Cookies and Tracking Technologies
We use cookies and similar technologies to support:
- Essential site and platform functionality
- Security and fraud prevention
- Preferences and session management
- Analytics and performance measurement
You can control cookies through your browser settings. Disabling certain cookies may limit functionality.
## 10. International Data Transfers
Information may be transferred to and processed in the United States and other countries where we or our providers operate. We use appropriate safeguards for international transfers where required.
## 11. Children's Privacy
Signals is not intended for individuals under 18. We do not knowingly collect personal information from children. If we learn that we collected information from a child, we will take appropriate steps to delete it.
## 12. California Privacy Rights
California residents may have additional rights, including the right to know what personal information is collected, used, shared, or sold; the right to delete personal information; the right to opt out of sale or sharing where applicable; and the right not to be discriminated against for exercising privacy rights.
We do not sell personal information. To make a California privacy request, email [privacy@returnsignals.com](mailto:privacy@returnsignals.com).
## 13. European Privacy Rights
Signals is currently offered to businesses located in the United States. If European privacy laws apply to particular processing, we will process information under an appropriate legal basis, such as contract performance, legitimate interests, consent, or legal obligation.
## 14. Third-Party Links and Services
Signals may connect to third-party websites, platforms, and services. We are not responsible for the privacy practices of those third parties. Business customers should review the privacy and compliance terms of their connected systems.
## 15. Changes to This Policy
We may update this Privacy Policy periodically. We will post the updated policy with a new "Last Updated" date. When required, we will provide additional notice.
Continued use of Signals after changes become effective constitutes acceptance of the updated policy.
## 16. Contact Information
For privacy-related questions, requests, or concerns, contact us:
**Material Model, Inc. (d/b/a Signals)**
2261 Market Street STE 85311, San Francisco, CA 94114
Email: [privacy@returnsignals.com](mailto:privacy@returnsignals.com)
Website: [www.returnsignals.com](https://www.returnsignals.com)
For general inquiries, contact [hello@returnsignals.com](mailto:hello@returnsignals.com).
---
# Security: Signals
> How Signals protects customer messaging data with enterprise-grade infrastructure, encryption, and privacy-first analytics.
Last Updated: July 3, 2026
Signals is committed to protecting customer messaging data with enterprise-grade security practices. Our infrastructure, built on Google Cloud Platform, implements defense-in-depth principles to safeguard information at every layer.
## 1. Infrastructure Security
Our platform is built on industry-leading security infrastructure with multiple layers of protection:
### 1.1 Encryption
- **Data in Transit:** TLS 1.3 encryption for all connections
- **Data at Rest:** AES-256 encryption for all stored data
- **HTTPS Only:** Automatic redirect from HTTP to HTTPS
- **Managed SSL Certificates:** Automatically renewed and maintained
### 1.2 Private Cloud Infrastructure
- **Google Cloud Platform:** Enterprise-grade infrastructure with SOC 2 Type II compliance
- **Virtual Private Cloud (VPC):** Isolated network environment with strict access controls
- **No Public Access:** Backend systems are not exposed to the public internet
- **Dedicated Infrastructure:** Resources allocated specifically for our services
### 1.3 Access Control
- **Multi-Factor Authentication:** Required for all administrative access
- **Role-Based Access Control (RBAC):** Least privilege principle for all team members
- **Audit Logging:** Comprehensive logging of all system access and changes
- **Regular Access Reviews:** Quarterly reviews of access permissions
### 1.4 Network Security
- **DDoS Protection:** Google Cloud Armor for distributed denial of service mitigation
- **Web Application Firewall:** Protection against OWASP Top 10 vulnerabilities
- **Rate Limiting:** API rate limiting to prevent abuse
- **IP Allowlisting:** Available for enterprise customers
## 2. Customer Data and Messaging Security
Signals processes sensitive customer data through iMessage, RCS, SMS, 10DLC SMS, WhatsApp, and connected commerce and support systems. We implement strict controls to protect this information:
### 2.1 Messaging Security
- **Encryption:** Message data is encrypted in transit and at rest
- **Provider Controls:** Messaging providers are selected and configured for production-grade delivery, webhook validation, and access controls
- **Route Evidence:** Message route metadata records the channel and provider evidence used for operational and compliance decisions
- **PII Protection:** Customer phone numbers and personal data encrypted with AES-256
### 2.2 Media Upload and Storage Security
- **Secure Upload:** TLS 1.3 encryption for media uploads
- **Content Validation:** Automated scanning to prevent malicious file uploads
- **Encrypted Storage:** Media encrypted at rest
- **Access Controls:** Attachments accessible only to authorized systems and personnel
- **Automatic Deletion:** Media deleted according to configurable or operational retention policies
### 2.3 Brand Data Isolation
- **Multi-Tenant Architecture:** Complete data isolation between brands
- **Role-Based Access:** Strict access controls ensure brands only see their own data
- **API Security:** OAuth 2.0 and API key authentication with rate limiting
- **Data Residency:** Options for geographic data storage requirements
### 2.4 Compliance Records
- **Consent Evidence:** Marketing consent, suppression, and opt-out evidence are stored for auditability
- **Suppression Controls:** STOP-style and ordinary-language opt-outs block future outreach
- **Marketing Gate Logs:** Blocked marketing sends and skip reasons are logged for operational review
- **Channel Evidence:** Provider service evidence supports channel-specific compliance decisions
## 3. Data Retention & Privacy
We implement data minimization principles and retain data only as long as necessary:
- **Messaging Conversations:** Retained according to business customer settings, operational needs, and legal requirements
- **Media Attachments:** Deleted based on configurable or operational retention policies
- **Analytics Data:** Aggregated analytics retained to provide trend insights to brands
- **Contact and Consent Information:** Retained as needed to provide the Service, honor opt-outs, and maintain compliance records
- **Security Logs:** Retained as required for security monitoring and compliance
### 3.1 Data Deletion Requests
You can request deletion of your personal data at any time by contacting security@returnsignals.com. We will:
- Acknowledge your request within 48 hours
- Complete deletion within 30 days
- Provide confirmation once deletion is complete
- Retain only data required by law or legitimate business purposes (e.g., financial records)
### 3.2 Automated Backups
- Daily automated backups with 14-day retention
- All backups encrypted with AES-256
- Backups stored in separate geographic regions for disaster recovery
- Regular backup restoration testing
## 4. Compliance & Standards
We adhere to industry-standard security frameworks and compliance requirements:
### 4.1 SOC 2 Type II Compliance
Signals, as part of Material Model, is working toward SOC 2 Type II certification. This includes:
- Security controls for protecting customer data
- Availability and performance monitoring
- Processing integrity verification
- Confidentiality protection measures
### 4.2 Security Best Practices
- **Regular Security Audits:** Quarterly internal security assessments
- **Vulnerability Scanning:** Automated scanning for known vulnerabilities
- **Penetration Testing:** Annual third-party penetration tests
- **Dependency Updates:** Regular updates of all software dependencies
### 4.3 Incident Response
- **24/7 Monitoring:** Automated alerts for security incidents
- **Response Team:** Dedicated security team with on-call rotation
- **Notification Procedures:** Affected users notified within 72 hours of confirmed breach
- **Post-Incident Reviews:** Comprehensive analysis and remediation after incidents
## 5. Security FAQ
### 5.1 What customer data do you collect?
Signals collects only the data needed to provide customer messaging and analytics services:
- **Customer Contact Information:** Phone numbers, emails, and messaging handles used for supported channels
- **Conversation Content:** Messages between customers, Signals, AI agents, and human operators
- **Media:** Customer-uploaded images or files used for support, returns, exchanges, and product questions
- **Order Information:** Product details, order IDs (provided by brand, not collected directly)
- **Compliance Metadata:** Consent evidence, suppression state, message route evidence, and opt-out records
- **Engagement Metadata:** Timestamps, resolution outcomes, sentiment data, and workflow outcomes
For detailed information, see our [Privacy Policy](/privacy) and [Messaging Compliance](/compliance) pages.
### 5.2 How do you secure customer messages?
Customer messages are protected through multiple security layers:
- **Encryption in Transit:** TLS 1.3 for API connections to messaging providers
- **Encryption at Rest:** AES-256 encryption for stored message content
- **Access Controls:** Only authorized AI agents and brand administrators can access conversations
- **Audit Logs:** All message access logged and monitored for unauthorized activity
### 5.3 How long do you retain media attachments?
Media attachments are handled with strict retention policies to protect customer privacy:
- **Configurable Retention:** Brands set their own retention policies based on their needs
- **Automatic Deletion:** Attachments deleted after the configured retention period where available
- **On-Demand Deletion:** Customers can request deletion through the applicable brand or by contacting Signals
- **Encrypted Storage:** Attachments encrypted at rest
### 5.4 How do you protect against common web attacks?
We implement multiple layers of protection:
- **XSS Protection:** Content Security Policy (CSP) headers and input sanitization
- **CSRF Protection:** Anti-CSRF tokens for form submissions
- **SQL Injection:** Not applicable (static site with no database)
- **DDoS Mitigation:** Google Cloud Armor and rate limiting
- **Clickjacking:** X-Frame-Options and CSP frame-ancestors headers
### 5.5 What should I do if I find a security vulnerability?
We appreciate responsible disclosure of security vulnerabilities. Please report issues to [security@returnsignals.com](mailto:security@returnsignals.com) with:
- Detailed description of the vulnerability
- Steps to reproduce the issue
- Potential impact assessment
- Your contact information for follow-up (we respect reporter anonymity if requested)
We commit to acknowledging your report within 48 hours and providing status updates throughout the resolution process.
### 5.6 How will I be notified of security incidents?
In the event of a security incident that affects your data, we will:
- Notify affected users within 72 hours of confirming the breach
- Send notifications via email to registered contact addresses
- Post a security advisory on our website
- Provide details about the incident, its impact, and remediation steps
## 6. Reporting Security Issues
We take security issues seriously and appreciate the security research community's efforts in keeping our users safe.
### 6.1 Responsible Disclosure
When reporting vulnerabilities, please:
- Email [security@returnsignals.com](mailto:security@returnsignals.com) with details
- Allow us reasonable time to address the issue before public disclosure
- Avoid accessing, modifying, or deleting data beyond what's necessary to demonstrate the vulnerability
- Do not perform actions that could harm our users or services
### 6.2 Our Commitment
- **Response Time:** Acknowledgment within 48 hours
- **Status Updates:** Regular updates on investigation and remediation
- **Recognition:** Credit for responsible disclosure (with your permission)
- **Safe Harbor:** No legal action for good-faith security research
## 7. Additional Resources
For more information about our data practices and policies:
- [Messaging Compliance](/compliance) - How we handle consent, marketing gates, and opt-outs
- [Privacy Policy](/privacy) - How we collect and use your information
- [Terms of Service](/terms) - Legal terms governing use of our services
## 8. Contact Information
For security-related questions, vulnerability reports, or security incident notifications, please contact:
**Material Model, Inc. (d/b/a Signals)**
2261 Market Street STE 85311, San Francisco, CA 94114
Security Email: [security@returnsignals.com](mailto:security@returnsignals.com)
Website: [www.returnsignals.com](https://www.returnsignals.com)
**Response Time:** Security inquiries are acknowledged within 48 hours
For general inquiries, you may also contact us at [hello@returnsignals.com](mailto:hello@returnsignals.com).
---
# Terms of Service: Signals
> Signals Terms of Service. Learn about your rights and responsibilities when using our customer messaging platform.
Last Updated: July 3, 2026
## 1. Agreement to Terms
By accessing or using Signals' services ("Service"), you agree to be bound by these Terms of Service ("Terms"). If you disagree with any part of these Terms, you may not access or use the Service.
## 2. Definitions
- **"Customer"** means the business entity that registers for, contracts for, or uses the Service.
- **"Customer Personal Data"** means data that Customer provides to Signals or that is generated through the Service on Customer's behalf, including order data, end customer contact information, consent evidence, and conversation content. Customer Personal Data does not include aggregated or de-identified data that can no longer reasonably identify a natural person.
- **"End Customer"** means a consumer or other individual with whom Customer interacts through the Service.
- **"Marketing Send"** means a brand-initiated message that promotes, encourages, or facilitates a purchase or review and is not a direct response to the End Customer's latest request.
## 3. Description of Service
Signals is a customer messaging platform for ecommerce brands. The Service supports customer conversations over iMessage, RCS, SMS, 10DLC SMS, WhatsApp, and related channels. The Service may include AI-assisted replies, human-sent replies for CX escalation, customer AI agent messages through Signals MCP, order support, delivery support, returns and exchanges, product and policy answers, customer-initiated purchase assistance, review workflows, optional brand-initiated marketing sends, analytics, and integrations with commerce, support, messaging, consent, fulfillment, warehouse, and return systems.
Signals' core service is customer support and responding to user-initiated queries. Customers may use Signals solely for support and customer-initiated conversations. Optional marketing workflows are subject to separate consent and eligibility controls.
## 4. Business-to-Business Service
Signals is a business-to-business service intended for businesses, enterprises, and professional entities. The Service is not intended for personal, household, or family use.
### 4.1 Geographic Availability
The Service is currently available only to businesses located in the United States. The Service is not offered to businesses or individuals located in the European Union, the European Economic Area, or the United Kingdom.
## 5. Customer Responsibilities
Customer is responsible for:
- Providing accurate account, billing, integration, and configuration information
- Maintaining the security of Customer accounts and credentials
- Obtaining and maintaining all rights, permissions, notices, and consents required for Customer's selected use of the Service
- Ensuring that Customer's privacy policy, terms, checkout flows, consent flows, and customer communications comply with applicable law
- Reviewing AI-assisted and automated workflows before enabling them for Customer's programs
- Providing accurate catalog, order, policy, support, return, exchange, and escalation information
- Ensuring that Customer's use of connected systems complies with those systems' terms
- Ensuring that any message Customer sends, initiates, or causes through the dashboard, API, MCP connector, or other Service interface complies with applicable law, carrier rules, platform rules, Customer's own customer-facing terms, and any required consent or suppression status
## 6. Messaging Compliance
Customer represents and warrants that it has obtained all rights, permissions, notices, and consents required for Customer's selected use of the Service, including under the Telephone Consumer Protection Act (TCPA), applicable state laws, carrier rules, messaging platform rules, WhatsApp rules, 10DLC rules, and any other applicable regulations.
Customer is not required to enable unsolicited marketing sends to use Signals. For SMS, 10DLC SMS, and RCS marketing sends, Customer must obtain explicit written consent before such sends occur. For iMessage and WhatsApp marketing sends, Customer must provide any consent required by applicable federal law, state law, platform rules, and Signals' operating policy. Where written marketing consent is not available, Signals' marketing-send eligibility controls limit brand-initiated marketing to confirmed iMessage conversations with no restricted-state signal. Signals' current public operating policy is described on the [Messaging Compliance](/compliance) page.
When Customer or its personnel, systems, or agents send, initiate, or cause messages through Signals, including through the dashboard, API, or MCP connector, Customer is responsible for the compliance of those messages. This includes the message content, purpose, timing, recipient selection, consent basis, suppression status, and any required disclosures.
Customer may not use rented, sold, purchased, scraped, or shared opt-in lists with Signals. Customer must keep connected consent and suppression systems accurate, including systems such as Shopify, Listrak, and other commerce or messaging consent tools used by Customer.
Customer shall indemnify Signals against claims, damages, penalties, losses, and expenses arising from Customer's failure to obtain required consent, Customer's messaging instructions, Customer's data, or Customer's violation of applicable messaging laws or platform rules.
## 7. Acceptable Use
Customer agrees not to use the Service to:
- Violate any applicable law, regulation, carrier rule, or platform rule
- Send unlawful, deceptive, abusive, harassing, harmful, or unwanted messages
- Upload content that infringes intellectual property rights
- Upload malicious code or interfere with the Service's operation
- Attempt to gain unauthorized access to any portion of the Service
- Upload content that violates the privacy rights of individuals
- Misrepresent consent, identity, brand affiliation, message purpose, or opt-out status
- Circumvent compliance gates, suppression controls, channel restrictions, rate limits, or security controls
## 8. Data Processing and Privacy
Use of the Service is governed by our [Privacy Policy](/privacy). By using the Service, Customer authorizes Signals to:
- Process Customer Personal Data to provide, operate, secure, support, and improve the Service
- Process conversation content, order data, product data, consent evidence, suppression records, and workflow data
- Use third-party infrastructure, AI, messaging, analytics, and integration providers as needed to provide the Service
- Store processed data for the agreed or operational retention period
- Use data as described in Section 9.3
## 9. Confidentiality, Ownership, and Data License
### 9.1 Confidentiality
Each party agrees to protect the other party's proprietary and confidential information using at least the same degree of care it uses for its own confidential information and not less than reasonable care. Each party agrees not to use or disclose confidential information except as necessary to perform under these Terms. This obligation survives for 5 years after disclosure. Standard exceptions apply, including information in the public domain, information known before disclosure, independently developed information, and disclosure required by law.
### 9.2 Ownership
Customer owns all right, title, and interest in Customer Personal Data. Signals owns the Service, platform, software, models, workflows, improvements, documentation, and related intellectual property.
### 9.3 Data License
Customer grants Signals a non-exclusive, perpetual, irrevocable, royalty-free, worldwide license to use Customer Personal Data and order data derived from the Service to improve, develop, and enhance Signals' products, offerings, technologies, and algorithms. To the extent such data is used for these purposes, Signals will use commercially reasonable efforts to aggregate or de-identify the data so it does not identify Customer or End Customers. Customer represents and warrants that it has obtained all rights and consents required to grant this license.
## 10. Account Registration
To access certain features of the Service, Customer must register for an account. Customer agrees to:
- Provide accurate, current, and complete information
- Maintain and promptly update account information
- Maintain the security of passwords and credentials
- Notify us immediately of any breach of security or unauthorized use
- Take responsibility for all activities under Customer accounts
## 11. Payment Terms and Subscription
### 11.1 Billing and Payments
If Customer subscribes to a paid plan:
- Customer agrees to pay all fees according to the selected plan, order form, or written agreement
- Customer authorizes us or our payment processor to charge the applicable payment method
- Annual plans are billed in full at the start of the subscription period unless otherwise agreed
- Monthly plans are billed on a recurring monthly basis unless otherwise agreed
- We may change pricing with 30 days notice for monthly plans or at renewal for annual plans
### 11.2 Cancellations and Refunds
- Customer may cancel a subscription according to the applicable order form or billing workflow
- Cancellations do not result in refunds for the remaining time in the current billing period unless the applicable order form says otherwise
- Pilot refund terms stated in an order form or written pilot agreement control for the pilot fees covered by that agreement
- Fees are otherwise non-refundable except as required by applicable law
### 11.3 Plan Changes
- Upgrades take effect immediately unless otherwise agreed
- Downgrades take effect at the end of the current billing period unless otherwise agreed
- Customer may request plan changes through the agreed account or billing process
## 12. Data Security and Retention
Signals implements security measures including TLS encryption, encryption at rest, private cloud infrastructure, role-based access controls, audit logging, monitoring, and backups.
Customer may request deletion of Customer Personal Data by contacting [privacy@returnsignals.com](mailto:privacy@returnsignals.com). If End Customers request deletion of their personal data, Customer may submit those requests on their behalf. Deletion requests do not affect data already aggregated or de-identified under Section 9.3, or records retained for legal, security, billing, fraud prevention, compliance, or dispute-resolution purposes.
## 13. Third-Party Services
The Service may connect to or rely on third-party services, including commerce platforms, support tools, messaging providers, consent systems, AI providers, analytics tools, fulfillment systems, warehouse systems, and payment processors. Customer is responsible for maintaining its accounts and permissions with those services and complying with their terms.
## 14. Disclaimers and Limitations of Liability
### 14.1 Service "As Is"
THE SERVICE IS PROVIDED "AS IS" WITHOUT WARRANTIES OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, NON-INFRINGEMENT, OR ERROR-FREE OPERATION.
### 14.2 No Legal Advice
Signals may provide compliance controls, documentation, and operating guidance, but Signals does not provide legal advice. Customer is responsible for its own legal review and compliance decisions.
### 14.3 Limitation of Liability
IN NO EVENT SHALL SIGNALS BE LIABLE FOR ANY INDIRECT, INCIDENTAL, SPECIAL, CONSEQUENTIAL, EXEMPLARY, OR PUNITIVE DAMAGES, OR ANY LOSS OF PROFITS, REVENUES, GOODWILL, DATA, OR BUSINESS OPPORTUNITY, WHETHER INCURRED DIRECTLY OR INDIRECTLY.
### 14.4 Maximum Liability
OUR TOTAL LIABILITY SHALL NOT EXCEED THE GREATER OF $100 OR THE AMOUNTS PAID BY CUSTOMER TO SIGNALS IN THE 12 MONTHS PRECEDING THE CLAIM.
## 15. Indemnification
Customer agrees to indemnify and hold Signals harmless from any claims, losses, damages, liabilities, penalties, costs, and expenses arising from Customer's use of the Service, Customer data, messaging compliance failures, violation of these Terms, or infringement of third-party rights.
## 16. Termination
We may terminate or suspend access immediately, without prior notice, for:
- Breach of these Terms
- Non-payment of fees
- Legal, regulatory, carrier, platform, or security risk
- Request by law enforcement or government agencies
- Unexpected technical or security issues
- Extended periods of inactivity
Customer may terminate its account according to the applicable order form or written agreement. Upon termination, Customer's right to use the Service will cease.
## 17. Governing Law
These Terms are governed by the laws of the State of California, without regard to conflict of law principles. Any disputes shall be resolved in the state or federal courts located in San Francisco County, California.
## 18. Changes to Terms
We may modify these Terms at any time. We will provide notice of material changes at least 30 days before they take effect where required. Continued use of the Service after changes become effective constitutes acceptance.
## 19. Severability
If any provision of these Terms is held unenforceable, the remaining provisions will remain in full force and effect.
## 20. Entire Agreement
These Terms, our [Privacy Policy](/privacy), and any order form or other written agreement between Customer and Signals constitute the entire agreement between Customer and Signals regarding the Service. Our [Messaging Compliance](/compliance) page describes our current public operating policy, but Customer's legal obligations are not limited to that summary. If an order form or enterprise agreement conflicts with these Terms, the order form or enterprise agreement controls for that Customer.
## 21. Contact Information
For questions about these Terms, contact us at:
**Material Model, Inc. (d/b/a Signals)**
2261 Market Street STE 85311, San Francisco, CA 94114
Email: [hello@returnsignals.com](mailto:hello@returnsignals.com)
Website: [www.returnsignals.com](https://www.returnsignals.com)
---