All articles

Opt-outs done right: STOP keywords, intent detection and why it matters

People don't always reply "STOP." Here's why honoring every real opt-out request protects your delivery, how to avoid removing people who still want to hear from you, and a checklist for your next program.

Every texting program gets opt-outs. Handled well, they're a small, healthy part of a clean list. Handled badly, they turn into spam complaints, carrier filtering and delivery problems for every message you send after. The difference usually comes down to one thing: whether your platform hears "stop" only when someone types the exact word.

Why opt-outs matter more than they look

When someone asks you to stop texting and the messages keep coming, they don't just leave quietly. They report the message as spam, and those reports feed directly into how carriers judge your traffic. Enough of them and carriers start filtering your messages, including to the people who wanted to hear from you.

That's why the industry treats opt-outs as a compliance requirement, not a courtesy. Carrier and CTIA guidelines expect every program to tell recipients how to opt out and to honor those requests promptly. And the FCC's updated consent rules, which took effect in 2025, expect opt-out requests made by any reasonable means to be honored, not only the exact keyword "STOP."

The simple rule: if a reasonable person would read a reply as "please stop texting me," your program should treat it that way.

The basics every program should get right

1. Say how to opt out, once, up front

Include clear opt-out language, like "Reply STOP to opt out," in the first message of a conversation. You don't need to repeat it on every follow-up reply; that clutters the conversation and can make a personal exchange feel automated.

2. Honor the standard keywords instantly

STOP is the one everyone knows, but people also reply STOP ALL, UNSUBSCRIBE, CANCEL, END and QUIT. All of them should opt the person out automatically, with a short confirmation and a way back in (for example, "Reply START to re-subscribe").

3. Make it stick across programs

An opt-out isn't just for the program that received it. If the same organization texts that person next month from a different program, the request should still apply. Opt-outs belong on a list that every future program checks before it sends.

4. Don't make volunteers guess

In peer-to-peer programs, volunteers and agents read replies all day. They need a one-click way to mark a reply as an opt-out, so a request buried in a longer message doesn't slip through.

The hard part: people don't talk in keywords

Real replies look like this:

  • "Please don't text me again."
  • "Take me off your list."
  • "Wrong person, stop messaging this number."
  • "I've already voted, leave me alone."

None of those contain the word STOP on its own, and every one is a clear request to stop. A keyword-only system misses all of them. On a program sending to hundreds of thousands of people, that can be thousands of ignored requests, and thousands of chances for a spam report.

The opposite mistake matters too. Plenty of replies use the word "stop" or sound negative without being opt-outs: "Can you stop by the office Saturday?", "I can't stop laughing," or a heated opinion about the other candidate. Treating every one of those as an opt-out removes people who still want to hear from you, and quietly shrinks your list.

How echo19 handles it

We built opt-out handling around both problems: catch every real request, and don't overreach.

  • Keywords first. STOP, STOP ALL, UNSUBSCRIBE, CANCEL, END and QUIT are honored automatically, with a confirmation and a START option to re-subscribe.
  • AI for everything else. If a reply doesn't match a keyword, a survey answer or an automated response, AI reads it for opt-out intent. It only acts when it's highly confident (90% or higher), and anything ambiguous is treated as not an opt-out. Every decision is logged, so it can be reviewed.
  • Common sense on angry replies. Profanity about a candidate or an issue isn't an opt-out. Abuse aimed at the sender, like "leave me alone," is, because that person clearly doesn't want more messages.
  • One click for P2P agents. Volunteers and agents can tag any reply as an opt-out from the conversation view.
  • Lists that carry forward. Opt-outs go onto your organization's opt-out list, so future programs skip those numbers automatically.

A quick checklist for your next program

  1. Does your first message include clear opt-out language? "Reply STOP to opt out" is enough.
  2. Are all the standard keywords honored automatically, not just STOP?
  3. What happens to "please stop texting me"? If your vendor can't answer, those replies are probably being ignored.
  4. Do opt-outs carry over to your organization's future programs?

Clean opt-out handling protects the people you're texting, your delivery rates and your reputation with the carriers. It's one of the least glamorous parts of a texting program, and one of the most important.

This article is general information, not legal advice. Talk to your counsel about the consent and opt-out rules that apply to your organization's programs.

Want a second look at your opt-out setup? We're happy to walk through how echo19 handles opt-outs, and what it would look like for your program. Talk to our team.

Plan your next cycle

Keep your lists clean and your delivery strong.

Tell us about your program and we'll show you how echo19 handles opt-outs, from STOP keywords to plain-English requests.

  • Real people who know political texting
  • Straight answers on pricing and 10DLC
  • No contracts, no subscriptions, no setup fees

Talk to our team

Tell us about your program. Someone from our team will get back to you shortly.

See our Privacy Policy for how we handle your information.

Book a demo Contact us