Short answer: Adding a “reply STOP to opt out” line to an outbound message does not reduce how many people reply — but in a live split test it multiplied the opt-out rate by roughly 5.8× and cut the booking rate to about a third. The opt-out line doesn’t suppress engagement; it redirects it, handing hesitant prospects an explicit exit before they can be warmed into a conversation. Where opt-out language is legally required, keep it — compliance is not optional — but don’t bolt it onto messages that don’t need it, and treat opt-out rate as a signal, not a scoreboard.
Every outbound team eventually runs into the same tension. Compliance frameworks — the TCPA in the United States, the Spam Act 2003 in Australia, and most platform policies — push you toward clear, explicit opt-out language on marketing messages. Good practice, and often the law. But sales instinct says the moment you tell someone how to leave, more of them will. So which is it? We ran the test on live client campaigns and measured it.
The test
Across the outbound programs we run for clients, we took a single high-intent sequence and split the very first message into two variants, classified by whether the sent body contained an explicit “reply STOP to opt out” line:
- Variant A — no STOP line. The message opens the conversation and invites a reply, with no opt-out instruction in the body.
- Variant B — identical message plus an explicit “reply STOP to opt out” line.
Everything else — audience, offer, timing, sender — was held constant, so the only difference a prospect saw was the opt-out sentence. Crucially, the conversations themselves are handled by a standardised AI setter, not a rotating cast of human reps: there is no variation in skill, mood, time of day, or script drift to muddy the result. The only thing that moved was the opt-out line, which is what makes the comparison clean. Here is the current read (opt-out figures shown as rates, since the raw opt-out volumes are small):
| Variant | Sample (first-message sends) | Reply rate | Opt-out rate | Booking rate |
|---|---|---|---|---|
| A — no STOP line | 54 | 50.0% | 3.7% | 20.4% |
| B — with “reply STOP” | 42 | 59.5% | 21.4% | 7.1% |
What the numbers actually say
Three things stand out, and the third is the one most teams get wrong.
- Reply rate is a wash. The STOP variant actually replied slightly more (59.5% vs 50.0%). Whatever the opt-out line is doing, it is not stopping people from engaging.
- Opt-outs jumped about 5.8×. The variant with the STOP line opted out at 21.4% versus 3.7% without it. That is not noise — it is the single largest movement in the test.
- Bookings fell to roughly a third. 20.4% of the no-STOP sends turned into a booked call, against 7.1% with the STOP line.
Put those together and a clear mechanism emerges. The opt-out line does not depress engagement — it redirects it. Hesitant people who would otherwise reply, get a question answered, and be warmed into booking a call instead take the explicit exit ramp you just handed them. Same total engagement, different destination: the exit instead of the calendar.
Is it statistically real?
Honesty matters more than a clean headline, so here is the statistical footing. On a Fisher’s exact test, the opt-out gap is significant at p = 0.009 — the increase in opt-outs is a real effect, not a small-sample fluke. The booking gap is suggestive but not yet conclusive at p = 0.085; call it a strong early signal on roughly 40 sends per arm rather than a settled fact.
The larger picture backs it up. Across the full history of this test (every send since the first opt-out variant went out), the no-STOP arm ran at a 9.6% opt-out rate against 19.2% for the STOP arm — the opt-out penalty holds at scale and stays statistically significant (p = 0.043). Bookings across that same span ran 15.1% without the line versus 8.5% with it. The direction never reverses.
The compliance backdrop: SMS and AI calls, US and Australia
The opt-out question matters so much right now because the ground under outbound is shifting quickly in both of our main markets. Two things are converging: regulators are tightening the rules on marketing messages, and they have explicitly pulled AI-generated voices into the same net as old-fashioned robocalls. Here is the broad shape of it.
United States
- The TCPA is the big one. The federal Telephone Consumer Protection Act carries statutory damages of roughly US$500–$1,500 per message or call, with no cap — which is why TCPA class-action filings climbed sharply through 2025. Marketing texts and calls generally require prior express written consent.
- AI voices are now covered. In February 2024 the FCC ruled that an AI-generated voice counts as an “artificial or prerecorded voice” under the TCPA. In plain terms, using an AI voice agent for outbound doesn’t sidestep the consent rules — and an existing business relationship doesn’t exempt it.
- Opt-outs got broader. Since April 2025, consumers can revoke consent by “any reasonable means” — not just the word STOP, but “quit”, “cancel”, “unsubscribe” or plain language — and businesses generally must honour it within about 10 business days.
- SMS has a second gate. On top of the TCPA, US carriers require A2P 10DLC registration for business texting and, since early 2025, block unregistered traffic outright. Add roughly 15 states with their own “mini-TCPA” rules and the text channel is now regulated infrastructure, not a free-for-all.
Australia
- The Spam Act 2003 governs commercial electronic messages such as email and SMS. Broadly, each message needs consent (the ACMA prefers clear, express consent), must identify the sender, and must carry a working unsubscribe that doesn’t force the recipient to log in or hand over extra details — actioned within five business days. Penalties run into the millions for serious or repeat breaches.
- AI calls are treated like any other call. The ACMA’s position is that an AI voice call is a telemarketing call: scrub against the Do Not Call Register, identify yourself, disclose that it’s an AI up front, honour opt-outs, and respect calling hours. In April 2025 one telemarketer was fined A$1.5 million — and its director personally fined — over calls to registered numbers, so the enforcement is real.
Australia’s new SMS Sender ID Register — and why it may not affect you
This is the ACMA change everyone is searching for, so here is the plain version. Australia’s SMS Sender ID Register opened on 30 November 2025 and becomes enforceable from 1 July 2026. From that date, only registered alphanumeric sender IDs — the brand-name “from” labels shown in place of a phone number, like a company name — may be used to text Australian mobiles. Unregistered brand-name messages get pushed into a single “Unverified” thread on the recipient’s phone and flagged as a likely scam. It is an anti-impersonation measure aimed at scammers spoofing trusted brands.
Here is the part that matters for most outbound: it only applies to brand-name (alphanumeric) sender IDs — not to messages sent from an ordinary long mobile number. If you text from a standard mobile number, the Sender ID Register does not gate you.
And that is the direction we would point you regardless. We don’t recommend brand-name sending, because in our experience an ordinary long mobile number simply converts better: it reads as a real person opening a two-way conversation, not a one-way broadcast from a brand. For the kind of conversational outbound that actually books meetings, a personal-looking mobile number is an advantage — and it happens to sit outside the new register entirely.
Notice the pattern: almost every rule adds a step between you and the conversation — a consent gate before the first message, an AI disclosure at the top of a call, an opt-out line in the body, a broadening of what even counts as opting out. Which is exactly where the uncomfortable part comes in.
The part nobody says out loud: compliance has a conversion cost
Here is the honest version, and it’s the whole reason we ran the test at the top of this article. Every compliance layer you add is, by design, an easier way for a prospect to leave — and prospects use it. Our split test put a number on one of those layers: an opt-out line multiplied opt-outs about 5.8× and cut bookings to roughly a third. An AI-disclosure line at the start of a call does something similar at the other end, handing the person a reason to hang up before you’ve said anything worth staying for. A consent gate shrinks the top of your funnel before the campaign even begins.
None of that is an argument to cut corners. The penalties are severe, the reputational damage is worse, and — bluntly — messaging people who never asked to hear from you is bad business regardless of the law. The point is the opposite: go in with your eyes open. Assume every required disclosure costs you something, budget for it, and win it back where you legally can — sharper targeting, a stronger first line, faster follow-up, and a consistent AI setter that works every conversation the same way. The teams that win at compliant outbound aren’t the ones who found a loophole; they’re the ones who accepted the tax and got so much better at everything else that it stopped mattering.
So should you drop the opt-out line?
No — and this is important. Where an opt-out mechanism is legally required, you keep it. The TCPA, Australia’s Spam Act, and the major messaging platforms exist for good reasons, and the conversion cost of a STOP line is never a licence to break the law or spam people who never asked to hear from you. What the data changes is how you think about it:
- Don’t add it where it isn’t required. Plenty of teams paste an opt-out line onto every channel and message reflexively. If a given touch doesn’t legally need it, that reflex is quietly costing you two-thirds of your bookings.
- Where it is required, the message has to earn the booking faster. A required STOP line means you are handing over the exit up front, so everything else — targeting, relevance, timing, the strength of the first line — has to do more work. The disclaimer is fixed; the message is your lever.
- Stop treating opt-out rate as pure loss. An opt-out cleans your list and removes someone who was never going to convert. Silence is not consent, and a low opt-out rate on a cold list usually means people are ignoring you, not warming to you. Opt-out rate alone is a poor quality metric — read it next to reply and booking rates, never on its own.
The deeper lesson is the one that runs through everything we do: outbound results are decided by who you contact and what you say far more than by any single disclaimer. If you want to see where the real leverage sits, our breakdown of speed to lead and the 5-minute rule and our look at pay-per-lead versus pay-per-appointment are the natural next reads.
Frequently asked questions
Does “reply STOP to opt out” reduce reply rates?
No. In our split test the variant with the STOP line replied slightly more often (59.5% vs 50.0%). The opt-out line does not suppress engagement — it redirects some of that engagement toward opting out instead of booking.
How much does an opt-out line affect bookings?
In our test the booking rate fell from 20.4% without the line to 7.1% with it — roughly a third — while the opt-out rate rose about 5.8×. The opt-out effect is statistically significant (p = 0.009); the booking effect is a strong early signal (p = 0.085) on a small sample.
Should we remove opt-out language to book more calls?
Only where it is not legally required. Opt-out mechanisms mandated by the TCPA, the Spam Act 2003 or your messaging platform must stay. The practical takeaway is to avoid adding opt-out language reflexively where it isn’t needed, and to invest harder in message quality and targeting where it is.
Is a low opt-out rate a good sign?
Not on its own. A low opt-out rate on a cold list often just means people are ignoring you. Silence is not consent. Read opt-out rate alongside reply and booking rates — a higher-intent audience can opt out more precisely because it engages more.
Do AI voice calls need consent under the TCPA?
Generally, yes. In February 2024 the FCC ruled that AI-generated voices count as “artificial or prerecorded” voices under the TCPA, so AI outbound calls face the same consent requirements as other robocalls — prior express written consent for marketing. This is general information, not legal advice; confirm your setup with a qualified adviser.
What are the SMS opt-out rules in the US and Australia?
Broadly: in the US, since April 2025 businesses must honour opt-outs made by any reasonable method (not only “STOP”) within about 10 business days; in Australia, the Spam Act 2003 requires a working unsubscribe on every commercial message, actioned within five business days. Rules vary and change — treat this as general information, not legal advice.
Does the ACMA SMS Sender ID Register apply to my business?
Only if you send SMS using a brand-name (alphanumeric) sender ID. Australia’s SMS Sender ID Register opened on 30 November 2025 and is enforceable from 1 July 2026; from then, only registered alphanumeric sender IDs can be used, and unregistered ones are flagged to recipients as unverified. It does not apply to messages sent from a standard long mobile number — which is what we recommend anyway, because long mobiles tend to convert better. General information, not legal advice.
Want outbound that books calls without burning your list?
Disclaimer: This article is general information, not legal advice. Marketing and telecommunications rules change often, differ by state and territory, and depend on the specifics of your setup. The compliance points above are orientation only — confirm your own program with a qualified lawyer or telco-compliance specialist before relying on it.