Free 1-month trial · No credit card · Live in 30 minutes
Marketing
One win-back campaign did nearly all the work. The honest reading of the other twenty-one is the more useful one.
7 min read · Updated September 2026
Two figures describe the same 8,359 win-back messages, and the distance between them is the finding. Across every campaign, 4.8% of messages were followed by a booking; drop the one campaign that carried most of those bookings and the rate is 2.0%.
Neither number is the answer to "do win-back texts work". The first one flatters us and the second one is closer to what a salon should expect from a campaign chosen at random. What sits between them is the actual lesson, which is that reactivation is not a channel you switch on. It is a small number of offers that land and a larger number that do not, and the only way to tell them apart is to look at them one at a time.
Own the limits before the findings. These are campaigns run through our own platform for two accounts, dominated by one multi-location studio in California, covering 2026-05-26 to 2026-09-15. Two accounts is not an industry. And the attribution is deliberately weak: this counts a client who booked after receiving a message, not a client who booked because of one. Someone who was going to rebook anyway, and happened to get a text that week, is in the numerator. Read every figure below as an upper bound.
Share of campaign messages followed by a booking
| Group | Share | Sample |
|---|---|---|
| campaign 27 | 3.7% | 945 |
| campaign 34 | 0.1% | 868 |
| campaign 32 | 0.5% | 843 |
| campaign 33 | 0.1% | 776 |
| campaign 5 | 35.9% | 597 |
| campaign 23 | 0.0% | 497 |
| campaign 22 | 0.0% | 492 |
| campaign 24 | 0.0% | 492 |
| campaign 1 | 9.0% | 410 |
| campaign 12 | 2.6% | 390 |
| campaign 8 | 2.9% | 344 |
| campaign 3 | 0.4% | 240 |
| campaign 19 | 0.0% | 193 |
| campaign 21 | 0.0% | 193 |
| campaign 42 | 0.0% | 110 |
| campaign 38 | 0.0% | 101 |
| campaign 15 | 2.2% | 91 |
| campaign 4 | 34.9% | 86 |
| campaign 37 | 0.0% | 71 |
| campaign 7 | 1.4% | 70 |
| campaign 10 | 10.9% | 64 |
| campaign 11 | 8.8% | 57 |
AdminifAI booking data · n = 8,359 · 2 salons · 2026-05-26 to 2026-09-15 · a multi-location studio in California
The shape is the point. Most of these campaigns converted nobody at all, or close enough to nobody that the difference is noise. A handful did modestly. One — campaign 5 in the table — ran away with it, on a send list smaller than most of the others.
This is why an average is the wrong summary statistic for campaign work. A salon reading "around five in a hundred" would plan a season of sends around a number that only one campaign in this set ever achieved. The median campaign here converted almost nobody.
The data says one offer worked. It does not say why, and we are not going to reverse-engineer a reason from a single campaign and present it as a principle.
That restraint is not modesty for its own sake. A win-back campaign bundles at least six decisions — who is on the list, how long since their last visit, what is offered, whether anything is discounted, the wording, and the hour it sends. Campaign 5 made all six at once, and it is one campaign. Any story about which decision mattered would be a story, and it would spread as a rule of thumb long after anyone remembered it came from one send.
What the distribution does support is narrower and more useful:
Plain texts converted at 7.9% across the same set; picture messages, in the table below, converted far lower.
Conversion rate, plain text versus picture message
| Group | Share | Sample |
|---|---|---|
| SMS | 7.9% | 4,423 |
| MMS | 1.3% | 3,936 |
AdminifAI booking data · n = 8,359 · 2 salons · a multi-location studio in California
That looks like a clean result and it is not one. The split is not controlled for campaign, and the campaign that dominated everything else sits in the plain-text arm. So this table mostly restates the previous section: one campaign carried the numbers, and it happened to be a text. It is not evidence that pictures hurt, and if you have been sending image promotions you should not stop on the strength of it.
We are including it because the confounded version is the version we have, and leaving it out would imply we had tested something we have not. If a salon wants to know whether images help, the test is the same campaign sent both ways to a split list.
Reactivation has a price that does not appear on an invoice. Across these campaigns the opt-out rate was 1.7%.
Low, but it is worth being precise about what it means, because an opt-out is permanent in a way a bad campaign is not. A campaign that converts nobody has cost you a morning. A campaign that converts nobody and collects opt-outs has cost you the ability to contact those people again — including the appointment reminders that actually work, which is the quiet way a marketing decision turns into a no-show problem. Reminders are a different job from promotions, and we have written about that boundary.
So the arithmetic on a speculative campaign is not "what if it does not convert". It is "what if it does not convert and costs me part of my list". That reframing kills most of the ideas that come out of a quiet week.
Most win-back lists are built on a guess about when a client has lapsed, and the guess is usually too generous.
Across 2,296 repeat visits at six salons, covering 2025-01-01 to 2026-09-14, 85.1% of them happened within sixty days of the visit before. That figure counts repeat visits only — clients who came once and never returned are not in it — so it describes the rhythm of people who do come back rather than how many do.
For list-building the reading is still practical. A client ninety days out is not slightly late; they are outside the pattern almost every returning client follows. Waiting until six months have passed to send anything means the first contact arrives long after the habit broke. We set out where that line sits, and what it costs to cross it, in the 60-day cliff.
The templates we would start from are on rebooking texts, and the mechanics of running the sends are on bookings. If the reason your clients are lapsing is that nobody picks up when they try to rebook, that is a different and usually larger problem — the call logs are in when salon clients actually call, and how to get more salon clients covers the acquisition side.
Some do and most do not. Across 8,359 messages sent for two salon accounts, 4.8% of messages were followed by a booking, but removing the single most effective campaign drops that to 2.0%. The realistic expectation for a new campaign is the lower figure, with a small chance of the higher one. Both numbers count bookings made after a message rather than because of it.
Sooner than most salons wait. Across 2,296 repeat visits at six salons, 85.1% happened within sixty days of the previous visit, so a client past ninety days is already outside the pattern that returning clients follow. A first contact at six months is a reintroduction, not a reminder.
Across these campaigns the opt-out rate was 1.7%. Treat it as a real cost rather than a rounding error: an opt-out also removes that client from appointment reminders and confirmations, which are the messages that reliably earn their keep.
Our own data cannot answer it. Plain texts converted higher in this sample, but the split is confounded — the one campaign that outperformed everything sits in the plain-text group, so the comparison mostly measures that campaign. The only way to know is to send the same campaign both ways to a randomly split list.
Nothing in this data settles it, and any campaign that worked made several decisions at once. Before discounting, note that a discount is the one variable that lowers the value of the bookings you win, so it deserves a test of its own rather than being included by default.
Software & pricing
7 min read
Missed calls
6 min read
Retention
8 min read
Free for a month, on whatever booking system you already run. Set up in under 30 minutes, with your services and staff loaded for you.
Start your free monthNo credit card · No contract · hello@adminif.ai
Free for one month
Tell us where to reach you and we will have your assistant answering within 30 minutes. No card needed.