Win-back messages that actually bring people back

A regular stops coming in. Nobody decides to leave; they just quietly stop, and three weeks later you notice the gap. The instinct is to send something, fast, usually a discount, usually to everyone who has gone quiet at once.

That instinct is why most win-back messages do not work. The fix is not a bigger discount. It is a message built around one genuine reason to return, sent at the moment it is still true, on the channel that suits how much this customer is worth.

Key takeaways

A blanket “we miss you” underperforms even at scale. Omnisend’s own 2026 benchmark puts customer-reactivation email at a 0.54% conversion rate, the lowest of any automation type it tracks.

Set the threshold from the customer’s own visiting rhythm, not a round number of days. What counts as “gone quiet” for a daily café customer is normal for a salon.

Lead with a specific fact, not a discount. The offer is a second message, only for whoever the first one did not bring back.

Save SMS for the customers worth its cost: it gets acted on the most and forgives the least.

Stop after two messages. A third win-back to someone who has not responded is not persistence, it is noise.

Why the generic version does so badly

Win-back is the hardest message a small shop sends, and the data on professionally run programs shows it. In Omnisend’s 2026 email marketing benchmark, drawn from over 20 billion campaign emails across more than 27,000 brands, customer-reactivation email converts at 0.54%, with a 33.11% open rate and the lowest revenue per send of any automation category it measures (Omnisend, published May 2026, 2025 data).

That is ecommerce data, not a café or a salon, and reactivation email is not the same job as a win-back text to forty people. But the pattern is worth taking seriously: even brands running this at scale, with real segmentation tools, get less than one reply in a hundred out of a generic “we miss you.”

The reason is not the channel. It is that “we miss you” carries no information. It does not tell the customer anything they did not already know, and it usually arrives with a discount that reads as an apology for bothering them. Loyalty segmentation already covers the group this message is for: the customers who have gone past their own usual interval, wanting one genuine reason to return, once, at the right moment. This post is that message.

Set the threshold from their rhythm, not the calendar

Whether a customer counts as “lapsed” is not a fixed number. It depends on how often they came in the first place.

The foundational marketing-science work on this, still the basis of most customer-value models used today, treats “has this customer gone quiet” as a question you can only answer against that customer’s own buying rhythm, not against a shared calendar date (Schmittlein, Morrison and Colombo, “Counting Your Customers: Who Are They and What Will They Do Next?”, Management Science 33(1), 1987, doi.org/10.1287/mnsc.33.1.1). A customer who visits daily and disappears for two weeks has gone further off pattern than one who visits every six weeks and is two weeks “late.”

Automated loyalty campaigns has the threshold table built from that idea: nudge around one to two visit-intervals past normal, make the real offer only if that first message does not work. The mechanics of building it are covered there. What that page does not have room for is what the two messages should actually say, which is the rest of this one.

Passumo’s own win-back trigger fires by default after 14 days without a visit. That default is a starting point, not a rule: a daily coffee shop should tighten it, a salon booked six weeks apart should loosen it considerably, and the delay is yours to change on the automation.

The first message: a fact, not a discount

The nudge is the one most shops skip, because it feels like it is not doing anything. It has no offer attached, which is the point. Most lapsed customers have not defected; they have just not thought about you, and a discount solves a problem they do not have.

Say one specific, true thing. Good candidates:

Progress already made. “Your card’s still got six stamps on it” costs the customer nothing to act on and reminds them of something they have already invested in.
Something genuinely new. A seasonal menu, later opening hours, a different Sunday roast. New information earns a look in a way “we miss you” cannot.
A specific fact about them. Not a first name dropped into a template. What they are actually close to, or actually missing.

A café, sent by push, three weeks after the last visit:

Your loyalty card's still got six stamps on it.
Two more for a free coffee.

A barbershop, sent by SMS, seven weeks after a six-week regular’s last cut:

Haven't seen you in a while. Your usual slot's
free this Saturday if you fancy it.

Neither offers anything. Both give the customer a reason to think about the shop again, which is the only job of message one. Writing a message people do not delete covers the craft of the sentence itself: lead with the shop, say the true thing, then stop.

The second message: where the offer earns its place

If the nudge does not bring them back, silence has cost you more than a discount will, and this is where an offer belongs. Not before.

A pub, sent by email, four weeks after the nudge:

Subject: A pint on us this week

It's been a while since we've seen you in. Come down
this week and your first drink is on the house, no
catch. We'd just like to see you back.

A salon, sent by SMS, to a client who booked every six to eight weeks and is now three months quiet:

We haven't seen you in a few months. Book this month
and 20% comes off your usual appointment.

Two things make both of these different from a blanket blast. Each names the gap honestly rather than pretending nothing happened, and each is sent to one person past their own pattern, not to a list segmented by nothing more than “has not visited recently.” The offer is the exception in this sequence, made once it is actually earned, which is also why it can afford to be a real one: you are not discounting your way to a sale that was coming anyway. What a stamp card is actually for covers why running a reward from margin, rather than as a habitual discount, is the sustainable version of this.

Choosing the channel

SMS is for the message that matters most and cannot be missed, and a customer worth winning back deserves exactly that channel. Push, email or SMS has the full comparison, including why SMS gets acted on more per message than the other two and costs more per message to match.

That trade-off is the reason to be selective rather than to text everyone who has gone quiet:

Customer’s value to you	Channel for the offer stage
High-frequency or high-spend regular	SMS. The cost is worth it and the message needs to be seen.
Occasional customer, moderate spend	Email. Survives being ignored, costs nothing to send.
New or low-engagement member	Push, if the wallet pass is still on their phone. Skip a paid channel until they have shown more.

The nudge stage can run on push or email for almost everyone, since it costs nothing and asks nothing. Save the paid channel for the customers where a missed message is a real loss.

What actually kills a win-back message

Leading with the discount. “20% off, come back!” tells the customer the relationship is transactional, which trains them to wait for the next discount rather than to come back because they wanted to.

Guilt. “We miss you so much!” reads as a shop talking about itself at the one moment it should be talking about the customer.

Sending to everyone who has gone quiet, on one schedule. A person who visits weekly and a person who visits quarterly are not the same kind of absent. One threshold for both means the message arrives too early for one group and far too late for the other.

A third message. Two is the sequence: a nudge, then an offer for whoever the nudge did not reach. Anyone still absent after both has told you something, and a third attempt reads as pestering rather than persistence.

Measuring whether it worked

Ignore the open rate. The only number that means anything is whether the customer actually returns within a couple of weeks of the message, which is the same standard automated loyalty campaigns sets for judging any win-back automation. What a healthy return rate looks like, and what it means when it is low, is covered in loyalty program metrics.

How Passumo runs this

The “when they stop coming” trigger on the automations page fires by default after 14 days without a visit, and you can change that number per card. Building the two-stage sequence above means setting up two automations on that same trigger with different day thresholds: an earlier one with no reward action attached for the nudge, a later one that adds a bonus stamp or applies a discount for the real offer. Each can send on a different channel, push or email for the nudge, SMS for customers where the offer is worth the per-message cost.

Frequently asked questions

How long should I wait before writing off a lapsed customer as gone for good? There is no fixed point where someone is definitely gone. Treat the second, unanswered message as the practical cutoff: keep sending regular updates by whichever free channel they are still on, but stop spending a paid message or a special offer on them until something changes.

Should the win-back message always include a discount? No. The nudge should not. The second message can, once the nudge has had its chance, and the offer works better for having been earned rather than default.

What if a lapsed customer never gave an email or phone number? Push still works. A wallet pass needs no address or number to reach, which is one reason to collect a name-only sign-up rather than none at all. What to put next to the till covers what a join form should actually ask for.

Is one win-back sequence enough, or do I need different ones per trade? The two-stage shape works for any local business. What changes by trade is the interval: a café’s “gone quiet” is measured in days, a salon’s in months. Set the numbers from your own customers rather than borrowing someone else’s.

Can I run this without checking the customer list by hand? Yes. Once the thresholds are set on the automation, it runs on every customer who crosses them, which is the entire point: nobody is checking a spreadsheet for who has gone quiet this week.

Where to go next

The threshold and the automation mechanics are in automated loyalty campaigns. Which channel a given message belongs on is push, email or SMS. The sentence-level craft, for this message and every other one, is in writing a message people do not delete. And knowing who has actually gone quiet, rather than guessing, starts with loyalty segmentation.

Sources: Omnisend, email marketing benchmarks (published May 2026, 20+ billion campaign emails across 27,000+ brands, 2025 data; skews toward ecommerce brands rather than local businesses). Schmittlein, D.C., Morrison, D.G. and Colombo, R., “Counting Your Customers: Who Are They and What Will They Do Next?”, Management Science 33(1), 1987, doi.org/10.1287/mnsc.33.1.1.