In an appointment business the client does not leave in a decision you can see. They pass their own return window, nothing happens, and by the time anyone notices they have been somewhere else twice. This system holds the interval and acts while the relationship is still recoverable.
Retention failures do not announce themselves. They accumulate as an absence, and absence is not a data point in any standard report.
Every service has a natural return window and every client has their own version of it. Nothing in a standard booking system holds that number, so nothing can notice when it passes. Overdue clients are found by accident, usually by a person scrolling through a list on a slow afternoon.
Monthly revenue can stay flat while the returning share of it erodes and new acquisition covers the difference. By the time the trend is visible in the totals, a year of relationships has already gone, and reacquiring those people costs several times what holding them would have.
The usual fix is a blanket campaign: everyone not seen in ninety days gets the same message. It reaches people who were never due, misses people who were badly overdue, and teaches the whole list to ignore you. The failure is attributed to win-back as an idea rather than to the rule.
A share of lapsed clients return on their own. Without a comparison group every one of those returns is claimed by the campaign, the number looks good, and nobody learns anything that survives contact with a serious question.
Four steps, in order. The first two are where the accuracy comes from and the last is where the credibility does.
Learns the expected return window from that person's own history and from the service they use, rather than applying one threshold across a mixed base. A quarterly client and a monthly client are both regulars and need different arithmetic.
Flags the gap as it opens rather than at an arbitrary cutoff, and separates the client who is drifting from the one who is travelling, seasonal, pregnant, injured or has moved away, because those are different conversations and two of them should not happen at all.
Contacts the client on the channel they actually respond to, within frequency caps and quiet hours, with content written to your market rules. Stops after a defined number of unanswered attempts rather than continuing until the person opts out in irritation.
Keeps a share of lapsed clients uncontacted and compares the two groups over the same period. This is the only construction that separates your work from clients who were coming back anyway.
Runs on the same integration as intake, conversation analysis and planning, which is why the second system is far cheaper to add than the first.
Who you do not contact determines whether this reads as attentive or as harassment.
Opt-outs are honoured across every system and every channel, not just the one that recorded them. A client who unsubscribed from email and then receives a message has been failed by architecture, not by policy.
Complaints, poor outcomes and disputed charges are excluded rather than swept into a win-back list. Where that information is not currently recorded, building the exclusion list is part of the work.
A client mid-course, on a plan, or with a future appointment already booked is not overdue, and contacting them as though they were undermines every later message.
Relocation, life changes and services that were never going to repeat. Reaching them costs goodwill and returns nothing, and treating the list as a volume exercise guarantees that it happens.
Retention is the easiest place in this business to produce a flattering number, which is exactly why the measurement design comes before the campaign.
Return rate by cohort, average interval by service, share of clients with a second visit, and revenue from returning clients, recorded before anything is sent.
A share of lapsed clients left uncontacted for the same period. Where volume is too low for a holdout to mean anything, another comparison design is agreed before launch instead of quietly dropped.
Clients returned, revenue from returned clients, interval movement by service, and the response rate by channel and by segment.
Returns we cannot attribute. If a promotion ran in the same window, that goes in the report rather than into a percentage nobody can defend later.
Answer five questions and the brief writes itself, or write to us directly. Either way you get an initial solution brief. A proposed architecture and a fixed quote follow a scoping call, once we have seen your systems.
Build a brief → Contact form