Customer Reactivation: Follow Up on the Right Schedule

A customer who comes back every six weeks and one who comes back once a year shouldn't receive the same overdue reminder.

Reactivation starts with a simpler question: when would it be reasonable to expect this person back, and is there a useful reason to contact them now? That question shapes the reactivation systems we build, with timing set for each service and AI helping with the conversation. This guide covers how to set timing rules for each service, check eligibility before anything is sent, use AI where it helps and measure what actually comes back.

There's no single definition of a lapsed customer. For each service, the business needs to agree the usual interval between visits, a grace period and the exceptions.

Separate the statuses that need different handling: overdue, cancelled, paused, already booked and genuinely unknown.

Each customer needs a date when it makes sense to look at their record again. Base it on:

Start with clear rules for each service. They're easy to explain and check, especially when history is sparse or messy. Where there's plenty of clean history, AI can suggest a likely pattern for a person to review. A predicted date isn't automatically better than a clear rule.

Personalisation should use relevant business history, such as the service and when it was last done. It shouldn't rely on intrusive detail or guess why someone hasn't been back.

Before any message goes out, check that contacting this person still makes sense:

If the history can't be trusted, send the record to a person for review instead.

Keep one coordination record per customer, so someone who uses several services doesn't receive competing reminders. Apply business-hours and frequency limits, and check eligibility again at the moment of sending, because a booking may have been made since the list was built.

AI can help in a few specific ways. It can draft a short, relevant message from approved facts, ask routine questions and offer valid booking times.

Include identification and opt-out handling suited to the message's purpose. The ACMA's guidance on avoiding spam covers consent, identifying the sender and unsubscribing for commercial messages.

AI shouldn't invent a past service, promise a result or pretend to be a person. Uncertain history, complaints, complex needs and requests to speak to someone go to the team. As with any automated booking, the system should only say a booking is made once it's confirmed in the calendar.

Track each stage: eligible customers, delivered messages, replies, confirmed rebookings, attended visits, and the net revenue and margin that follow. Record cancellations and unsubscribes too.

A rebooking isn't necessarily a recovered customer or an extra sale. Some people would have come back anyway. Where it's practical, compare against a holdout group or a similar group that wasn't contacted before crediting the workflow.

Review results by service and by how overdue people were when contacted. A booking that's never attended isn't retained revenue.

Reactivation workflows need looking after:

Automation can reduce overlooked follow-up, but it can't guarantee that nobody is ever missed. Data feeds break, phone numbers change and bookings get entered in the wrong place, so a way to recover belongs in the process.

Start with one service and a measurable group of customers. Review the timing, tone and replies with the team before expanding. For the economics behind retention, see the growth ceiling.