Most stores treat every customer the same way when it comes to winback: wait some fixed number of days (often thirty), then blast a generic discount to everyone who has gone quiet. It feels like a reasonable rule. It is also a rule that ignores something obvious the moment you look at real order data: your repeat customers do not share a single rhythm. They each have their own.
The pattern hiding in real order data
Look at any perfume or beauty store's order history for long enough and the pattern becomes visible. One customer restocks her signature scent roughly every 23 days, almost like clockwork, because that is how long a bottle lasts her. Another orders every 41 days, because he buys for himself and his wife in the same trip and one bottle covers more ground. A third orders erratically around specific occasions: Ramadan, a birthday, National Day. None of these are wrong. They are just different clocks, and a single 30-day rule will always be too early for some customers and too late for others.
Too early means you nag someone who was never going to reorder yet, and the message reads as spam. Too late means you wait until a genuinely lapsed customer has already found a substitute, or simply forgotten you exist, by which point winning them back costs far more than reaching them the week they were actually due.
Rizecart computes each customer's own cycle from their real order history: the average number of days between their past orders. From there, "overdue" is personal. A customer with a 23-day cycle who has gone 35 days without ordering is meaningfully overdue. A customer with a 60-day cycle at day 35 is right on schedule and should not hear from you yet. The customer view in Rizecart shows this as a simple overshoot marker: how far past their own clock someone has drifted, in plain days and percent, not a generic "30 days since last order" column that means something different for every customer it is applied to.
The lever matters as much as the timing. Not every overdue customer responds to the same nudge. Some respond to a plain reminder. Some respond better to being shown a bundle. Some need to see a new product first, a soft "have you seen this" rather than a hard sell. Rizecart learns which lever has worked for a given customer in the past and uses that one again, rather than sending the same generic discount to everyone who has gone quiet regardless of what actually moved them last time.
Every winback still keeps its holdout, so "we brought her back" is a measured claim, not an assumed one. The point is not to message people more. It is to message the right person, in the week their own rhythm says they are actually due, with the lever that has already worked on them once.