The assumption is that guests stop coming because of a bad experience. Occasionally true. Far more often they simply drifted — a habit changed, somewhere new opened, life got busy — and the restaurant fell out of mind without any decision being made.
That is good news, because drift is much easier to reverse than dissatisfaction.
There is no universal number. A city-centre lunch spot might consider four weeks lapsed; a special-occasion restaurant might not worry until six months. The useful definition is relative to that guest's own pattern: someone who came every fortnight and has not been in for two months has changed behaviour. Someone who comes twice a year has not.
This is why blanket "haven't seen you in 90 days" campaigns underperform — they treat both guests identically.
Reach out too early and it is odd. Too late and the habit has been replaced. The window is usually shortly after the guest's own normal gap has passed, which is only knowable if you have their history.
The reflex is money off. But a discount to someone who drifted for no particular reason spends margin you did not need to spend, and sets an expectation for next time.
More effective: acknowledge them, mention something relevant — a dish they ordered before, a new menu in a category they liked — and give a reason to come now rather than eventually. Save discounting for guests who genuinely need the nudge.
Email open rates in hospitality are low. Messaging channels get read far more reliably, which matters more than reach for a win-back message aimed at a small, specific list.
Not every lapsed guest is winnable. Some moved, some changed jobs, some did have a bad meal and did not say. A win-back campaign that recovers a meaningful minority of a well-targeted list is doing its job — measure it against the cost of acquiring the same number of new guests, which is the honest comparison.