When is a restaurant guest actually lapsed?
Ninety days is the number you will be sold. It is a quarter of a calendar year, and calendars have nothing to do with how people eat. Here is how to find yours.
There is no universal number, and any provider who gives you one without looking at your data is giving you their reporting convenience, not your answer. A guest is lapsed relative to their own rhythm. Take the guests who have visited at least twice, measure the gaps between their visits, and your threshold sits at roughly two to three times the typical gap. For a weekly lunch spot that lands near three weeks. For a special occasion restaurant it can be most of a year. Both are correct, and neither is ninety days.
Why ninety days is the number everyone repeats
Because a quarter is convenient to report on, and because it is safely in the middle. It is not wrong so much as arbitrary: it was never derived from anything about your restaurant, and it will be roughly the same number whether you serve twelve covers a night or two hundred.
The tell is that nobody who quotes it asks how often your regulars come. If the answer does not change when the question changes, it was not an answer.
The number you actually want, and where it lives
It is already in your point of sale, and you do not need a new system to find it. What you need is one list: every guest who has visited at least twice, and the number of days between each of their visits.
A guest with a single visit has no gap, so they cannot help you here. They matter for a different problem, which is whether a first visit ever becomes a second.
Three steps
- Pull the gaps. For each returning guest, the days between visit one and two, two and three, and so on. One row per gap, not per guest.
- Find the middle. The median gap is your restaurant's natural rhythm. Use the median rather than the average, because a handful of guests who came back after two years will drag an average somewhere useless.
- Find the edge. Sort the gaps and look at where the top ten or twenty percent begins. That is the point where a gap stops being normal and starts being unusual for your own guests.
| What you measure | What it tells you |
|---|---|
| Median gap | The normal rhythm. Contacting inside this is nagging. |
| 80th percentile | Drifting. Still warm, still cheap to bring back. |
| 90th percentile | Lapsed. The habit is at risk. |
| Beyond that | Dormant. Recoverable, but it is a different message. |
The useful part is that this gives you three moments instead of one. The guest who is drifting does not need a discount, they need a reason and a reminder. The guest who is dormant needs to be told the place still exists. Sending both of them the same twenty percent off is how a list gets tired.
One number for the whole list is still wrong
Most restaurants have at least two populations that share a dining room and nothing else. The weekday lunch regular who comes every Tuesday and the couple who come for their anniversary are not on the same clock, and averaging them produces a threshold that is too late for one and too early for the other.
If you can split the gap calculation by anything meaningful, do it. Lunch against dinner is usually the cleanest split and the one your data already supports. Two thresholds that fit are worth more than one average that fits nobody.
What it costs to get wrong, in both directions
Too early
You contact people who were already planning to come. You hand a discount to guests who needed none, and over time you teach a slice of your regulars to wait for the offer before booking. This is the more expensive error and the harder one to notice, because the report looks excellent: the contacted group came back at a high rate. They were coming anyway, and without a holdout group nothing in the numbers will tell you that.
That trap has its own page: how to tell if a win-back campaign actually made money.
Too late
The habit has already been replaced. Nobody decides to stop going to a restaurant. They go somewhere else a few times, and the old place quietly stops being on the list of things they think of on a Thursday. Reaching someone while the habit is still warm is a different and much easier conversation than reminding someone that you exist.
Questions owners actually ask
Is ninety days a reasonable default for a restaurant?
It is a reasonable default only in the sense that any round number is. Ninety days is roughly a quarter, which is a reporting convenience rather than anything about how people eat. For a neighbourhood spot whose regulars come weekly, ninety days is far too late and the guest is long gone by then. For a special occasion restaurant, ninety days is far too early and you are chasing people who were never leaving.
How much visit history do I need before this works?
You need guests with at least two visits, because the whole calculation is about the gap between visits and a single visit has no gap. A few hundred guests with two or more visits is enough to see the shape. If almost nobody in your data has a second visit, that is itself the finding, and it points at capture rather than at campaigns.
Should the threshold be the same for everyone on the list?
No, and this is the most common way the calculation gets wasted. A weekday lunch regular and a twice-a-year anniversary guest are not on the same clock. Compute the gap distribution separately for the groups that behave differently and you get two or three thresholds, which is far more useful than one average that fits neither.
What happens if I set the threshold too early?
You contact people who were already planning to come, so you give a discount to guests who needed no discount, and you train a portion of your regulars to wait for the offer. It is the more expensive mistake of the two, and the harder one to see, because the campaign report looks excellent: the people you contacted came back at a high rate. They were coming anyway.
What happens if I set it too late?
The habit has already been replaced by the time you reach out. Nobody decides to stop going somewhere. They just go somewhere else a few times and the old place stops being an option they think about. Reaching a guest while the habit is still warm is a different conversation from reaching one who has to be reminded the place exists.
Where this fits
Betancur Studio builds automated retention systems for independent restaurants in the United States, using the guest data a restaurant already has. The system works the point out per guest where the records allow it: a guest with at least three visits is contacted at two and a half times their own usual gap. Without that history it uses the ninetieth percentile of the restaurant's guests who have one, and only when neither exists does it fall back to forty days.
Timing is one of several levers a restaurant can pull before spending anything on ads. The full list is on how to increase restaurant sales without advertising.
The mechanics of the campaigns are on the email marketing page. If you want a straight answer on whether your guest list can bring people back before buying anything, the free fit check is yours to keep either way.
Get my free fit checkNo card, no call required.