There is a version of this article that opens with "the average restaurant loses 40% of its regulars every year." We are not going to write that, because we do not have data to support it and neither does most of the internet that repeats it.
Your lapse rate is a number you can calculate from your own reservation history in about an hour, and yours is the only one that should change what you do. Here is the method.
What you are measuring
Lapse rate is the share of guests who met your definition of a regular at some point and have since stopped coming, measured over a window you choose deliberately.
It is not churn. Churn implies a cancelled subscription — a decision, with a date. Restaurant guests almost never make that decision. They drift, which is why the measurement has to be based on an absence rather than an event.
You need a definition of "regular" first
This measurement is meaningless without one, and there is no universal answer — four visits a year is devotion at a destination restaurant and indifference at a coffee bar.
If you have not set one, do that first. The method for choosing a threshold from your own visit distribution takes an afternoon.
The calculation
Two cohorts, one division. It is genuinely this simple, and the simplicity is why it is worth doing consistently rather than elaborately.
Pick a reference date. Call it TODAY. COHORT Guests who met your "regular" threshold based on visits between 24 and 12 months ago. (A full year of history, ending a year ago.) RETAINED Of that cohort, how many visited at least once in the last 12 months. LAPSED = COHORT − RETAINED LAPSE RATE = LAPSED ÷ COHORT The 12-month offset is the point. It gives every guest in the cohort a full year to come back before you count them as gone.
Why the offset matters
The obvious approach — "regulars who have not visited in 90 days" — measures something real but different. It is a snapshot of who is currently quiet, and it moves with seasonality, which makes it useless for comparing one period to another.
The cohort method asks a cleaner question: of the people who were genuinely regulars a year ago, how many are still with us? That number you can track across years and act on.
Keep both. The 90-day list is your working list for outreach. The cohort figure is your health metric.
Cuts worth making
A single group-wide number tells you almost nothing about what to do. The same calculation split four ways usually points straight at a cause.
- By location. One site lapsing far faster than the others is usually a management or staffing story, not a marketing one.
- By first-visit occasion. Guests acquired through a special event often lapse at a very different rate to walk-ins.
- By spend band. If your highest spenders lapse fastest, that is the most expensive sentence in your business and it should reorder your priorities today.
- By whether they ever visited a second concept. Groups often find that multi-concept guests lapse considerably less, which is an argument for cross-promotion you can actually evidence.
What to do with the number
Do not benchmark it against anyone. You have no comparable, and the restaurants publishing figures are not running your concept in your neighbourhood at your price point.
Benchmark it against yourself. Calculate it once a quarter with the same definition and watch the direction of travel. Four consecutive quarters of improvement is a genuine result you can defend in a board meeting. "Better than the industry average" is a sentence with no information in it, and whoever you are quoting almost certainly made it up.