Second-Visit Rate

Second-visit rate is a cohort measure rather than a period measure, and that distinction is the reason it is useful. A repeat rate describes everybody in a dining room at once. A second-visit rate follows one group of first-timers forward in time and asks a single question about them. Producing it takes first visits that were identified and later visits matched back to the same guest. That requirement is harder to satisfy than it sounds.

How do you measure second-visit rate?

Take every guest whose first visit fell in a given month, look forward across a fixed window, and count how many returned. Sixty and ninety days are the usual windows. The window is not optional, because without one the rate climbs forever: a guest returning after four years still eventually counts.

The reporting trap is comparing cohorts whose windows have not equally closed. Last month’s cohort will always look weaker than the cohort from six months ago, since most of its window has not happened yet. A dashboard listing recent months beside older ones therefore shows a decline that is entirely an artefact. Read only cohorts whose full window has elapsed, and treat the current month as incomplete rather than as bad.

Why is the second visit worth working on?

It is the cheapest point of intervention a restaurant has, because everything needed to act is already known. The guest’s identity is fresh, the date of their visit is on the record, what they ordered may be too, and the absence of a return is unambiguous. Later in a relationship the signals get murkier and the guest’s own rhythm has to be inferred first.

It also compounds, since each additional visit makes the next one likelier. Moving the first step moves every step behind it, which is why operators who work this metric concentrate outreach on recent first-timers instead of spreading it across the whole guest base. The window where the meal is still remembered is short, so this usually runs as a journey firing per guest rather than as a monthly send.

Why does my second-visit rate look worse than it is?

Anonymity hits this metric harder than any other, and the mechanism is specific. A first visit that was never identified means the guest’s actual second visit gets recorded as somebody’s first. One unmatched visit produces two errors: a missing first-timer and a phantom new guest. Rooms with heavy walk-in and bar trade undercount here more than anywhere else in their reporting.

Two real populations also depress the number honestly. Out-of-town guests are structurally unlikely to return whatever the restaurant does, and a venue near a hotel or a tourist district carries a large share of them. Guests brought by somebody else are the other group: they dined, they may have loved it, and the record belongs to whoever booked. Splitting local volume from visitor volume is often the difference between a number that looks alarming and one a team can act on.