Transparency
Our real numbers
How many watched slots saw a table free up, and after how long. Watches that came to nothing are counted like the rest: without them these figures would mean nothing.
This page measures the service: what a watch bought from us delivers. For what we observe of the restaurant — the pace at which slots open and how long a seat stays bookable — see the observed availability.
We will publish neither a success rate nor a delay before 100 completed slots and 20 observed availabilities. Below that, a percentage would measure only luck: a handful of slots does not make a statistic, and a fragile figure published today would be quoted tomorrow as if it were solid.
Where the sample stands on 19 September 2026 :
- 36 completed slots, meaning those whose outcome is known
- 29 that saw at least one table free up
- 5 still being watched, and therefore excluded from the calculation
- Last updated
- Sample
- 36 completed slots, of which 29 with at least one availability
- Method
- A measurement line is opened when each watched slot starts and closed either at the first availability detected, or at the date of the meal if no table freed up. Both cases count towards the denominator.
- Scope
- Les Grands Buffets, Narbonne. Measurements are GetTable’s own; the service is independent of the restaurant, which shares no booking data.
- Reuse
- Published under the Creative Commons Attribution 4.0 licence (CC BY 4.0): quote, reuse and republish freely, including commercially, provided you credit GetTable and state the date of publication.
How these figures are established
Every watched slot opens a measurement line the moment watching starts. It is closed either at the first availability detected, or when the date has passed without any table freeing up. Both cases are counted.
Delays are segmented by how far off the date was at the start: watching a date three months away and one five days away are not comparable, and a single average would mislead.
Watches stopped by the customer partway through are excluded from the denominator: observation ceased, and counting them as failures would understate the result as much as ignoring them would overstate it.
These measurements contain no personal data: no number, no email, no order reference. The table that stores them has no link to customer accounts, and only totals are published, never an individual slot.
Why a publication threshold
A rate calculated on a handful of slots mostly measures the luck we had. Until the sample reaches the threshold stated above, this page says where collection stands rather than showing a percentage that would flip at the next slot.
Watches still running are not counted: their outcome is unknown, and including them would skew the rate one way or the other.
To understand why tables free up, and at which moments: the cancellation calendar at Les Grands Buffets. For what our readings show of the restaurant itself: the availability observatory.