Small Tables Book Today, Groups Book Ahead: What 1,582 Reservations Show

2026-08-26 · Research

A table for two is stock you sell on the day. A table for eight is stock you sold last week. That is the clearest thing in the data we pulled from Stolio on 25 August 2026: 1,582 reservations from 13 restaurants, and a straight line running from small parties to large ones. The smaller the booking, the later it arrives.

Here is the whole finding in one table. The percentage is the share of bookings of that size that were made within 24 hours of the meal.

There is no wobble in it. Every step down the list is a step towards booking earlier, and the gap between the ends is enormous: seven out of ten bookings for one or two people are made inside a day, while just over one in three large groups is. A party of eight has to agree on a date with eight diaries. Two people decide at six that they cannot face cooking.

Across everything, 63.5% of bookings arrive inside a day. 40.1% of the 1,582 reservations were made on the day of the meal itself, 63.5% within 24 hours, and 89.6% within a week. One caveat belongs right here rather than further down: anything staff typed in at the door on the night lands in the same-day bucket by definition, and we cannot separate a walk-in written into the system at 20:10 from a table booked that morning, so part of the 40.1% is paperwork rather than a booking decision. The median lead time is one day. The mean is 3.5 days, and the difference between those two numbers is a small tail of bookings made more than a week ahead — 10.4% of the sample fell outside a week — which pulls the average up and tells you very little about a normal Tuesday.

Large tables skew earlier: only 35.5% of bookings for seven or more arrived within 24 hours. Small tables do not. 71.4% of bookings for one or two people and 66.3% of bookings for three or four were made within a day of the meal. Even at five and six covers, 57.1% were booked within 24 hours — fewer than for one or two, but still most of them. Those are two different businesses happening in the same dining room, and they do not respond to the same things. If you hold anything back from the online diary, hold back two-tops, not the big table.

Bookings build through the afternoon and peak at four o'clock, an hour before service. Converted to Prague time, 16:00 is the single busiest hour for taking a booking — 10.0% of the 1,582 arrived in that hour alone, with 17:00 next at 8.9%. And 28.9% of all bookings arrived between five and nine in the evening: while the dining room was filling up and the phone was least likely to be answered. Only 8.2% arrived before nine in the morning or after ten at night, so this is not a story about people booking in the small hours.

Saturday is both the busiest day and the most planned-ahead one. Of the 477 Saturday reservations — far more than any other day — 57.2% were made within 24 hours, the lowest share of the week. Tuesday is the opposite: 73.2% of its 153 bookings arrived inside a day. Monday, Sunday and Wednesday sit in the high sixties, Friday at 63.9% across 299 bookings, Thursday at 61.4%. We did not break party size down by day of the week, so we cannot say why Saturday sits where it does.

The channel barely matters, which surprised us. Bookings made through the restaurant's website were 63.4% within 24 hours across 1,187 reservations. Bookings staff entered by hand — phone, in person, walk-ins written straight into the system — were 62.8% across 352. That is a gap of 0.6 points. If anything the manual figure is flattered, because anything typed in at the door on the night lands in that bucket by definition, and it still did not come out ahead. On this data, channel does not separate early bookers from late ones. Why that is, we cannot say from these numbers. (A third channel, reservations arriving from the Syrve POS, sat at 74.4%, but that is only 43 bookings and we would not lean on it.)

It also holds up restaurant by restaurant. Looking only at the six venues with 30 or more bookings, the share booked within 24 hours ranged from 45.2% to 83.3%: 64.2% at the venue with 723 bookings, 70.1% at 321, 45.2% at 250, 56.1% at 114, 72.4% at 98 and 83.3% at 42. That is a wide spread, and it is worth saying plainly that a restaurant's own number depends on its own guests. But even the most planned-ahead venue in the set had nearly half its bookings arrive inside a day.

One thing you should know before trusting any of this. The largest restaurant accounts for 723 of the 1,582 reservations — 45.7% of the sample. That is a lot of weight in one place. The reason we are still comfortable publishing the headline is that this venue's own figure is 64.2%, almost exactly the 63.5% overall, so it is not dragging the average anywhere. It is a large voice saying the same thing as everyone else. The party-size gradient is what it is, but on a smaller base once you set that venue aside, so treat the exact percentages as indicative rather than precise.

Three more limits worth stating. These are 13 restaurants that use Stolio, not a sample of Czech restaurants. Nothing here should be read as "restaurants in the Czech Republic book X% within a day" — it is what these 13 did. The window is short: services from 17 March to 24 August 2026, five months of spring and summer. December is not in this data at all, and December is precisely when large group bookings, work parties and set menus behave differently. And we applied no filter on status: the 1,582 includes bookings later marked cancelled, we have not recalculated the headline with them removed, and because the status field in these accounts is only updated occasionally we would not trust it to strip them out cleanly.

The thing to do today costs nothing. Work out your own split. Take the last four Saturdays and the last four Tuesdays, put the date each booking was made next to the date of the meal, and subtract one from the other. Twenty bookings a day is enough to see the shape. Count how many came in on the day, how many within 24 hours, and how many more than a week out. If your diary is on paper, that is one coffee, not an afternoon.

Then do it again, split by party size. Four buckets is enough — 1–2, 3–4, 5–6, 7 and up. If your gradient looks like ours, you now know which half of your room is decided in advance and which half is still up for grabs at four in the afternoon. If it does not look like ours, that is more interesting still, and your number is the one that should drive your decisions, not ours. A restaurant doing a lot of set-menu group work will look nothing like a neighbourhood bistro, and both of them are right about themselves.

What you do with the number depends on what it says. A high same-day share means the last two hours before service are worth staffing properly, because that is when your remaining covers are being decided. A low one means your covers are largely settled before service starts, so the phone at seven matters less than it feels — whether that is good or bad depends on whether you are full. Either way, you are working from your own guests rather than a rule of thumb.

We publish this kind of thing because we have the data and we would rather write down what it actually says. Stolio is the reservation system these bookings came through, and it stores the time each booking was made, which is where these numbers came from.