The problem is staleness
Ask anyone running a busy floor which tables are free and they will answer confidently. Ask again four minutes later and the first answer is already wrong.
This is the whole of table management. Not layout, not seating charts, not reservation software — staleness. The picture someone is seating from is always a few minutes behind the room, and during a rush a few minutes is the entire margin.
It produces two expensive symptoms, and neither shows up anywhere in your sales figures because they are covers you never served:
Cleared tables nobody notices. A table paid and left at 8:20. It gets wiped at 8:24. At 8:31 a party of four is told there is a twenty-minute wait, and they leave. The table was free for eleven minutes and nobody with a waiting guest knew.
Parties seated at tables that were not ready. The reverse error. The table looked free because it was empty, not because it was reset.
Both come from the same cause: the state of the floor lives in a head, and heads go out of date.
Four states, not two
Most restaurants informally run two states — busy or free. The missing ones cost real money.
| State | What it means | Why it needs to be separate |
|---|---|---|
| Available | Clean, reset, ready to seat | The only state a host should seat into |
| Reserved | Held for a booking | Otherwise a held table looks free and gets given away |
| Occupied | Guests seated, order open | Carries how long they have been sitting |
| Cleaning | Paid and vacated, not yet reset | The state that prevents both symptoms above |
Cleaning is the one to add if you add nothing else. Without it, a vacated table is either invisible to the host or visible and dirty, and there is no third option.
Whether these live on a magnetic board, a printed grid the host marks up, or a floor map that updates itself as orders and payments happen, the requirement is the same: one shared view, current, that everyone on shift reads from. Not one per person.
Laying out the floor
Two decisions do most of the work.
Number the tables the way the room reads. Sequentially, in the order someone walking the floor encounters them. Not by when they were bought or which corner they are in. A new captain should be able to find table 14 without asking. This sounds trivial until you watch a trial-shift server lose ninety seconds finding a table during a rush.
Record real seating capacity per table, not a generic count. The host is deciding where to put a party of six. A floor plan that says "20 tables" answers nothing; one that says "four 2-tops, nine 4-tops, five 6-tops, two 8-tops" answers it immediately.
Then identify your problem tables honestly. Every restaurant has two or three: the one by the kitchen door, the one in the draught, the one under the speaker. Do not pretend they are equivalent. Seat them last, seat them with parties who will not mind, or fix what makes them undesirable. A table nobody wants is capacity you paid rent for and do not have.
Seating without stranding capacity
The rule is simple and routinely broken under pressure: seat the smallest table that fits the party.
A party of two at a six-top strands four seats for the whole sitting. On a night when you turn a table twice, that is eight covers you cannot serve, and it usually happens because the six-top was the nearest free thing when the host was busy.
A worked example. A restaurant with five 6-tops, average order value ₹850, turning twice a night. If two of those six-tops carry parties of two for both sittings:
- Stranded seats: 2 tables × 4 seats × 2 sittings = 16 covers
- At ₹850, roughly ₹13,600 a night of capacity that could not be sold
That is not a rounding error. Over a month of weekends it is more than most restaurants spend on marketing.
Two practical rules that prevent most of it:
- Hold your large tables for large parties during peak — with a cut-off time after which they go to anyone, so you are not holding empty capacity all evening either.
- Combine and split deliberately, not opportunistically. If two 4-tops make an 8-top, know that in advance and know how long it takes to reset them.
Waitlists people accept
When the room is full, the waitlist is your product for the next twenty minutes, and it is mostly a communication problem.
Quote pessimistically. If you think it is twenty minutes, say twenty-five. Beating your own quote by ten minutes generates goodwill out of nothing; missing it by ten generates a bad review — even when the absolute wait was identical. Quoting optimistically to stop someone leaving converts a walk-away into a complaint, which is a worse trade.
Take a number, do not make people stand. A party standing at the door watching your service is a party auditing your service.
Update proactively. If the estimate moves, tell them before they ask. The wait is rarely what annoys people. Being forgotten during it always is.
Know which table is turning next. This is what makes an honest quote possible, and it comes straight from tracking how long each occupied table has been sitting.
Measuring it properly
Two numbers, and the second is the one most restaurants skip.
Turnover — how many parties a table serves in a service. Track it per table, never as a floor average. The average tells you nothing you can act on; the per-table figure tells you that table 14 turns once while table 6 turns three times, which is a specific problem with a specific cause.
Occupancy — the share of seats actually filled across the service. This is the number that catches stranded capacity. A restaurant can have excellent turnover and poor occupancy if it is seating twos at six-tops all night, and turnover alone will never reveal it.
Then break the table's cycle into four segments and time them for a fortnight:
- Seated → first order placed
- First order → food delivered
- Last course → bill requested
- Bill settled → table reseated
Segments 1 and 4 are where the recoverable time is. Segment 2 is your kitchen, and the fix is in restaurant operations. Segment 3 is the meal, and compressing it is the one form of turnover improvement that costs more than it earns.
Common mistakes
Running the floor from one person's memory past twenty tables. It is not a discipline failure. Memory genuinely cannot hold a changing 25-table floor during a rush.
Skipping the cleaning state. Guarantees you will either strand cleared tables or seat dirty ones.
Seating the nearest free table. Under pressure this is the default, and it is how large tables get stranded.
Quoting optimistic wait times. Trades a walk-away for a bad review.
Measuring turnover as a floor average. Hides exactly the tables you need to know about.
Treating the floor plan as the host's private tool. Its value is that captains covering each other's sections see the same picture.
What to do this week
- Add the cleaning state, however you track the floor. It is the cheapest fix on this page.
- Renumber tables so they read in walking order, and put real seat counts against each.
- Name your three worst tables and decide what you are doing about each.
- Set a large-table hold rule for peak, with a cut-off time.
- Time the four segments for a fortnight on every table.
- Quote wait times five minutes longer than you think for two weeks, and watch what happens to the mood at the door.
The short version
- Table management is a staleness problem. One shared, current view beats several private accurate-a-few-minutes-ago ones.
- Track four states. Cleaning is the one most restaurants are missing and it causes both expensive symptoms.
- Seat the smallest table that fits. Stranded seats at large tables are invisible and costly.
- Quote wait times pessimistically. Beating your own estimate is free goodwill.
- Measure turnover per table and occupancy alongside it. Averages hide the tables costing you money.
- Recoverable time is before the meal and after it, never during.
