Why the drawn plan always promises more
A 700 by 700 mm four-top with four chairs tucked under it occupies about 1.0 square metre of drawn area. Put four guests in it and the same table needs 1.55 square metres, because each chair has to come out roughly 450 mm to be sat on and again to be left. That is a 55 percent increase on the identical piece of furniture, and it is the entire mechanism behind a seat count that nobody can serve.
The error compounds rather than averaging out. Each table is drawn in isolation, so each table is under-sized in the same direction. A room of 30 tables does not under-count by a random amount that cancels; it under-counts by 55 percent of the chair zone on every single table, then subtracts circulation from whatever is left, which is already too little. By the time the plan reaches a contractor the shortfall is structural.
This matters commercially because the seat count is not a drawing detail. It drives the covers forecast, the staffing rota, the kitchen pass width and often the rent negotiation. A 118 seat room quoted as 154 seats is a business case built on a number that exists only in CAD.
The three clearances that decide the count
Chair pull-out is the first. Allow 450 mm from the front edge of the seat to the nearest fixed object, 500 mm for armchairs and for a guest who is older or less mobile, and 600 mm where a wheelchair has to transfer. Measure from the seat front and not from the table edge, because a chair that is 480 mm deep eats most of the gap before the guest has moved at all.
Chair back to chair back is the second. Two guests seated facing away from each other need 900 mm between the backs at minimum, 1100 mm to be comfortable, and 1200 mm if a server has to pass between them carrying anything. Add the 450 mm pull-out on both sides and the honest table-edge to table-edge distance for facing rows is 1800 mm, not the 1300 mm that appears on most drawings.
The service lane is the third, and it is the one most often deleted when the seat count comes in short. A single server with a tray needs 1200 mm of clear width; two servers passing need 1500 mm. Drop the lane below 900 mm and the room still seats the same number of guests, but the covers per hour fall by roughly a quarter, because the bottleneck moves from the kitchen to the floor. Selling 118 seats at 3.5 turns beats selling 140 seats at 2.6 turns every week of the year.
Working the arithmetic on a real room
Take a 260 square metre shell with 180 square metres of usable dining floor after the entry vestibule, the restrooms and the service station. Reserve 26 square metres for the circulation spine and the cross aisles at 1200 mm. That leaves 154 square metres of table zone.
At the drawn, pushed-in density of 1.0 square metre per cover, 154 square metres reads as 154 seats. At the serviceable density of 1.55 square metres per cover for a four-top, it is 99 seats. Real rooms land between the two because the mix is never uniform, and a typical full-service layout settles around 118 covers, which is a 23 percent overstatement. If the plan is heavy on two-tops the gap widens to 30 percent, because a 700 mm two-top is the least efficient table in the building at roughly 2.05 square metres per cover once pull-out is counted.
The same room laid out with 1200 by 750 mm rectangular six-tops comes back at 1.35 square metres per cover, and a 1000 mm round four-top at about 1.45. That is why round restaurant tables keep appearing in rooms where the brief says maximise covers: a round table has no dead corner, so the chair zone wraps instead of colliding. It is also why the table list and the seat count have to be produced together. restaurant tables ordered from a catalogue before the plan is fixed will always be the wrong size for the room they land in.
Dead seats and the table mix
Not every seat on a plan is sellable. Four categories are effectively dead at peak: the seat inline with the queue at the entry, the seat beside the pass where plates are landed, the seat at the end of a banquette run next to a service door, and the seat at a two-top pushed against a wall where only one chair can be pulled out. In a 118 seat room that is 6 to 9 seats, or 5 to 8 percent, and a host who seats them anyway damages the turn time of everybody around them.
A banquette is the exception that proves the rule. Fixed bench seating runs at about 0.9 square metres per cover because there is no pull-out to allow on the wall side, which makes it the densest seating in any restaurant. The trade is flexibility: a banquette cannot be recombined for a party of nine, and a 30 metre run of it locks the room into one table rhythm for the life of the fit-out. Rooms that host private events want 60 to 70 percent loose tables and 30 to 40 percent fixed, not the other way round.
Then there is the reconfiguration allowance. If four-tops have to become a ten-top twice a week, every four-top in that zone needs 600 mm of slack on one side, and that slack is drawn as aisle but functions as nothing. Count it as lost seats, not as circulation, or the same square metre gets used twice in the forecast. A room built mostly from four-cover dining sets has the least slack of any layout, because a 700 mm square with four chairs has no side that can open without borrowing from the next table.
Chairs, spares and what a seat actually costs
Once the seat count is honest, the chair order is the count plus a spare allowance, not the count. A 118 seat room wants 130 chairs: 12 spare covers the stack that lives in the back for the terrace, the four that get taken out of service for a loose joint, and the two that a delivery driver will eventually break. Ten percent is the working figure, and it is cheaper than a second production run in month fourteen, which is when a container minimum order and a fresh freight bill make a 12 chair top-up cost 40 percent more per unit.
Chairs also have a footprint that the seat count ignores. A stack of eight chairs in the corner of the room is 0.9 square metres of floor doing nothing at 20:00 and everything at 11:00, and if the plan has no store for it the stack ends up in the aisle it was supposed to clear. Rooms without a chair store usually buy 8 to 12 fewer chairs than they need and simply accept a lower peak, which is a quiet way of paying for the omission every service.
When the count and the mix are settled, the take-off becomes mechanical: 20 four-tops at 700 mm square, 14 two-tops at 700 by 700, 6 six-tops at 1200 by 750, and 118 covers plus 12 spare. That list is what we quote from, and it is the only version of the project where the chair, table and banquette quantities reconcile. Buying restaurant chairs before the plan is signed is how a room ends up with 40 chairs that fit the drawing and not the tables. Ordering restaurant table and chair sets for your floor plan avoids half of that problem by fixing the pair at the source, but only if the plan is already counted honestly.
Reading the plan against the specification
Two documents have to agree before production: the floor plan with clearances, and the furniture schedule with dimensions. They disagree more often than not, and the disagreement is always in the same three places. The plan draws a 700 mm table and the schedule specifies a 750 mm top with a 40 mm overhang, so the chair-to-chair gap shrinks by 80 mm in a zone that had no slack. The plan assumes a chair 450 mm deep and the schedule specifies an upholstered armchair at 560 mm. The plan assumes chairs tuck fully under and the schedule specifies arms.
The fix costs a morning. Overlay the schedule dimensions on the plan, redraw the pull-out zone at the specified chair depth rather than the assumed one, and recount. If the number drops by more than the model tolerates, the answer is usually to change one table size rather than to delete an aisle. This is the same discipline we describe as restaurant furniture layout planning, and it is why we ask for the plan and the chair depth together rather than the plan alone.
There is a shortcut that catches the remaining errors, and it is worth the two days it takes. Tape the table footprint on the actual floor at 1:1, put one real chair on it, sit down, stand up, and walk past with a tray. Then do it with a second chair opposite, and then with the row of tables either side in place. A 700 mm four-top that looks generous on paper is tight the moment two guests are seated and a third walks behind them. Every project that has done this has changed at least one table size before signing, and none of them regretted the two days.
Send the plan with the clearances marked and we will return a seat count, a table mix and a chair quantity that survive a Saturday. If the room also has a fixed banquette or a private dining partition, mark those as fixed and we will count around them instead of through them. send your floor plan with the chair depth you intend to use written on it, because that one figure moves the count more than any other. The number that comes back will be lower than the one on your drawing, and it will be the one you can staff.