Ask around and you will hear "four to six tables per server" as if it were a law. It is a rough average across very different restaurants, and following it blindly is how you end up either overstaffed or drowning. Here is a more useful way to think about it.
Why the standard number misleads
Four tables in a fine-dining room with a five-course tasting menu is a full workload. Four tables in a café where people order a coffee and a croissant is close to idle. The number of tables is not the workload — the number of interactions is.
The four things that actually decide it
- Covers per table. A four-top takes roughly twice the work of a two-top, not the same.
- Courses per meal. Each course is a trip out and a trip back. Three courses roughly doubles the movement of one.
- Table turn time. A table that turns twice in a service is served, cleared and reset twice.
- How much of the ordering the guest does. If guests order from a QR menu at the table, the taking-the-order trip disappears entirely.
A rough calculation
Count the trips one table generates over a full sitting: greet, take order, bring drinks, bring each course, check back, clear, present the bill, reset. In a typical bistro that is around 10 to 12 trips for a three-course table.
Now assume a server can sustain about 60 to 70 purposeful trips per hour of service without quality dropping. Divide, and you get the number of tables one person can hold at that turn rate. For most bistros it lands between five and seven — higher than the folk wisdom, because the folk wisdom came from full-service restaurants with more courses.
Measure it instead of guessing
You do not need a stopwatch. If your order taking is on a tablet, the data is already there: which server took each order, at what time, and how long each table stayed occupied. Two weeks of that tells you more than any rule of thumb:
- Average time a table stays occupied, by day and by service.
- Which sections consistently run late.
- Which dishes stretch the kitchen and therefore the floor.
Modern Resto records the server's name and a timestamp on every ticket, and reports time spent at table, so you can size sections from your own numbers rather than from a forum post. It is part of the free order-taking side.
Two adjustments worth making
Uneven sections beat equal ones. Give your fastest server the section nearest the kitchen and one extra table; give a new starter one fewer. Equal sections feel fair and serve guests worse.
Cut trips before adding staff. A QR menu that lets guests read the card without waiting, and a kitchen screen that removes "is it ready?" trips, both reduce workload per table. Often that is cheaper than another shift.
The signal you are over the line
Not complaints — those come late. Watch for drinks arriving after starters, or check-backs being skipped. Both happen well before guests say anything, and both mean the section is one table too big.