Every room has seats that fill last and earn least. They are rarely bad seats — they are seats nobody has measured. How to find them, and the cheap changes that usually work before moving furniture.

There is a table by the service door. It fills at 23:30 on a Saturday and never otherwise.

Everyone knows it is the bad table. Nobody has ever measured how bad, or why, and both answers are usually surprising.

The number that does not exist in most venues

Revenue per seat, by position.

Total revenue is known. Revenue per night is known. Revenue by where people actually sat is almost never known, because tabs attach to a table number and nobody ever groups by it.

⚠️ It is one query away in most systems and it changes the floor plan conversation completely — from opinions about which tables are good to a ranked list.

What the ranking usually shows

Three findings recur:

The worst table is not the one everyone names. The obvious bad table is filled last but spends normally once occupied. The genuinely worst is frequently a table that fills early and turns slowly.

One or two positions carry a surprising share. Usually a specific banquette or corner, and it is worth knowing which, because that is the table you should be protecting for the right guests.

The middle is undifferentiated. Most tables perform within a narrow band, which means the conversation should be about the extremes, not about redesigning everything.

Why a table underperforms

Four causes, in rough order of how often they turn out to be it:

It is never offered. Staff assign from habit. A table that gets offered fourth fills fourth regardless of whether it is any good.

It is hard to serve. Distance from the bar shows up as slower rounds and fewer of them. ⚠️ This is the one that is genuinely structural and the one people notice least.

It is exposed. Next to the door, the toilets, the service route. Guests sit there and do not settle.

It is the wrong size. A four-top used by twos all night is earning half of what the footprint could.

The cheap fixes, before moving anything

Change the offer order. Free, reversible, and it tests the "never offered" hypothesis in one week.

Change what it is for. The awkward two-top by the door is a bad dinner table and a good waiting-for-a-friend table. Assigning a purpose beats trying to make it a normal table.

Change what it is worth. If it genuinely is worse, pricing it differently is honest and works — but only after the first two, because otherwise you are discounting a table that was simply never offered.

Then move furniture. Expensive, disruptive, and frequently unnecessary once the first three are done.

The measurement trap

Revenue per seat rewards tables that turn quickly, which is not always what you want.

⚠️ A table hosting a four-hour party at high spend may rank below a table doing three fast rotations, and the four-hour party may be the better guest. Look at revenue per seat and revenue per visit, and understand which one your room is optimising for.

The thing worth doing once a year

Sit at the worst table for twenty minutes during service.

Not a metaphor. Most owners have never done this, and it answers "why" faster than any report — the draught, the noise, the sightline, the fact that nobody walks past.

Seating decides it before the guest does

Where the host walks first  the tables that fill
Where the tray route runs  the tables that get attention
Where the section line sits  the table that belongs to nobody

⚠️ A table on a section boundary is served by whoever notices, which on a busy night is nobody.

The check

⇒ Look at the section map. Any table touching two sections is a candidate before you conclude anything about the table itself.

Three cheap causes before the expensive one

1. It is always seated last  a habit, not a property of the table
2. It is out of sightline   fixable with a mirror or a route change
3. It is genuinely bad     cold, loud, by the toilets, in the way

⚠️ One is the most common and costs nothing to fix. Rotate which table gets seated first for two weeks and re-measure.

What "underperforming" should mean

Covers per available hour  not revenue
Revenue per cover      separately

⚠️ Combining them hides which problem you have. A table with normal covers and low spend is a service problem; low covers is a seating problem.

The fix that makes it worse

Discounting the bad table  now it attracts the guests who spend least

⚠️ You have not fixed the table; you have changed who sits there, and made the number look better while the room earns less.

When to actually move furniture

After the seating habit has been tested and ruled out
After the sightline has been tried
And only in a quiet period

⚠️ Moving furniture is the most visible and least reversible option, which is why it tends to be chosen first and regretted.

The yearly walk

⚠️ Sit at every table for two minutes, in service, once a year. 　 ⇒ You will find things that no report contains: the draught, the speaker, 　　 the angle where nobody can catch a server's eye.

Three to hold

Revenue per seat by table, last quarter. The ranking.

Offer order, observed. Which tables get filled first, and whether that matches the ranking.

Occupancy rate by table. Separates "bad table" from "table nobody offers."

Where the record has to sit

If tabs are attached to tables but nobody groups by table, the data exists and the answer does not. That is the situation in most venues — the measurement is not missing, the query is.

tasteck records sales against tables and visits, so revenue and occupancy per position is a report rather than a floor plan argument.

The bad table might be a good table nobody has offered since the layout changed two years ago.

The Table Nobody Sits At