Staff

Kitchen Ticket Times: The 4 Numbers That Explain Why Guests Still Complain

Your ticket printer or POS has shown it for months: an average of 14 minutes from ticket to plate. Comfortably inside what you consider normal. And yet there's a table getting visibly impatient mos…

In this article
  1. The four numbers
  2. Which of the two problems do you have?
  3. Calculate your own kitchen's tail and its value
  4. What to do this week, this month and this quarter

Your ticket printer or POS has shown it for months: an average of 14 minutes from ticket to plate. Comfortably inside what you consider normal. And yet there's a table getting visibly impatient most busy nights, and another review this month calling the service slow — when service wasn't the problem at all. The average tells you the kitchen is running fine. It tells you nothing about the one ticket a night that decides everything.

Every kitchen with a POS or a kitchen display system knows its own average ticket time — the time between a ticket firing and the plate leaving the pass. Most owners look at that single number, see it sits inside their own norm, and assume the kitchen is 'fast enough'.

That average is the least interesting number your system keeps. A guest who is served on time doesn't remember it. A guest who once waits twelve minutes longer than expected remembers it — and tells people. Your restaurant's reputation isn't set by the average. It's set by the tail: the tickets that blow past it.

This article walks through four numbers that together explain why a kitchen with a perfectly fine average still generates complaints — and, more importantly, which of the four is actually your problem. Because the fix is completely different depending on the answer: a genuinely slow kitchen is fixed with extra hands or tighter prep. A kitchen that is fast enough on its own but lets tickets pile up against each other is fixed by changing how tickets are sequenced — and those two fixes cost wildly different amounts.

The four numbers

1. The average — the number you already have

Average ticket time is the sum of every ticket divided by the number of tickets. It's the number every POS and kitchen display system shows by default, and it's a decent signal for the general health of your pass — an average that slowly climbs over weeks points at a real capacity problem. But an average is a summary, and a summary hides exactly the tickets a guest remembers. Two kitchens with the exact same 14-minute average can deliver completely different guest experiences: one kitchen's tickets all land between 12 and 16 minutes, the other swings between 8 and 28.

2. The tail — the number nobody tracks

What a guest remembers isn't the average, it's the worst case that happened to THEM. Statistically that's the 90th percentile of your ticket times: the duration within which 90% of tickets are ready, and therefore also the point above which the slowest 1 in 10 tickets fall. At a 14-minute average, that 90th percentile realistically sits between 22 and 30 minutes — a gap of 8 to 16 minutes between the number you see and the number one in ten guests actually feels. That gap, the 'tail gap', is the first figure in the calculator at the bottom of this article: the bigger the gap between your average and your worst case, the less predictable the kitchen feels to a guest — even while the average stays exactly the same.

The average hides the tail

The same 14-minute average ticket time, two completely different guest experiences: a predictable kitchen where almost every ticket lands between 12 and 16 minutes, against an unpredictable kitchen that averages the exact same 14 minutes, but where one in ten tickets takes well over 25 minutes.

Predictable kitchen
Unpredictable kitchen

3. The held-ticket share — usually the real cause

A long tail rarely comes from a kitchen that is structurally too slow. It almost always comes from tickets that WERE ready on time but got held at the pass, waiting for another element of the same table to catch up so everything can go out together. One table ordering a steak 'well done' while the rest want medium, one starter held back until the main of an adjacent table is also ready for simultaneous service, one plate sitting a minute under the heat lamp because expo is dealing with something else: those aren't slow tickets, they're held tickets. Count, for one week, how many tickets sat at the pass longer than it took to actually cook them, divided by total tickets. Above 15%, the held-ticket share is usually the main cause of your tail — and that means the problem is in sequencing, not cooking speed.

4. Revenue per minute saved — what it's worth to fix

A shorter average ticket time means a shorter table occupancy, and a shorter table occupancy means more turns within the same service. Your service window in minutes, divided by table time plus the ticket time it depends on, gives the number of turns per table that night — exactly the same floor-division that decides how many times a table can realistically turn. Every minute you cut from the average ticket time can raise that turn count by exactly one on your busiest nights, on the tables sitting right at the edge. One extra turn on a handful of tables, at your average spend per cover, adds up faster over a full year of service than most owners expect — and that's the number that justifies investing in extra staff or a different pass system.

What a faster pass is worth on a busy night

How every minute shaved off the average ticket time converts into extra turns on the tables sitting right at the edge of an extra round — and therefore into extra revenue over a full year of service.

Minutes saved off the average ticket timeExtra revenue per year

From 4 minutes saved: +€58,656 extra revenue per year on the tables sitting right at the edge.

Which of the two problems do you have?

The four numbers above split into two families, and they call for two completely different fixes. If your held-ticket share (number 3) is high while your average (number 1) is genuinely fine, the problem is in how the pass synchronises tickets — fixed with tighter expo discipline, a different way of grouping tickets on the screen, or simply the agreement that a table gets served in two rounds when one dish needs an exception. If your held-ticket share is low but your average itself is high, the problem is raw capacity — too few hands on the busiest station, not enough prep before service, or a menu with too many à la minute dishes. A kitchen that fixes the wrong problem — adding a cook while tickets were already ready on time and simply sitting there — pays for a fix that never shortens the tail.

Calculate your own kitchen's tail and its value

Fill in the numbers your POS or kitchen display system already tracks. The calculator shows how big your tail gap is, whether your problem is sequencing or capacity, and what a shorter ticket time is worth per year.

Tail gap
The difference between your average and what your slowest 1 in 10 guests feel
Extra turns per busy night
Extra revenue per week
Extra revenue per year

Based on the tables you flagged as 'right at the edge' — the real gain depends on how many of your tables genuinely sit at that point.

What to do this week, this month and this quarter

  • This week: for one busy night, count how many tickets sat at the pass longer than it took to actually cook them. That's your held-ticket share — the number that decides which of the two fixes you need.
  • This month: if your held-ticket share is above 15%, change nothing about staffing. Change how expo groups tickets first — for example, by firing a dish that systematically waits two minutes later instead of letting it sit ready and waiting.
  • This quarter: measure again. A falling tail gap with an unchanged average confirms it was a sequencing problem. If the tail stays just as wide, the problem was capacity after all — and that's where your next investment should go.

A kitchen that is fast enough on paper can still feel slow — not because the average is lying, but because an average never tells you what the one guest in ten who defined the evening actually felt. Measure the tail, not just the average, and work out what every saved minute is genuinely worth before you invest in a fix that solves the wrong problem.

Frequently asked questions

What's a good average ticket time for a restaurant?

It depends heavily on the type of kitchen and the dish. A sandwich counter aims for a few minutes, a brasserie for 12 to 18 minutes on a main course, and a fine-dining kitchen finishing dishes à la minute can deliberately take 25 to 35 minutes because the guest expects that pace. There is no universally good number — only a number that matches what your menu and concept promise, and what a guest has already inferred about the evening's pace from your website or booking confirmation.

How do I measure the 90th percentile without a kitchen display system?

Most modern POS systems log the time a ticket fires and the time it's marked ready, even without a separate kitchen screen — ask your supplier for an export of those two timestamps. If you don't have that, a week of hand-tallying on a clipboard at the pass is enough: fire time, ready time, per ticket. A hundred tickets is enough for a reliable picture of your tail.

Why is the 90th percentile better than the maximum?

The absolute maximum is usually an exception — a ticket that got lost during a POS glitch, an order that had to be remade, or a table that itself asked for a delay. That number says nothing about a normal night. The 90th percentile only counts tickets that were slower than the rest under normal pressure, which is why it's the number you can actually manage down — the maximum will always be an outlier, however well your pass runs.

Can a faster kitchen display system fix the tail by itself?

A kitchen display system usually shortens the average, because less time is lost hunting for a ticket or asking about a status. But it doesn't automatically fix the held-ticket problem — that stays a matter of how expo groups and synchronises tickets, screen or no screen. A screen only helps against the tail once expo actually uses it to decide, deliberately, when a ticket genuinely needs to wait and when it doesn't.

Does every table need to be served at exactly the same time?

No, and that assumption is exactly what leaves many tickets waiting unnecessarily. A table that knows one guest ordered a dish that takes longer usually doesn't expect perfect synchronisation — a short heads-up ('your pasta will take a little longer, shall we start the rest?') costs nothing and takes a ticket out of the queue without the guest experiencing it as slow service.

How often should I re-measure these numbers?

Once a quarter is enough to see whether a change worked, unless you just changed something structural — a new menu, a new cook on the busiest station, or a switch to a different pass system. Measure the first two or three weeks after a change like that, exactly the way you keep a closer eye on a new staff schedule in its first weeks.