There is no correct number of dishes. There is a number at which your guest still chooses, your kitchen still delivers and your last dish still sells — and for your restaurant that number is not the one next door's.
Almost every menu grows. A dish goes on because a regular asked for it, a vegetarian alternative appears, a classic stays because taking it off feels like a waste. Nobody ever adds five dishes at once — and yet after three years there are twice as many.
On the question of how many there should be, the internet repeats one thing: fewer is better, because of choice overload. That is half right. The research everybody points at says something far more useful than "fewer is better", and anyone who reads it in full reaches a different conclusion about their own menu.
This guide walks through seven rules with the figures and the sources attached, and carries three interactive pieces: a slider that grows and shrinks your menu live, a board where you switch dishes on and off to see what they cost you in purchasing, and a calculator at the bottom that runs your own menu.
Everything runs in your own browser: nothing is sent anywhere and nothing is stored. The bands below are guide values for independently run European venues — your own till is the final judge, and the calculator exists precisely to check it.
Why the number counts on both sides of the pass
On the dining-room side, the number of dishes is a decision you impose on your guest. On the kitchen side it is stock, mise-en-place and a shelf-life clock. Almost every article on this subject looks only at the first side, and that is precisely the side where the effect is smallest and most contested.
The kitchen side is not contested at all. Every dish you add brings purchase lines, prep before service, space in the fridge and a chance that something spoils. Those are not psychological effects but arithmetic ones: they sit in your order sheet and on your scales.
And there is a third side nobody measures: the number of dishes decides how thinly your sales are spread. The same one hundred and forty-four portions divided over 24 dishes look nothing like the same portions divided over 48. Past a certain point the bottom of your menu sells less than once a service — and that dish still costs you prep, purchasing and fridge space.
Same venue: five categories, 90 covers a service, 1.6 dishes per guest. Drag the slider and watch what moves with it — the guest on the left, the kitchen on the right.
Per category
9.6
Sells < 1× a service
9
Unique ingredients
113
Prep hours a service
8.0
The four chips are the four conditions from the meta-analysis by Chernev and colleagues (2015). Notice what happens: well into the forties only one of them lights up — because this menu is grouped and described. Choice overload is not a question of how many, but of how many without structure. The kitchen columns, by contrast, climb from the very first dish.
The 7 rules, and where they come from
The first three are about your guest, the middle three about your kitchen and your till, and the last one puts them together into a single number that is yours.
1. Don't count your menu, count your categories
A forty-dish menu in eight categories is a completely different object from forty dishes in two. Your guest never reads forty things: they pick a category first and read a handful of lines inside it. So the number that counts is dishes per category, not the total.
The working band the trade uses is five to nine per category. It lines up with the classic idea that a person can hold about seven things at once, and it is exactly wide enough to give everybody at the table something without forcing anyone to compare.
In practice: write your menu out in columns and count each column. If every one sits between five and nine, then your total — whatever it is — is probably not your problem. If one column is at fifteen, you do not have a menu problem but a layout problem, and you fix that with a subheading, not with scissors.
2. What the research actually says about choice overload
This is where nearly every article on the subject goes wrong. "Too much choice paralyses" is presented as settled fact, invariably with one study attached. But when Benjamin Scheibehenne and colleagues pooled all the available studies in 2010, almost nothing of that effect survived on average.
The meta-analysis that followed — Alexander Chernev, Ulf Böckenholt and Joseph Goodman, Journal of Consumer Psychology 2015, 99 observations across more than 7,200 participants — explained why. Choice overload is not a law that always holds; it appears once four conditions coincide: the assortment is unstructured, the decision is hard to make, the chooser is unsure what they want, and they want to put as little effort in as possible.
Read those four again and look at your menu. All four are design choices. Group your dishes under clear headings and the first falls away. Write one sharp line under each dish and the second falls away. Have your floor recommend one dish per category and the third falls away. The number of dishes is not the problem — the number of dishes that say nothing about themselves is the problem.
3. The menu that attracts is not the menu that sells
The study everybody quotes is Sheena Iyengar and Mark Lepper's (2000): in a delicatessen, a tasting table carried either six or twenty-four kinds of jam. What usually gets left out is that the big display won on one measure — more people stopped at it, sixty per cent against forty.
And then the figure that does get quoted: of those who tasted at the small display, thirty per cent bought something. At the big display it was three per cent. Ten times fewer sales, with more lookers.
Translated to a menu that is not an argument to cut everything, but it is an argument to know what your menu has to do. A broad menu in the window or on your website brings people in. That same breadth at the table, at the moment a decision is actually required, works against you. It is exactly why an extensive à la carte menu with a short prix fixe beside it so often outperforms either one alone.
4. Your kitchen doesn't count dishes, it counts ingredients
Two forty-dish menus can differ threefold in what they cost your kitchen, and the difference is not in the dishes but in the overlap. A dish built from things you already carry costs you almost nothing extra. A dish with six ingredients of its own costs you six purchase lines, six places in the fridge and six clocks ticking.
Switch dishes on and off below and watch the unique-ingredient count move. The first four dishes share almost everything; the last two share almost nothing.
The practical rule that follows is not "fewer dishes" but "fewer loose ingredients". Before you cut a dish, check whether you can rebuild it out of what is already on the shelf — that lowers your purchasing without anything leaving the menu. The recipe costing tool shows, per dish, which ingredients you use nowhere else.
Six dishes with six ingredients each. The first four come out of the same mise-en-place; the last two bring their own shopping list.
Four dishes out of the same kitchen share nine ingredients across twenty-four lines. Add the last two and you go from nine to twenty-one — two dishes that grow your purchasing by more than half.
5. The tail: the dishes that sell less than once a service
Menu sales are never evenly spread. A handful of dishes carries the bulk, a middle group ticks along, and at the bottom sits a tail of dishes that are there but barely move. In the venue used throughout this article — 48 dishes, 144 portions a service — nine of them sell less than once a service.
Such a dish is not free. It costs prep before every service, it holds its own ingredients in the fridge, and it occupies a line on a menu your guest has to read through anyway. And it is exactly the kind of dish nobody ever decides about, because its revenue is too small to notice and too large to ignore.
The figure you need for this is in your till and nowhere else: pull a sales report per dish over three months and divide by your number of services. Everything under one is your tail. That is also the only genuinely reliable input the calculator below takes — the rest are estimates you can adjust.
6. Cutting doesn't delete revenue — but it isn't free either
On the question of whether cutting pays, two mirror-image mistakes get made. The first: assuming a cut dish's revenue disappears. That is almost never true — somebody whose favourite is no longer listed orders something else. The second: assuming all of it moves across. That is not true either; some of your guests came for that one dish.
A realistic assumption sits around seventy per cent: of every ten portions you cut, seven come back on another dish and three do not. That percentage is an input field in the calculator, because for a neighbourhood restaurant with regulars it is lower than for a tourist one.
Only then can you do the sum. In the example venue, cutting the nine tail dishes frees about an hour and a half of prep a service and costs the margin on the portions that do not move across. Net, a little over three thousand euro a year is left — no fortune, but free money, and it is the rest of the mise-en-place that really gains. Nudge the assumptions and the same sum can come out negative, and then the right answer is: leave them on.
7. The right number follows from your format, not from a rule of thumb
A bistro with forty covers and a starred restaurant with forty covers should not carry the same number of dishes, and a pizzeria should do something different from both. The guide values below are not law but a starting point: they differ because the format differs, not because one venue is better than another.
What they all share is the underlying test. Pick the number at which your last dish still sells at least once a service. If you sell four hundred portions a service you can carry a far wider menu than someone selling eighty — the same menu is roomy for one and a thirty-dish tail for the other.
And revisit that number with the seasons rather than with the years. A seasonal menu that changes four times a year can be narrower each time, because the variety sits in time rather than on the paper.
| Format | Categories | Per category | Total | Why |
|---|---|---|---|---|
| Bistro / brasserie | 4 – 6 | 6 – 9 | 28 – 45 | Wide enough for a mixed table, narrow enough to keep everything fresh daily. |
| Fine dining à la carte | 3 – 4 | 4 – 6 | 14 – 22 | Every preparation is labour-intensive; here breadth comes out of execution. |
| Tasting menu | 1 – 2 | 1 – 3 | 2 – 5 | You sell a <a href="/en/blog/menu-and-drinks/tasting-menu-fine-dining-strategy.html">journey</a>, not a choice; the menu is an announcement. |
| Pizzeria / trattoria | 3 – 5 | 8 – 14 | 30 – 60 | High overlap: twenty pizzas often share five more ingredients than the first ten. |
| Café / lunch venue | 4 – 6 | 4 – 8 | 18 – 35 | A fast decision at the counter; anyone who has to read too long orders coffee only. |
Run your own menu
Fill in what is on your menu, what you turn per service and what a dish costs you in prep. The fields carry the brasserie from this article — 48 dishes across five categories, 90 covers — so you can see straight away how it reads.
One field matters more than the rest: the spread of your sales. Pick "flat" if almost everything sells about as well, "normal" if there is a clear top, and "sharp" if a handful of dishes carries the whole menu. The calculator shows what share your top fifth then takes — compare that with your till report and pick the setting that is true.
Menu-size scan
Your own menu, your own covers — and the number at which your last dish still sells.
Your menu
Your kitchen
Money
Fits your venue
—
Tail dishes
—
Top fifth
—
Net of cutting
—
—
The model spreads your portions across your dishes according to the chosen distribution, normalised — so the total never changes when you adjust the setting, only the shape of the curve. Unique ingredients grow by (100% − reuse) per extra dish. When cutting, the set percentage of portions moves to another dish; the rest is lost. The amount is a net and may be negative. Everything runs in your browser; nothing is sent or stored.
Two things to know when you read it. The amount is never the revenue of your tail — it is the prep time saved minus the margin on the portions that genuinely fall away when you cut. That is why it can be negative, and why "cut" here is an outcome rather than advice.
And the first tile is the only number in this whole article that is entirely yours. Rules of thumb say 25 to 45; your covers and your sales spread say something more precise. If that lands far from the rule of thumb, believe your own figures — but check first that your top fifth looks in your till the way it looks here.
How to right-size your menu in three steps
Shortening a menu on instinct almost always ends in an argument with the chef. This order measures first, decides second and only cuts at the end.
This week — pull the report
- Run a sales report per dish over the last three months in your till and divide each count by your number of services.
- Mark every dish that stays under one sale a service. That is your tail; count them and count nothing else.
- Write your menu out in columns and count each category. Anything above nine gets a subheading rather than scissors.
- Put your own figures into the scan above and note what comes out. Change nothing on the menu yet.
This month — fix what is free first
- Give every dish without a description a one-line one; that switches off the second condition for choice overload without cutting a single dish.
- Group everything under clear headings and put no more than nine lines under one heading.
- Run your tail through the recipe costing tool: which of those dishes use ingredients you deploy nowhere else?
- See whether two tail dishes can be rebuilt from your existing mise-en-place. That removes the purchasing cost and leaves the choice standing.
This quarter — only now cut and reprice
- Cut the tail dishes that both sell little and carry ingredients of their own. Leave the rest.
- Measure again after a month: did the revenue move across to the surviving dishes, or did it genuinely fall away?
- Turn the freed prep time into something measurable — a better mise-en-place or one more service-ready dish from the existing menu.
- Revisit the prices on the surviving menu with menu engineering and the menu price calculator: a shorter menu carries sharper pricing.
The number that counts is not in this article
Anyone looking for one correct number of dishes finds a rule of thumb online and leaves it at that. Five to nine per category is a perfectly good starting point and it is right more often than not — but it is an average over thousands of venues, none of which looks like yours.
The usable answer is a test rather than a number: does your last dish still sell at least once a service? If it does, your menu is not too long, however long it is. If it does not, you know exactly which dishes deserve the conversation — and thanks to the previous rule you also know that cutting is not always the answer.
Then move on to what your menu earns rather than what it costs. Menu engineering tells you which dishes to push, price psychology how to present them, menu design where the eye lands, and food cost what is left at the bottom. This page only answers the question that comes before all of them.