Most of Your Menu Is Reaching the Same Guest Twice

No operator ever sat down and added thirty items in an afternoon. The menu got long the way every menu gets long — one individually defensible decision at a time. A regular asked. A chef had something worth trying. A soft Tuesday needed a reason to post. Four years later there are ingredients in the walk-in that exist to serve two percent of covers, and nobody can point to the moment it happened.

What makes this hard to correct is that abundance feels like health. A group with capital, a deep bench and a big kitchen can carry ninety items, so it does. The constraint that would have forced a choice never arrives. What ends up on the table is not a concept anyone chose — it is an accumulation, and it is expensive in ways the P&L never labels.

Most operators know the menu is too long. Knowing it is too long changes nothing, because the moment you try to cut, the argument starts. Somebody defends every item, and the loudest defense usually wins.

Units sold is the wrong way to decide what to cut

The standard tool is menu-mix analysis — rank every item by units sold and contribution margin, then cut from the bottom. It is the right tool for pricing and the wrong tool for cutting, because it measures the item instead of the guest.

Two failures follow, and they run in opposite directions. The first is the redundant bestseller. An item can sell well and add nothing, because the guests ordering it would have been perfectly happy with two other things on the same menu. It is popular, and it is also replaceable. Cutting it costs you almost no guests.

The second is more dangerous — the veto item. Somewhere on your menu is a dish with terrible numbers that is the only acceptable option for one person in a party of six. The vegetarian entrée. The gluten-free handheld. The plain chicken a ten-year-old will actually eat. Menu mix sees a slow mover and marks it for death. What it cannot see is that the item is not selling six covers, it is holding the whole table. Cut it and you do not lose its units. You lose the reservation.

Reach is not the same as popularity

The tool that can tell them apart is TURF — total unduplicated reach and frequency. It comes out of media planning, where the question was never which ad ran the most, but how many distinct people a combination of placements reached, and how often. Two placements hitting the same audience are not twice the reach. They are one audience, paid for twice.

A menu works the same way. Every item reaches some set of guests, those sets overlap heavily, and the overlap is invisible in a sales-mix report. The question TURF answers is the one that actually governs a cut: what is the smallest set of items that still reaches nearly everyone who walks in?

The answer, on most long menus, is uncomfortable. A large minority of the menu adds no incremental reach at all. Those items are not serving a segment. They are serving guests who had three other acceptable options on the same page, which means the kitchen is carrying their full operating cost for nothing.

How to run TURF on your own menu

You do not need research software. You need ninety days of item-level ticket data and a spreadsheet. The one thing that matters is that you work at the ticket level, not the item level, because overlap only shows up when you can see what else was on the check.

• —Pull tickets, not sales mix. You need every item on every check, keyed to the check. A sales-mix summary has already thrown away the information you need.
• —Define the lanes. Group items the way a guest chooses, not the way the kitchen files them — handhelds, shareables, vegetarian, kids, gluten-free. A guest picks within a lane.
• —Count reach per item: the number of tickets it appears on. This is the number menu mix already gives you.
• —Count unduplicated reach: only the tickets where that item was the sole item from its lane. Those are the checks that item actually held.
• —Rank by unduplicated reach, never by units. A bestseller with almost no sole-item tickets is redundant. A slow mover with a high share of them is a veto item and it stays, whatever its margin looks like.
• —Cut from the bottom of the unduplicated list, then hold the shorter menu for one full period before you judge it. The first two weeks feel like loss and the numbers have not moved yet.

What the tail costs once you have found it

Here is the math on an illustrative three-unit group. Run it with your own numbers, not these.

Say the group does $4M across three units at 30% food cost — roughly $1.2M in annual purchases — carrying 78 items. Menu mix says the bottom 24 items are 6% of covers, and the reflex is to cut all 24.

TURF sorts that 24 into two very different piles. Nineteen are redundant, ordered almost entirely on tickets that carried two or three acceptable alternatives. Five are veto items, appearing as the sole choice in their lane on a meaningful share of the checks they touch. Cutting all 24 saves the most money and quietly costs you tables. Cutting the 19 costs roughly half a point of reach.

Now cost the 19. Those items depend on 28 ingredients that appear nowhere else — call it $70,000 of annual purchasing. Fast ingredients turn and waste at maybe 2%; slow ones sit, and 9% is generous. Seven points of excess waste on $70,000 is $4,900 a year. Then the labor nobody books to the menu: each item still needs its own prep, its own place on the line, its own training for every new cook. Three prep hours a week per unit at $18 fully loaded is 3 × 52 × 3 × $18, or $8,424 a year.

That is $13,324 a year to carry items that reach nobody who was not already reached. And it is the conservative version — it counts none of the ticket time, none of the comps on a dish the newest cook has made twice, and none of the manager attention spent holding a standard across 78 items instead of 59.

The other half of TURF is frequency

TURF has a second letter, and it is the one that keeps this from becoming an argument for permanent narrowness. Reach tells you how many distinct guests you can serve. Frequency tells you how often the same guest can come back before the menu runs out of reasons. They are different problems with different fixes, and cutting is only the answer to the first one.

A menu can reach almost everyone in the market and still cap the business, because the regular who has worked through everything worth ordering has no fourth visit in them. That is not a reach failure and no amount of subtraction fixes it. Knowing when to widen deliberately — and how to do it without becoming a concept your regulars no longer recognize — is a separate decision with separate evidence behind it.

The sequence matters, though. Cut for reach first. A menu carrying nineteen items that hold nothing is not a menu with a variety problem. It is a menu with a decision problem, and adding to it makes both worse.

The bottom line

Sales mix measures items. TURF measures guests, and guests are what you are actually cutting.

Pull ninety days of tickets, sort every item by the checks where it was the only acceptable option in its lane, and the argument about what to cut mostly settles itself. The redundant bestseller loses its defense. The slow mover that holds a six-top gets the protection it always deserved. And the items in between — the ones nobody would order if they were not already printed — stop costing you a third of a point to reach nobody at all.