Four boxes. High popularity and high margin, high popularity and low margin, low popularity and high margin, low popularity and low margin. Stars, plowhorses, puzzles, dogs, or whatever the version you were handed calls them. Promote the first, reprice the second, reposition the third, kill the fourth.
It is the most widely taught analytical instrument in the restaurant business, and it is the clearest example in the industry of a framework with blanks in it doing the operator’s deciding for him.
What The Matrix Actually Is #
You supply the items. The grid supplies everything that matters.
It decides that two variables are the relevant ones. It decides where the dividing lines fall. It decides that popularity and margin are independent inputs rather than things that cause each other. It decides that four categories exhaust the possibilities, and it decides what the correct action is for each box before it has seen a single thing about your building.
You typed. It ruled.
The inference: you filled in your own numbers; therefore the output is your analysis.
That is the sibling move to a template, and it is harder to argue with, because the operator now believes the answer is his. He did the typing. What he did not do is decide the shape, and the shape is where the ruling lives.
What The Two Variables Cannot See #
Margin percentage instead of margin dollars. Most versions of the grid run on contribution margin, which is closer to right than food cost percentage, but the dividing line is still an average of your own menu. An item lands on the wrong side of the line because of what sits next to it, not because of what it produces.
Why the Guest is in the building. The item that brings a Guest through the door does not have to be the item that makes the money, and killing it because it scored badly removes the reason for the visit that funded everything else on the check.
What the item does to the rest of the check. An item that pulls a second round, a shared starter, or a dessert is carrying revenue the grid credits to other rows.
What the item does to the kitchen. Two items with identical margin can cost completely different amounts of stage time, station space, skill, and holding capacity. One of them is a bottleneck at seven-forty on a Friday and the grid cannot tell them apart.
Shared product. Kill the low performer and the yield on three other items changes, because that product was being used across them. Grids treat rows as independent. Kitchens do not work that way.
What you are chosen for. Some items are the operation’s position. They can score badly on both axes and still be the single most valuable thing on the menu, because they are the reason a Guest picks you instead of the place down the street.
How the item was priced and where it sits. Popularity is not a property of the item. It is a result of price, placement, description, and what your cast says about it. The grid reads the outcome of your own past decisions and hands it back to you as data about the food.
Daypart and mix. One blended grid across three dayparts describes an operation that does not exist at any hour of the day.
The Failure Mode It Produces #
Run the grid faithfully for a few cycles and the menu converges. Low scorers come off, high scorers get promoted, prices move toward what the matrix rewards. Every operator running the same instrument on the same category of menu ends up in the same place.
The items that get killed first are the ones with narrow appeal and strong loyalty, because narrow appeal reads as low popularity. Those are frequently the items a specific cohort of Guests comes in for. The grid cannot see a cohort. It sees a row that did not sell enough.
So the instrument systematically removes the reasons particular people choose you, in the name of optimization, and the operator watches his numbers improve on paper while the operation gets more ordinary every cycle.
What Replaces It #
Margin dollars per item, ranked. Not percentage, not a box. Dollars. Then look at the ranking against what you thought you knew, because it is usually different.
Margin dollars per hour of stage time. Your room and your hours are the finite thing. What an item produces per hour it occupies the operation is a truer read than anything the grid outputs.
Kitchen cost per item, named. Station time, skill required, holding behavior, what it does to throughput at your peak. Some of your best-margin items are the reason tickets stack.
Check-level reading, not item-level. Pull the checks that contain an item and read what else is on them. That tells you what the item is actually contributing.
Cohort reading. Which Guests order it, how often they come, and what happens if it is gone. Ask your cast — they can tell you which items have people attached to them.
The position question, asked out loud. For every item, what is this doing here. Bringing people in, making the money, holding the standard, or nothing. An item can earn its place on any of the first three. Nothing is a real answer and it means the item comes off.
Your own ruling on what the menu is for. Before any of the arithmetic. The menu is an expression of what your operation is chosen for, and that is a decision only you can make. Every number afterward serves the ruling instead of replacing it.
What Changes Tomorrow #
Take your top twenty items by volume and build one column: margin dollars per item. Rank them. Look at where your assumptions were wrong, and they will be.
Then add a second column that nobody’s grid contains — minutes of your busiest station’s time per item at your peak hour. Sort by margin dollars divided by those minutes.
That ranking is the one that matters, because it reads money against the only thing you genuinely cannot make more of on a Friday night. Items will move. Something you have been promoting because it scored well will turn out to be the reason your kitchen backs up at seven-forty, and something you were about to kill will turn out to be the cleanest money in the building.
Then take the three lowest and ask the position question on each before you touch them. If any one of them is why a specific group of Guests picks you, it stays, and it stays on purpose with the reason written down.
Digging Deeper #
Every term in my framework lives in the Knowledge Base: kb.jeffreysummers.com
Terms used here: [Measurement Lock-In], [Cover Blindness], [Ordinary By Design], [Meaningfully Differentiated Value], [Differentiation Economics], [Sameness Machine], [Default Architecture], [Designed Architecture], [Configuration Arbitrage], [Safest Mediocre Execution], [Product Constraint], [Constraint Architecture], [Transactional Cost-Plus], [Occasion]
More of my work on this, all of it in the Knowledge Base:
- 5.X What Your Menu Is Actually Telling You — the read that replaces the grid
- 5.X The Price Is Not The Formula — why the cost-percentage axis is the wrong axis
- 2395 The Focused Menu — the actual menu decision
- 2465 The Menu Development Cycle — how a menu is supposed to move
- 2420 The Menu That Makes The Guest Work — what the grid never measures
- 5.X The Lost Opportunity Tax — the cost of the item that stays because it scored
