Menu engineering without the jargon
Popularity against contribution margin gives you four groups of dishes and four different decisions. Here is what to actually do with each.
Plot every dish on two axes: how often it sells, and how many dollars it contributes after food cost. Percentage margin is the wrong axis — a 90% margin on something nobody orders pays no rent. Contribution in dollars, multiplied by covers, is what lands in the bank.
The four groups
| Group | Sells | Contributes | The move |
|---|---|---|---|
| Winners | High | High | Protect. Don't reprice, don't reformulate, don't hide it on the menu. |
| Crowd-pleasers | High | Low | Fix the cost, not the price. Portion, spec, and prep waste first. |
| Hidden gems | Low | High | A merchandising problem. Move it, describe it better, train the servers. |
| Dead weight | Low | Low | Cut it, unless it is holding inventory for something profitable. |
Price changes have an elasticity assumption in them
Raising a dish $1 does not add $1 × current covers. Some people order something else. Model it explicitly: at what drop in covers does the price rise stop paying? For most casual dishes a 5% price rise survives a 5–10% cover loss and still comes out ahead — but you should see that number before you print the menu, not after.
Run the reverse test too
Ask what happens to each dish if one ingredient goes up 10%. The dishes that fall apart under that test are the ones to re-spec now, while there is no pressure.