Four kinds of dish, and what to do about the dogs

There is a dish on your menu that everybody loves and nobody has ever costed properly. It has been there since you opened. Regulars ask for it by name. The chef would take it personally if you mentioned it, and it is losing you money on every plate that leaves the pass.

Usually it is the Sunday roast. Sometimes it is a burger with three garnishes that were added one at a time over four years, none of which went onto the price.

Two axes, and one of them is usually the wrong unit

Menu engineering plots every dish on two measures. How often it sells, and what it contributes. Draw a line through the middle of each and you get four boxes, which the trade has been calling stars, workhorses, puzzles and dogs since the early eighties.

The measure people get wrong is the second one. Contribution has to be in rand per plate, not gross profit percentage, and the difference is not academic.

The four boxes

Stars: sells well, earns well

Leave the recipe alone. Feature them where the eye lands, train the floor to recommend them, and watch their ingredient costs more closely than anything else on the menu, because a star that drifts is the most expensive drift you have. This is the box where complacency costs the most, precisely because nothing looks wrong.

Workhorses: sells well, earns badly

Your volume, and the reason people walk in. You cannot pull them and you probably cannot put ten percent on the price without a conversation on Facebook about it.

So take the cost out where the guest does not meet it. Change the cut, not the dish. Change the supplier, not the recipe. Look at the garnish that goes back on ninety percent of plates. A workhorse improved by eight rand a plate at three hundred plates a month is R2 400 that nobody notices leaving, or arriving.

Puzzles: earns well, barely sells

The most interesting box, because the food is usually fine and the problem is elsewhere. It is fourth in a list of nine. It is described in a way that means nothing to somebody who has not eaten it. It is next to something cheaper that does a similar job. Servers do not mention it because they have not tasted it.

Move it, rename it, describe it better, put it on as a special for a fortnight and see what happens. A puzzle that becomes a star is the single best return available on a menu, and it costs you nothing but attention.

Dogs: sells badly, earns badly

Cut them. Most of them, most of the time. A menu with nine dogs on it is not offering choice, it is offering hesitation, and it is carrying stock for dishes nobody orders.

Before you do, three checks. Is it a dog or is it new, with fourteen covers behind it and no verdict yet? Is it there for a reason that is not profit, such as the one vegetarian main or the kids' plate that decides where a family of five eats? And is anything else on the menu leaning on its ingredients, because dropping a dish that was carrying half the yield of a whole fish will make its neighbours worse.

A dog with a good reason to exist is not a dog. It is a cost of doing business, priced accordingly.

Where the exercise usually goes wrong

Not in the theory. The theory fits on a napkin. It goes wrong in the inputs, and always the same three ways.

Stale costs. A menu ranked on recipe costs from January is a ranking of January, and if your protein has moved eleven percent since then the ranking is fiction. This is the most common failure by a long way, because recosting a hundred and forty recipes by hand is a fortnight of somebody's life and nobody has a fortnight.

Missing sub-recipes. If the mother sauce, the pickle and the house dressing are costed at nothing, every dish that uses them looks better than it is, and the dishes that use the most of them look best of all. Which means your ranking is not a profitability ranking, it is a ranking of how much uncosted labour a dish hides.

Doing it once. A menu review in March that nobody repeats until the following March is a photograph of a moving thing. Prices moved. So did what people ordered.

Doing it without losing a fortnight

This is the part Dyner does. Because invoices are read and matched as they arrive, every ingredient carries a current price, and every recipe and sub-recipe built on it recosts when that price moves. The ranking is therefore never older than the last delivery.

The menu profitability screen puts each item next to the quantity sold, the rand value of that quantity, the average selling price and the average recipe cost per unit sold, and the sales value less recipe cost. You can group by item, by category, or by super category, which is how you find out that it is not one dish, it is the whole grill section.

There is a sort that pulls the biggest margin leaks to the top, which is a blunt instrument and the right one to reach for first. Ten minutes with that list on a Tuesday morning will tell you more about next month's profit than a fortnight of anything else.

What to actually do on Monday

1. Rank last month's dishes by units sold and by rand contribution per plate. Two columns, nothing clever.
2. Take the bottom eight by contribution and check whether the recipe on file is the dish the kitchen is sending out. It usually is not.
3. Pick one puzzle. Move it up its section and get the floor tasting it this week.
4. Pick one workhorse and find eight rand in it that the guest cannot see.
5. Cut one dog. Only one, and tell the chef why using the number rather than your opinion.

Do that every month and the menu drifts towards the things that pay. Do it once a year and you will spend the other eleven months guessing, in a business where a point of margin is a wage.