Menu Engineering for a Small Café — Finding What Actually Earns

Every line on your menu occupies shelf space — a line a guest reads, a slot in the display case, a crate in the fridge. Menu engineering is the habit of asking two blunt questions of each line: how often does it actually sell, and how much does each sale leave behind once the ingredients are paid for? Neither answer requires a consultant. Both come out of numbers you already have, and the whole exercise fits in one honest evening with a spreadsheet.

The Two Numbers Each Item Needs

Popularity is the simpler one: how many units left the till in the period you are examining. Count sit-in and takeaway versions of the same drink separately, because their costs differ and their audiences differ. Take a window that includes your quiet days as well as your best Saturday, so one payday or one festival does not distort the picture.

Contribution is menu price minus ingredient cost. Not the wholesale price of the whole dish — just what went into this serving: the milk, the beans, the syrup, the slice of cake, the takeaway cup. Labour and rent are deliberately left out of the per-item figure, because in a small café they arrive in broadly the same lump whether the order was an espresso or a club sandwich. Contribution is the number the item itself controls.

Here is a six-item example menu for one trading month. The costs are invented for the arithmetic and clearly labelled as such — yours come off your own supplier invoices.

Drawing the Two Lines

With the table in front of you, compute two averages. Total units sold across the six items is 2,705, so the average item sells about 451 units a month — that is the popularity line. Add the contributions per sale and divide by six: $14.35 divided by six is $2.39 — that is the profitability line. An item above both lines behaves differently from one below them, and that is the whole of the method.

The Four Quadrants

What to Do With Each Quadrant

• Stars — guard them. The latte is $2,695 of the month's contribution. Keep the beans it uses in stock even when the supplier is late, and never let a new trainee learn on the drink that pays the rent.
• Plowhorses — nudge, do not jump. Twenty cents on the croissant takes its contribution from $1.80 to $2.00 and adds $128 a month at 640 sales. If the count holds for a fortnight, it holds for the year — that is $1,536 of extra contribution from one line.
• Puzzles — market them before you kill them. The walnut cake has the best contribution per slice on the menu and still sold 95. Move it into the counter display with a photo on the menu and describe it properly; getting it from 95 to 150 slices adds 55 times $2.80, another $154 a month.
• Dogs — be honest about why they exist. The iced chai sold 60 times for $138 of contribution and ties up a bottle of syrup. That is a retirement candidate. The espresso is quieter than you expected, but it is also the base of your coffee credibility and the drink people buy while their friend buys the latte. Decide deliberately, not by spreadsheet alone.

Where the Counts Come From — and Where They Stop

The popularity half of the table is mechanical, and a till should hand it to you. In Kafe360 the items sold report lists best sellers by quantity and by revenue, names the ten slowest movers, and splits revenue by category, over whatever date range you set — which is precisely the popularity column of this exercise, without any typing. The period comparison page puts two months side by side with the percentage change, so you can see whether last month's Puzzle is becoming this month's Star.

The Mistakes That Skew the Quadrants

• Costing the menu once, in the year you opened, and never again. Contribution decays quietly as supplier prices drift.
• Analysing a month with a promotion in it. A half-price cake week makes the cake look popular and unprofitable at the same time, and both readings are wrong.
• Deleting a Dog that carries a ritual. The plain espresso that regulars order while deciding is doing quiet work the count cannot see.
• Counting the takeaway latte inside the sit-in latte. The cup, the lid and often the bigger size belong to a different cost base — recipe-based stock control treats them as separate items for exactly this reason.
• Judging a dish in its first fortnight. New items need a menu cycle of familiarity before the count means anything.

A Quarterly Hour, Not a Project

• Pull item counts for a full trading month, takeaway and sit-in counted separately.
• Refresh ingredient costs from the latest supplier invoices — the three or four that moved, not all of them.
• Compute the two averages, sort the ten items into quadrants, and act on exactly two of them this month.
• Write the action and the date next to each item you touched, so next quarter you can see whether the croissant rise held.

Done that way, menu engineering is not a rebrand of the menu — it is a recurring, small correction in the direction of the items that pay for the room. The till counts the popularity; your invoice folder prices the ingredients; the two averages do the sorting.