Coffee shop menu engineering: sizes, milk options and rush-hour capacity

Build one row per sold drink combination

A large oat latte is not a small dairy latte plus free complexity. It has a selling price, ingredient quantities and packaging cost of its own. Start with a defined sales period and keep refunded or voided transactions out of sold-unit totals. Decide whether your report shows net sales after discounts and apply that basis consistently.

The coffee-shop costing guide helps establish the drink recipe. This review uses those costs to decide what to promote, reprice or investigate across the menu. The general menu-engineering guide explains the popularity-and-contribution matrix. Here, the important work comes before the matrix: reconstructing variants without counting one order twice.

For example, a report may show one latte sale and one oat-milk modifier. That is one drink, not two. Join the modifier to its parent ticket item, retain its charged amount, and attach the matching incremental cost. If modifier records cannot be linked reliably, mark the resulting mix incomplete rather than guessing which size received them. Marji does not import POS sales; use your own exported or manually recorded counts.

Sales total $970 and the listed costs total $370, leaving $600. Multiplying all 200 drinks by the small dairy latte’s $3.20 contribution would report $640. That shortcut hides $40 because the variant mix is different. The large oat drink contributes less in this example even though it has the highest selling price; those assumed costs are not a claim about oat drinks generally.

Review the modifier as its own decision

In the sample, moving from large dairy to large oat adds $0.50 of revenue and $1.50 of cost, reducing contribution by $1 per converted order. A $1.50 surcharge would preserve the large dairy drink’s contribution if quantities, purchasing costs and demand stayed unchanged. It would not prove customers will accept that price or that it is the right decision for this café.

First inspect the recipe and invoice basis: pack volume, usable amount, waste allowance and portion used. A unit-conversion error can make a modifier appear unprofitable. If the costs are valid, compare several operator choices: a price change, a revised tested recipe, a different supplier specification, or deliberately retaining the option for customer choice. Show the trade-off instead of deleting a popular option automatically.

Check the station that limits the rush

EHL’s menu-engineering discussion cautions that item contribution alone misses preparation time and operational constraints. A café can apply that warning by measuring the station that actually holds up orders. Total recipe elapsed time is not always the same as occupied barista or machine time, especially when tasks overlap.

Across ten fully occupied station minutes, ten standard drinks would contribute $30 and five specialty drinks $20 under these assumptions. This is a capacity comparison, not a sales forecast. It requires enough demand for the alternative, interchangeable use of that station, and no other bottleneck. During a quiet period with spare capacity, rejecting a $4 contribution order because of its slower rate could make the result worse.

Measure service times over several comparable orders and record batching, equipment and staffing conditions. Do not deduct the entire hourly wage from every minute of drink preparation and then deduct that wage again from shop contribution. If an option requires an additional paid shift, show that added payroll separately in the decision.

Keep pastry waste outside the drink count

Suppose a café buys 50 pastries at $2 each, sells 40 at $4, and discards the remaining ten with no salvage. Revenue is $160 and the batch costs $100, leaving $60 before other costs. Counting only the 40 sold pastries at their $2 purchase cost would imply $80 and hide $20 of waste. Record unsold output beside sales rather than quietly changing the pastry’s recipe cost.

Compare pastry attachment using receipts, not pastry units divided by a drink total that includes duplicate modifier rows. A guest may buy two drinks and one pastry; specify whether the metric is per receipt, per guest or per drink. Use the combo-pricing guide before offering a pastry discount to guests who already buy both.

Turn the review into one measurable change

Select one uncertain variant, verify its inputs and write down the decision you want to test. Compare contribution dollars, units, modifier mix, waste and service capacity over comparable periods. Record a holiday, weather change or promotion that could explain a difference. A better result after repricing is observational evidence, not proof that price alone caused it.

The existing worksheet can compare your cleaned rows, but it cannot reconstruct missing ticket relationships or measure your station times. Keep those records with the analysis. The useful outcome is a defensible change to one menu decision, followed by a check of what guests actually ordered.