How to cost and price a cup of coffee properly
Short answer: build the cost of each drink from your own figures, line by line — coffee including purge and dial-in waste, milk including what goes down the drain, consumables, the seconds of staffed time at a loaded rate, energy, water and treatment, a share of the equipment, and remakes. Then price from that, not from what the shop next door charges.
Every figure this exercise needs is in your own records, and none of the ones you might borrow would be right. What follows is the method: which lines belong in the model, how to get each one from your own operation, and what to do with the result.
Why does pricing from the competition go wrong?
Because a price is only meaningful in relation to a cost, and the cost belongs to you.
The usual approach is to check what nearby venues charge and position yourself inside that range. What that does is adopt another business's economics as the basis for yours — their rent, their lease terms, their wage bill, their milk contract, their volume, their equipment, their waste.
Any one of those can differ enough to move your cost per drink materially. If several differ, the price that works comfortably for them may not cover your costs at all: a drink that loses money at every volume. Selling more of it makes the position worse, and because the shop looks busy the diagnosis usually lands on marketing rather than arithmetic.
The competition still constrains what customers will accept, so it belongs at the end of the process as a reality check rather than at the start as the method. Cost it, decide what to charge, then test whether that price is defensible in your market. If it is not, the problem is your cost structure or your positioning, and repricing alone solves neither — which is one of the recurring ways coffee shops fail.
What goes into the cost of one drink?
Work through these lines for each of your main drinks. Do the best sellers first; they tell you most of what you need to know.
Coffee
Start with the dose. Weigh it — the actual dose your grinder delivers into the actual basket you use, not the number written on the recipe card, and check both single and double where you serve both.
Then add the coffee you consume without selling, which is the part almost everyone omits:
- Purge shots. Whatever gets pulled and discarded after a grind adjustment, at the start of service, or between beans.
- Dialling in. Every morning, and again whenever the grind moves, a new bag is opened, or the weather changes. On a busy specialty bar this is not a trivial quantity.
- Grinder retention and spillage. Coffee left in the chute, on the counter, in the knock box.
- Discarded shots. Anything pulled and poured away because the timing was wrong.
To capture this without guessing: over a week, record the weight of coffee used and count the drinks sold. Multiply drinks by your target dose, compare with actual consumption, and the difference is your real waste. Divide it back across drinks sold for a per-drink waste figure from your own bar. Repeat quarterly, because it moves.
Milk
For a milk-heavy venue this is normally the largest recurring consumable, and it is also the least accurately known.
Per drink, measure the actual volume poured into the cup rather than assuming it, because the volume in a jug and the volume in a cup are different quantities.
Then the waste. Jug texturing makes milk waste structurally invisible: milk is poured by estimate, the excess is tipped away, and nothing counts it. To measure it, collect what would otherwise go down the drain for a week, weigh it at the end of each service, and divide by the count of milk drinks sold. Most operators find the result larger than expected. Milk waste in a busy café covers the causes and what reduces them.
Where the measurement shows waste is a serious line rather than a rounding error, the structural answer is to dose milk per drink rather than per jug, which removes the estimate that creates the waste. Your own measured figure is what tells you whether that is worth acting on, and it is worth having either way.
Cost plant-based milks separately. Their cost per litre differs from dairy, and averaging the two hides a real difference between menu items.
Consumables
Individually negligible, collectively not, particularly on takeaway-weighted sales.
Count everything that leaves with the drink or is used to make it: cup, lid, sleeve, carrier, stirrer, napkin, sugar, syrup, chocolate, a paper filter if you use one. Consumption is not the same as specification — sugar and napkins get taken in quantities nobody controls, so derive per-drink usage from purchase records over a period rather than from what a drink theoretically needs.
Where you serve both takeaway and in-house, cost them as separate items. The difference sits entirely in this line, and it can be the difference between a good item and a marginal one.
Direct labour
Cost the seconds of staffed time a drink consumes, at a loaded hourly rate.
The loaded rate is the wage plus employer contributions, holiday, any pension obligation, and the non-productive staffed hours — opening, closing, cleaning, restocking, the quiet stretches that still require someone present. Divide your total staff cost for a period by the productive hours actually worked; that is the rate to use. Using the bare wage instead is a common way a costing exercise flatters itself.
For the seconds, time the drink during a real service rather than a demonstration: grind, dose, tamp, lock, extract, texture, pour, lid, hand over, plus attributable wiping. Time it at peak, and note separately whether the drink is the one that holds up the queue — a drink that consumes a barista's whole attention during the rush costs more than its own seconds, because it displaces other sales.
Energy
You need energy attributable to producing the drink, which is not the same as your electricity bill divided by drinks sold.
Take the machine's active draw during production and the length of the production cycle to get energy per drink, then handle the standing load separately: idle draw across the trading day is a fixed cost of being open, not a variable cost of one more coffee, so it belongs in overhead. Treating it as per-drink makes quiet days look like a costing problem when they are a utilisation problem.
Include the grinder, the water boiler, milk refrigeration and any milk system, taking figures from equipment specifications or a plug-in meter rather than estimating.
Water and treatment
Water in the cup is small. Treatment is not necessarily small: filter or softener cartridges are replaced on a schedule set by your hardness and throughput, and reverse osmosis has both consumable and reject-water costs. Divide the cost of a cartridge by the litres it is rated to treat for a cost per litre, which converts easily to a cost per drink. Where water is metered, include what goes down the drain in backflushing, rinsing and purging, plus any volume-based wastewater charge.
Equipment amortisation
This is the line almost everybody omits, and omitting it means your costed drink is not the cost of making a drink; it is the cost of making a drink on equipment somebody gave you.
The method:
- Total the acquisition cost of the equipment used to make the drink — machine, grinder, water treatment, milk system, refrigeration — plus installation, site works and initial training.
- Add lifetime servicing and wear parts: planned maintenance visits, burrs, gaskets, seals, showers, valves, cartridges. Ask suppliers for service intervals and replacement lives in writing — these are inputs you cannot invent.
- Deduct any realistic residual value.
- Choose a realistic service life — how long you intend to run it, which is usually not the warranty period.
- Estimate total drinks over that life from your current volume, adjusted only for growth you can defend.
- Divide. Net cost over life, divided by drinks over life.
On a well-utilised bar the result is usually a modest per-drink figure — smaller than milk, smaller than labour. On a low-volume bar it is not modest at all, because the same cost spreads across far fewer drinks. That asymmetry is the argument for matching equipment to volume, and it only becomes visible once amortisation is in the model. The total cost of ownership of a commercial coffee machine covers how to establish the inputs.
If you lease rather than buy, the lease payment is a fixed cost of trading rather than a per-drink one — but for comparing machines, converting it to a per-drink equivalent over the term makes the options comparable.
Waste and remakes
Beyond the coffee and milk waste already counted, there is finished product that never sells: remakes, orders abandoned at the counter, drinks made for the wrong ticket, and anything produced speculatively before a rush.
Count remakes for a week. Staff will tell you if you make clear that the count is for costing rather than for blame. Every remake consumes a full drink's worth of everything above, including the labour, and it is the one waste line that improves immediately when the cause is identified.
Should overhead be allocated to each drink?
You can, and it is worth doing once to see the whole picture — but it is the wrong basis for most decisions.
A fully absorbed cost spreads rent, standing charges, insurance, management salaries and marketing across every drink, usually by volume. It gives a total cost per drink and a comfortable sense of completeness. Its weakness is that the allocation is arbitrary and volume-sensitive: sell fewer drinks and the absorbed cost per drink rises, which can make an item look unprofitable when the real problem is a quiet venue.
Contribution per drink — price minus the costs that vary with making one more — answers operating questions. Whether to keep an item, promote it, reprice it, or staff a busy hour all turn on contribution, because the rent does not change when you sell one more flat white.
Hold both. Use contribution for decisions about individual items and the mix; use total contribution against total fixed costs to find your break-even day, which tells you whether the business as a whole works. A drink with healthy contribution in a venue that never reaches break-even is still a failing business, and the two figures together tell you which problem you have.
How do you build a model you will actually keep using?
A model that takes a day to update will be built once and abandoned. Structure it so maintenance takes minutes.
One sheet of inputs. Every cost you buy — coffee per kilogram, milk per litre, each consumable per unit, your loaded labour rate, energy per unit, treatment per litre, amortisation per drink. Nothing else in the workbook holds a bought-in price, so a supplier increase means changing one cell.
One row per drink. Quantities only — grams of coffee, millilitres of milk, units of each consumable, seconds of labour — referencing the input sheet for cost. Quantities change rarely; prices change often, and separating them is what makes the model survivable.
A waste factor per input, held as its own cell, derived from your measurements rather than assumed. When you re-measure, you update one number and the whole menu re-costs.
A results view showing cost, price, contribution and contribution as a share of price for each drink, sortable by contribution and by volume.
A date on the sheet, and a habit of revisiting it whenever a supplier price or a recipe changes, and otherwise on a fixed schedule. A model reflecting last year's milk price is worse than no model, because it is trusted.
A spreadsheet is sufficient unless your till system will do this properly. Sophistication is not the constraint; being maintained is.
What is the model actually for?
Three uses, in ascending order of value.
Pricing. With a real cost you can set a price deliberately: cover the cost, contribute to fixed costs, then check the result against what your market will bear. Where the defensible price sits below the cost, the item needs re-specifying, the cost needs attacking, or it should not be on the menu — all better outcomes than discovering it a year later.
Menu decisions. Sorting drinks by contribution against volume shows which items to feature, which to simplify and which to remove. Items with low contribution that also slow the peak are doubly expensive, occupying the barista during the only hour when capacity is scarce. Items with strong contribution that customers rarely order are a display and training opportunity rather than a pricing one.
Knowing which drinks carry the business. In most venues a small number of items generate most of the total contribution, and it is common for the item an operator is proudest of not to be among them. Once you know which drinks pay the rent, decisions about layout, training, equipment, prep and promotion get easier: you protect the throughput and quality of those items above everything else.
Where this exercise is not worth doing
If you sell a very small number of drinks a day, the per-drink model will tell you what you already suspect: fixed costs dominate and no pricing decision fixes them. The problem is volume or the cost base, and the effort is better spent there.
Nor should the model become the business. A costed menu supports decisions; it is not a strategy. Customers do not buy your contribution margin — they buy a drink they like, in a place they like, from people they like. The arithmetic exists to make sure doing that well also makes money.
And if the model shows a high equipment cost per drink, the answer is not automatically to spend. It may be to sell more drinks on what you already own.
Frequently asked questions
How often should I re-cost the menu? Whenever a significant input price changes, whenever a recipe or dose changes, and otherwise on a schedule you actually keep. Milk and coffee move often enough that a costing left untouched will be wrong in the direction that hurts. With a single input sheet, updating it takes minutes.
Should I include the owner's own labour? Yes, at a rate you would have to pay someone to do the same work. An operation that only shows a profit because the owner works unpaid is subsidised rather than profitable, and it cannot be delegated or sold. Costing your own time also tells you whether hiring is affordable.
Is contribution or gross margin the better figure to watch? Contribution per drink for decisions about individual items and the mix, because it isolates what changes when you sell one more. Total contribution against total fixed costs for the health of the business. A margin expressed as a share of price compares easily but hides the fact that two drinks with the same share can contribute very different amounts in absolute terms.
Do I need to cost every item on the menu? No. Start with the drinks that make up the bulk of your sales, and any item you suspect is being sold at a loss. Together those cover most of the value. Rare items can wait, though anything complicated and slow deserves attention regardless of volume, because it consumes peak capacity.
How do I handle a drink whose cost comes out above what customers will pay? Establish the fact, then choose deliberately: re-specify it (different milk, different size, simpler build); reprice it and accept lower volume; keep it as a loss leader, knowing consciously what it costs; or remove it. All four are legitimate. Discovering the problem is what matters.
This article is part of the Vea Group knowledge base. Vea Group S.p.A. is an Italian designer and manufacturer of professional and premium coffee machines, with heritage dating to 1919 and production in Chignolo d'Isola, Italy, and Suzhou, China.