Miscalculated prime cost: the capital leak we fixed with the Restaurant Model Canvas and the Standard Recipe Generator

The starting picture: 78,000 USD in sales, cash still thin

The mistake wasn't having a 68% prime cost: it was not knowing it. This full-service trattoria with 26 tables and 19 employees was selling close to 78,000 USD a month, a 14.80 USD average ticket, six years in business and 72% of sales in the dining room. The owner arrived with the line I hear over and over: «I'm billing more than ever and the bank gets thinner every month». Real prime cost —food cost plus labor cost— was near 68% of sales. To size the leak: full-service food cost for restaurants under 2 million dollars runs a 33.7% median (National Restaurant Association, 2025); he thought he sat at 61% total and was seven points above. Every point over 60 eats your EBITDA. The symptom was tight cash; the root cause, a number nobody was measuring right. The first leak was costing recipes with old prices.

Leak one: costing food with last year's price list

The kitchen built recipe costs off the prior year's supplier list, and input inflation had pushed the real cost up almost 6 points without anyone noticing. Food cost «on paper» read 32%, close to the full-service median of 32.0% (National Restaurant Association, 2025); the real one had drifted to 38%. That gap, on 78,000 USD a month, is more than 4,600 USD evaporating into the plate with no explanation. In a country where 54% of the 130,000 food establishments operate informally (Acodrés, 2025), many never re-cost at all. The Masterestaurant rule is firm: food cost per dish caps at 32%, recalculated every time a key input moves more than 5%. Not once a year. The second leak was conceptual and it warped every pricing decision: they folded part of rent and electricity into each recipe's «cost». A restaurant's typical electric bill runs around 2,300 USD a month (Toast, 2025); prorating it dish by dish made gross margin look worse than it was.

Leak two: loading rent and utilities into the plate cost

With that inflated number, the owner raised prices where he shouldn't and gave away margin where there was room. Fixed costs are paid with volume, not recipe by recipe: they belong in the break-even, never in food cost. Once we stripped plate cost down to input and waste only, real gross margin surfaced and menu decisions straightened out. The MASTERESTAURANT method keeps two separate buckets: variable cost to the plate, fixed cost to the break-even. Mixing them is the mistake I see over and over. The third leak was timing: the P&L landed 45 days out. Deciding with numbers a month and a half old is driving by the rearview mirror; by the time the owner finally saw the drift, he had been bleeding cash for six weeks. And cash rules: poor cash-flow management is tied to roughly 82% of small-business closures (Inc., U.S.

Leak three: deciding off a P&L that arrived 45 days late

Bank study). We set up a weekly prime cost dashboard with the Masterestaurant costing tool (herramientas_restaurantes.html): food cost and labor cost recalculated every Monday off the week's real purchases, not off the late accounting P&L. Run that way, the owner went from finding out at 45 days to correcting at 7. The first month he already cut three points of prime cost; that cadence change alone is worth more than any rushed price hike. The measurable result was cutting prime cost from 68% to 59% in three months, without raising the 14.80 USD ticket or letting anyone go. The lever wasn't selling more: it was measuring right and splitting the buckets. We re-costed the menu with the month's real prices, pulled fixed costs out of the plate, and renegotiated two inputs that had spiked. Real food cost fell from 38% to 33%, back inside the healthy full-service band (32.0% median, National Restaurant Association, 2025).

The result: prime cost from 68% to 59% in three months

Labor cost dropped from 30% to 26% by matching shifts to the dining room's real demand curve, which carries 72% of sales. On 78,000 USD a month, nine points of recovered prime cost is close to 7,000 USD a month that used to evaporate. The cash stopped lying. The transferable lesson is that prime cost is measured, not guessed, and the first step depends on your size. Small independent (one bar, one location): this week re-cost your five best-selling dishes with last month's real invoices, not the old list; that's where 80% of the leak usually hides. Mid-size (full service, 15-25 employees like this trattoria): build a weekly prime cost dashboard —food cost plus labor cost over sales— and put one person in charge of it; correcting at 7 days beats any promotion. Multi-unit group: standardize costing per site with the same engine and compare prime cost across locations; the one running five points worse is pointing at the leak.

Transferable lessons: what to do this week by your size

In all three, the rule holds: fixed costs to the break-even, variable to the plate, and prices recalculated when the input moves, not when the year closes. This case is an anonymized composite from Diego F. Parra's practice (+8,400 restaurants across 43 countries) and it doesn't promise the same jump in every context. First: a limited-service or high-volume operation already starts from lower food cost —a 32.4% median in 2025, down to 31.0% at sites over 2 million dollars (National Restaurant Association, 2025)—; there the re-costing upside is smaller and the real lever is labor and waste. Second: if the leak is structural rather than measurement —disproportionate rent, a dead location, a ticket below break-even— no dashboard fixes it; you fix the model first. Third: in markets with violent input inflation or an unstable supply chain, food cost moves so fast you must re-cost almost weekly, and the result holds only if that discipline stays.

Limits of this case: where I would NOT expect the same result

Without the habit, prime cost climbs right back within a quarter. Measuring food cost on old prices: input inflation had raised the real cost nearly 6 points and no one saw it. Prime cost «on paper» was 61%; the real one, 68%. Loading fixed costs into the plate: pushing part of rent and utilities into each recipe's «cost» made gross margin look worse than it was and skewed pricing decisions. Fixed costs are paid with volume, not plate by plate. Late P&L: deciding on 45-day-old numbers is driving by the rearview mirror. By the time the owner saw the leak, cash had been bleeding for six weeks.