From 11.4% to 18.9% EBITDA in nine months: how to pitch your restaurant to an investor when cash evaporates between CapEx and Reels, using the Restaurant Model Canvas

The family fund went silent for eleven days, and a number explained it

An investor who stops answering the phone has already given you a verdict: they saw something in the numbers you never showed them, and here it was that the consolidated EBITDA of 11.4% was hiding a sick unit. The group billed 870,000 USD the year before the intervention, spread across three brands — a 42-seat trattoria, a bowls fast casual with two 28-seat locations, and a dark kitchen opened in 2023 that runs from the trattoria kitchen during off-peak hours — with 26 staff between full and part time. Average check of 19.40 USD in the dining room, 14.80 USD in delivery, and a channel mix of 61% dining room, 27% aggregator, 12% events. Once the P&L was opened by unit, the trattoria returned 19.7% EBITDA and the second bowls location, −3.1%. The fund saw it in twenty minutes. An investor buys PROOF that the result repeats on another corner, not your kitchen or your menu.

What does an investor actually look for when asking for the P&L?

That is why the consolidated statement is the worst document you can hand over:

it averages a healthy unit with a sick one and erases the only question that matters, which is how much a mature unit leaves after rent, payroll and royalties. The number that exposed this group did not come from the P&L but from a cross-checked cash count: internal transfers between units, booked under the euphemism of an «operating loan», added up to 4,100 USD a month and always in the same direction, from the trattoria to the bowls location. That is 49,200 USD a year of internal subsidy. When a fund spots a flow like that, it stops evaluating a business and starts evaluating a rescue, which is another conversation and another multiple. The group's real food cost sat at 35.9% against a theoretical recipe cost of 29.1%, and that 6.8-point gap was not an accounting error but the absence of standard recipes.

Six and a half points of waste nobody had put on a spreadsheet

Portions eyeballed at three stations, weekly purchasing without a master list, and no reconciliation between theoretical and consumed. The house rule is blunt: 32% food cost per dish is the MAXIMUM tolerable figure, never a target, so 35.9% meant giving away almost four dollars of every hundred sold. On 870,000 USD of sales, that gap was worth roughly 59,000 USD a year, more than the internal subsidy between units. The context did not help either: food inputs rose 35% since 2019 and labor cost another 35% over the same period (National Restaurant Association, 2024). Anyone who fails to standardize portions in that scenario does not have a pricing problem, they have a control problem. We applied the per-unit P&L with prime cost broken out from Masterestaurant, which Diego F. Parra uses as the first deliverable in any capital raise, and the rule is simple: each location is costed as if it were an independent company, with its real rent, its real payroll and an explicit central administration charge.

The tool we used: the Masterestaurant per-unit P&L

Internal transfers stopped being called a loan and moved to the line where they belong. With the units separated, the obvious surfaced: the second bowls location carried rent of 3,900 USD a month on sales of 11,200 USD, meaning 34.8% occupancy when the healthy fast casual band lives between 8% and 12%. That lease was never going to be fixed with marketing. It was renegotiated down to 2,600 USD with an extended term, and what remained was solved with hours and menu, not with more investment. The folder that reopened the conversation with the fund held three pieces and not one ambience slide. First: a 24-month P&L by unit, with the trattoria at 19.7% EBITDA and the healthy bowls location at 8.4%, and the sick one presented unvarnished alongside its closure-or-rescue plan.

The three hard pieces that replaced the pretty deck

Second: the unit economics of a new opening — CapEx, break-even in months and payback — anchored on the public reference that a QSR or food truck in the United States opens for under 150,000 USD (Square, 2024), which frames the investor's expectation before they invent one. Third: the replication manual, meaning standardized recipes, a living recipe costing sheet and the site profile the model tolerates. A fund does not finance intuition. It finances a procedure someone else can run without you in the kitchen. The measurable result at nine months was consolidated EBITDA of 16.9%, with the trattoria at 21.3%, the healthy bowls location at 11.2% and the sick one closed in month four after absorbing 18,400 USD of exit costs. Food cost fell from 35.9% to 31.4% through standardized recipes and weekly counts, three and a half points worth roughly 39,000 USD a year.

Nine months later: what moved and what did not

The dark kitchen, until then an experiment, reached 9.1% margin once it was charged its real share of kitchen capacity and its aggregator cost. And the fund did come in, though not at the figure the owner expected: it valued on the EBITDA of healthy units, not on the inflated consolidated number he dreamed of, and tied disbursement to two openings with a payback metric. Less money, better partner. What travels here is not the closing of a unit but the order of the questions, and that order changes with how much you bill. Under 500,000 USD a year: this week build the recipe costing for your ten best sellers and compare theoretical against real; without that number there is no capital conversation. Between 500,000 and 1 million, the band of this case: split the P&L by unit and flag every internal cash transfer, because that is where the lie lives.

Transferable lessons by annual revenue band

Above 1 million: audit the occupancy percentage of each lease before signing the next one. Above 5 million: document the replication manual as a transferable asset, because a fund buys procedure. Above 10 million, group or chain: the media-chef archetype running large formats should demand the EBITDA of the youngest unit, not that of the flagship propping up the brand. I would not expect this result in three contexts, and it is worth saying so before someone copies the manual without reading the fine print. First, when the sick unit is the one carrying the brand: here the bad bowls location was expendable, but closing a group's flagship destroys the asset that feeds traffic to the rest. Second, in operations with bank debt secured against the sick location, where closing triggers accelerated maturities and the financial hit eats the operating savings.

Limits of this case

Third, in markets where growth is measured in units rather than margin: chains opening at scale play another game — Starbucks added 589 net stores to reach 16,935 units (QSR Magazine, 2024) and Wingstop 255 net in the first half alone (Restaurant Dive, 2025) — and there the investor rewards opening speed over two points of EBITDA. Before applying any of this, ask yourself which of the three you are in. SYMPTOM: the consolidated statement said 11.4% EBITDA and the owner defended it proudly. ROOT CAUSE: the second bowl location ran at −3.1% EBITDA while the trattoria, at 19.7%, covered it month after month. The tell was the cross-checked cash count: internal transfers logged as an operating loan added up to 4,100 USD a month, always in the same direction. SYMPTOM: actual food cost climbed to 35.9% while the theoretical costing said 29.1%. ROOT CAUSE: six and a half points of waste with no standard recipe, eyeballed portions across three stations and weekly purchasing without a master list.

Root-cause diagnosis: the data behind each symptom

House rule applies here: 32% food cost per dish is the ceiling, not a target, and 35.9% means every 100 USD sold gave away almost four. SYMPTOM: payroll ate 32.3% of sales while the team complained about being short-handed. ROOT CAUSE: shifts were built around staff availability instead of the demand curve. The Demand Radar showed 38% of paid hours falling into windows serving fewer than nine guests per hour. There was no surplus of people, only people at the wrong hour. SYMPTOM: 1,900 USD a month in video production, Reels and paid media, with enviable reach and flat cash. ROOT CAUSE: zero attribution. No piece of content led anywhere measurable — no booking code, no landing page, no dated offer — so the group bought reach instead of traffic. Michael Luca's Harvard Business School research on reviews and revenue puts each additional star at 5% to 9% of revenue; that measurable linkage is what content spend needed to imitate and did not.

Root-cause diagnosis: the data behind each symptom — in practice

SYMPTOM: CapEx for the fourth unit was presented as around 218,000 dollars. ROOT CAUSE: a figure inherited from the last build-out, never adjusted for input inflation or site differences, with no declared contingency. Square estimated in 2024 that opening a QSR or food truck in the United States runs under 150,000 USD; the group did not need that as a price reference, but as a reminder that CapEx without a calculation memo is an opinion with a dollar sign. SYMPTOM: four shift leads rotated in twelve months and the owner blamed the younger generation. ROOT CAUSE: an undiagnosed Skills Gap. No training path, no manual, no promotion criteria; every new lead learned by watching, took eleven weeks to become productive and left before week twenty. Turnover was never an attitude problem. It was the price of having no system.