From 8.1% to 14.6% EBITDA in Eleven Months: How to Present Your Restaurant to an Investor When the Brand Sells on Reels but Bleeds in Cash, Using the Restaurant Model Canvas

Forty-one slides and only seven talked about money

The deck rejected twice had 41 slides and barely 7 touched money, and 5 of those 7 showed gross revenue. That is the full diagnosis of the case, stated before the premises. The group ran 4 gourmet fast-food locations, 112 employees between full and part time, an average ticket of 18.40 USD and consolidated revenue of 4.6 million USD a year, with 7 years of operation since the first opening. Their content asset was real: 34 million accumulated views and 61% of orders coming through the app after an organic campaign. EBITDA, 8.1%, showed up on slide 39 in footnote type, and the word CapEx never appeared at all. A committee does not reject a restaurant for being unattractive; it rejects a story where cash sits in the margin notes.

Does an audience matter to an investor or not?

An audience matters, and it matters a great deal, but it never stands alone: a committee funds repeatable margin, not reach. Sector numbers back the asset.

Each additional star in review ratings moves between 5% and 9% of an independent restaurant's revenue, according to Harvard Business School (Michael Luca), and bookings rise 30% in the week after a creator posts, according to Marketing LTB (2025). That is the tension of this case and it has to be resolved, not dodged: the group owned the asset and lacked the mechanism that turned it into auditable margin. An investor buys the repeatability of that mechanism, not the platform metric. When someone shows me views before contribution by channel, I already know where the hole sits, and it sits in the same place almost every time.

P&L by channel: 11.3% against 34.8% contribution margin

The number that cracked the case open was the P&L by channel, a report the group never produced. Content-driven delivery —61% of orders— left an 11.3% contribution margin, while the dining room left 34.8%. Three charges never entered plate costing: the platform commission, the packaging and the courtesy item riding along with every campaign order. The group closed a consolidated monthly P&L with a 40-day lag, and that lag hid real cash flow: sales climbed, cash stayed flat, and nobody could explain why inside a boardroom. That explains the rejection. Growing your most visible and least profitable channel is not growth; it is buying revenue with margin, and a committee reads that in 4 minutes.

Prime Cost at 71.9% and a 6.8-point portioning gap

Prime Cost sat at 71.9% in a format that should live near 60%, and the difference walked out right there. We worked the Masterestaurant method over the standard recipe, which existed in writing in none of 3 of the 4 locations. Theoretical food cost for the 8 star dishes came in at 29.4%, under the 32% ceiling we hold at Masterestaurant as a maximum —never a target— and the real figure from inventory came in at 36.2%. Those 6.8 points of deviation were not theft: they were eyeballed portioning, a ladle with no gram weight and a house sauce poured from a free piping bag. Input pressure amplifies any slip, because food and labor costs have both climbed steadily since 2019. With no written gram weight, margin gets decided by whoever cooks the night shift.

What the Masterestaurant method did, in order?

Plate costing with standard recipe and gram weights first, weekly P&L by channel second, and the deck only at the end. Order matters more than tooling here.

The Masterestaurant menu engineering matrix surfaced 3 star dishes that survived the delivery channel and 2 that destroyed it, and those 2 came off the app without touching the dining room menu. The courtesy item moved from every order to orders above 25 USD. P&L lag dropped from 40 days to 7. For years I got this wrong by recommending a price increase as the opening move; price lifts whatever gram weight lets escape, and that inverted order costs customers. The 18.40 USD ticket was never touched here, and Prime Cost still gave up 8.4 points in 5 months.

The result and what the committee approved on the third pass

He walked into the committee on the third round with an 18-slide deck whose first 6 slides were cash. Prime Cost from 71.9% to 63.5%, EBITDA from 8.1% to 13.9%, delivery contribution margin from 11.3% to 22.6%, and P&L by channel with a 7-day lag, auditable location by location. Consolidated revenue barely moved, from 4.6 to 4.7 million USD, and that was precisely the argument: same volume, 5.8 more points of EBITDA, with the mechanism written down and replicable for location number 5. Content went untouched across those 5 months. The 34 million views stayed where they were, and for the first time they had a line to land on. What if the founder had pushed a fourth time with the community deck? He would have burned the investment circle of his city, which is small and talks.

Transferable lessons by revenue band

The recommendation shifts with annual revenue band, and so does this week's first step. Under 500 thousand USD: write the gram weights of your 5 best sellers, by hand if needed, and weigh real yield for 7 days. Between 500 thousand and 1 million: split the delivery P&L from the dining room P&L, even on a spreadsheet, before Friday. Above 1 million —the band of this case—: pull closing lag under 10 days and compute contribution by channel, not just food cost. Above 5 million: audit CapEx per location and the real opening cost, which starts around 150,000 USD for a QSR according to Square (2024). Above 10 million, group or chain: this is where the media-chef archetype with large-format venues lives, and the risk is a personal brand covering units that bleed money; demand P&L per unit and close the 2 worst rows of the table.

Limits of this case

I would not expect this result in three contexts, and saying so matters as much as the numbers. First, in an operation with no delivery volume: when 90% of orders come from the dining room, the channel-margin finding does not exist and the work sits elsewhere, probably in table turns. Second, in a group carrying tight debt: giving up 8.4 points of Prime Cost in 5 months demands working capital to weigh, retrain and reprint menus, and whoever pays payroll with today's sales has no such room. Third, in large chains already running weekly P&L per unit: there the lever is unit development rather than costing, and scale proves it, with Chipotle opening between 315 and 345 locations in 2025 according to Chain Store Age (Q4 2024) and Wingstop adding 255 net restaurants in the first half of 2025 according to Restaurant Dive (2025). This case worked because hidden margin was there; where it is not, it does not appear.

Root-cause diagnosis: what each symptom revealed

Symptom: rising revenue against flat cash. Root cause: content-driven delivery grew on an 11.3% contribution margin versus 34.8% in the dining room, because platform commission, packaging and the courtesy item never entered plate costing. What exposed it was the per-channel P&L, which the group never produced: it ran a consolidated monthly P&L with a 40-day lag, and that lag buried the real cash flow. Symptom: a 71.9% Prime Cost in a format that should live near 60%. Root cause: three of four units had no written standard recipe; theoretical food cost on the eight star plates came in at 29.4%, inside the 32% ceiling we hold at Masterestaurant, while the storeroom actual hit 36.2%. Those 6.8 points of variance were not theft.

Root-cause diagnosis: what each symptom revealed — in practice

They were free-hand portioning and a purchase size that ignored turnover. Symptom: 94% annual kitchen turnover. Root cause: an undiagnosed Skills Gap. The group hired line cooks and handed them a 34-item menu with no manual, so learning happened by imitation and walked out with every resignation. Each departure cost 1,100 to 1,400 USD in recruiting and learning curve, and that line item sat buried inside OpEx without a name of its own. Symptom: the deck failed despite spectacular social metrics. Root cause: the founder was presenting traction while the committee was pricing risk. Two different languages. Traction answers 'why open a restaurant here at all'; risk answers 'what happens if the founder is out sick for three months'. Nobody had written the second answer, and that absence reads as institutional fragility rather than missing data.