Digital vs traditional marketing: how a casual dining venue recovered 3.1 EBITDA points by fixing its spending mix with Masterestaurant's Radar Gastronómico

The file that came in through the bank's door, not marketing's

A casual dining restaurant built on market cuisine, 22 tables and 74 seats, had grown sales for seven straight years and still could not sustain a credit line, and that was the starting point of the file. The details, so nobody extrapolates too far: 19 employees on formal payroll, 14 of them full time, a mid-sized Andean city of nearly a million people, an average check of 21.40 USD in the dining room and 16.80 USD in delivery, annual revenue between 500 thousand and 1 million USD. The dominant channel was no longer the dining room but the delivery marketplace, with 38 % of all orders. The assignment reached SATE Institute through the least glamorous route, a commercial bank with an MSME portfolio that wanted to understand why sales climbed while cash did not. The answer sat entirely inside commercial spending. Seventy-one percent of the commercial budget went to acquisition advertising, and there was the hole, because buying first visits in this trade amounts to filling a tub with no plug.

Seventy-one percent of the budget bought first visits and funded a leak

Restroworks (2025) documents that 70 % of first-time diners never return, so a plan tilted toward capture funds, by design, a leak of seven out of every ten pesos invested. The operator was putting between 4 % and 7 % of sales into attracting people without ever measuring what share came back, and once that happens the spending stops being investment and turns into unproductive OpEx. The aggregate effect is what worries multilateral banks: early business mortality and destruction of formal employment in an industry that, according to the International Labour Organization, concentrates much of the region's youth employment. Picking a side between digital and traditional is the wrong question, and arguing otherwise has cost thousands of operators their margin. The real difference was not the channel, it was the UNIT OF MEASURE: the old plan bought first visits and counted them as wins, while the method we applied buys second and third visits, which is where margin lives because no discovery commission gets paid again.

Digital versus traditional is the wrong question, and it burns cash

I got this wrong for years, recommending channel mixes when the problem was never the mix but the denominator. Digital advertising measured by reach and a billboard measured by impressions commit exactly the same accounting sin. Paytronix (2024) shows that operators in the 90th percentile draw more than 37 % of their transactions from loyalty members, and that share does not care whether the guest arrived through Instagram or a paper flyer. The lazy consultant's temptation is to recommend shutting down everything offline, and in this case that would have destroyed the most profitable commercial asset in the operation. The partnership with the gym next door and the arrangement with the office three blocks up produced 62 lunch covers a week at practically zero acquisition cost, and nobody had ever counted them because they showed up on no dashboard. What we did was instrument them with an agreement code at the point of sale, track them for eleven weeks and scale them to two more partners.

Traditional was not switched off: it was measured for the first time

At the same time the digital front was reordered toward what can actually be capitalized: WebFX (2026) reports that a complete Google Business profile is seven times more likely to receive clicks, and that asset charges no commission on every order it delivers. The budget did not rise by a single dollar, and I insist on that because half the trade believes retention is a new line of expense. From the 71 % that went to acquisition we came down to 34 % in two quarters, and the difference moved to three concrete destinations: a repeat-purchase program with a per-visit benefit, an owned ordering channel, and contact data capture in the dining room and in delivery. The direct-ordering lever was served on a platter, since Paytronix (2024) documents that 70 % of consumers prefer ordering straight from the restaurant rather than a third party, a preference this operator had never put to work while handing 38 % of its orders to the marketplace.

Reallocating without spending a peso more: money moved from acquiring to retaining

First-party data is no luxury for big chains. It is the only way to stop renting the relationship with your guest. The MASTERESTAURANT method led by Diego F. Parra attacked this with the return-per-commercial-peso matrix, the Masterestaurant ecosystem tool built to split spending into two thankless columns: pesos that buy a transaction and pesos that buy a relationship. It was applied like this, and the order matters: first every commercial spending line from the previous twelve months was tagged, then it was crossed against the point-of-sale recurrence history, and only at the end was the current budget touched. Restroworks (2025) reports that 72 % of people use social media to research restaurants, a figure that inside the matrix justifies discovery content at marginal cost rather than paid advertising. The condition without which none of this holds up is having your repeat guest identified. Without that identifier, any return calculation is literature.

Transferable lessons by annual revenue band

The recommendation shifts with the revenue band, because the first step for an operator at 400 thousand looks nothing like the first step for a fifteen-unit group. Under 500 thousand USD: this week add a phone or email field to every closed check, by hand if you must, and count how many guests repeat within 60 days. Between 500 thousand and 1 million, the band of this case: audit what share of your commercial spending buys first visits and set yourself a ceiling of 40 %. Above 1 million: turn on an owned ordering channel and measure its share against the marketplace. Above 5 million: install a loyalty program with an explicit target of 37 % of transactions, the Paytronix (2024) benchmark. Past 10 million, the celebrity-chef archetype running large-format themed venues: consolidate the database into a single instance before buying one more impression. I would not expect this result in three contexts, and I would rather say so before somebody copies the plan blind.

Limits of this case

First, a venue with under two years of operation: with no recurrence history there is nothing to reallocate, and there acquisition genuinely is the right investment until a base gets built. Second, an operation with unstable kitchen or service quality, because retaining an unhappy guest only accelerates negative word of mouth, and no repeat-purchase program fixes a badly costed dish or a poorly run dining room. Third, markets with very high digital penetration and aggressive price competition, where Circana (2025) found that 50 % of those who stopped eating out would come back with lower prices, a sign that the problem lies in the value proposition and not in the media mix. The mid-sized city in this file also looks nothing like a saturated capital. The difference was not the channel, it was the unit of measurement. The traditional method bought first visits and booked them as success; the method we applied buys second and third visits, where margin lives.

What actually separated the two methods?

Restroworks (2025) documents that 70 % of first-time guests never return, so a budget tilted toward acquisition is financing, by design, a 70 % leak.

Traditional marketing did not vanish from the plan, and that nuance matters, because the lazy consultant's temptation is to recommend switching off everything offline. The alliance with the neighboring gym and the deal with the office three blocks up produced 62 weekly lunch covers at practically zero acquisition cost; what we did was measure them for the first time and scale them, while cutting flyers, which over eleven tracked weeks brought 9 identifiable guests. First-party data changed the nature of the asset. A restaurant without a contact base rents its demand from a marketplace; with an owned base, demand becomes an asset you can reactivate without paying a toll. Paytronix (2024) measured that 90th-percentile operators draw over 37 % of transactions through loyalty members, and that is the ceiling we aim at, not an average reachable in one quarter.

What actually separated the two methods — in practice?

Online reputation turned out to be the cheapest channel and the worst attended, an imbalance repeated across almost the entire 500 thousand to 1 million band.

WebFX (2026) reports that complete Google Business profiles are seven times more likely to receive clicks, and completing nine fields plus answering 41 reviews cost 11 hours of one person's work: no paid campaign competes with that return per hour invested. From a development standpoint, the difference reads in employment. Commercial spend dropping from 6.8 % to 4.3 % of sales while EBITDA climbed freed cash to formalize two part-time positions and raise the kitchen base wage, and turnover fell from 96 % to 61 % annually. That is SDG 8 measured on payroll, not in a statement of intent.