Restaurant sales growth plan: myth vs reality, with measurable deliverables

Step 1: calculate your real acquisition cost before spending a single dollar on ads

The cost of acquiring a new diner is calculated by dividing ALL of the month's marketing money —paid ads, agency fees, creator barters, welcome discounts— by the number of guests who show up in your POS for the first time that month, and the deliverable is one single number, written in one cell. At the Medellín restaurant that opens this guide, that division came out at roughly US$10 per new guest against US$9.20 of contribution margin on the first visit: the operation was buying customers at a loss. ChowNow measured a paid CAC of US$27 in fast food and close to US$180 in fine dining for 2025, so if your figure lands far below your segment's band, you are probably counting the denominator wrong. Verify it on Friday: export unique guests for the month, subtract the repeats and divide again.

Step 2: build monthly repeat cohorts, because the average is lying to you

A cohort is the group of guests who came in for the first time during the same month, and you track it separately across the following three months to see how many return for a second and a third visit. Your overall average hides this: in the Medellín case, the third visit happened for barely 19% of new guests, and that 19% carried the entire profitability of the business. I got this wrong for years by watching aggregate frequency. The deliverable is four rows of data per cohort —new guests, how many returned within 30 days, within 60, within 90— plus an explicit target for the second visit, which in my judgment should not drop below 30% in a chef-driven room with a mid-to-high average check. Cadence rules here: review cohorts every seven days, not monthly.

Step 3: plug the repeat leak first, even if leaving the ad budget idle hurts

Before buying one more click, spend three full weeks bringing back the people who already know you, and the reason is plain arithmetic: if a stranger costs you US$10 to attract and someone already in your database costs US$1, the second dollar works ten times harder. Run the counterfactual all the way. Say you lift the second visit from 24% to 34% over a cohort of 900 new guests: that is 90 extra covers with zero acquisition spend, at a US$15 check, roughly US$1,350 of incremental sales at full margin, and that money pays next month's advertising without touching the till. The deliverable is a three-message sequence with send dates, a defined segment, and a return rate measured inside the POS through a reservation code. Cheap acquisition in restaurants almost never comes from ads: it comes from winning the search made by someone who wants to eat NOW, which is a different kind of traffic altogether.

Step 4: make acquisition cheaper by squeezing local search before paid media

BrightLocal measured that 76% of mobile "near me" searches end in a physical visit within 24 hours, and Malou reported that 79% of restaurant searches carry no brand name, meaning people ask for a category and a neighborhood, not for your room. Diego F. Parra insists on this sequence inside the Masterestaurant method because organic CAC sits around US$9 per ChowNow against US$27 for paid. The deliverable: a verified Google profile with real hours, 20 photographed dishes with prices, and 12 new reviews a month with written replies. Weekly verification in the search panel: non-brand queries that generate calls and directions. Average check moves when you change the architecture of the menu, and the deliverable is a new menu with four decisions already made: two anchor dishes pulled for weak margin, one fixed-price pairing set mid-band, a shareable dessert with food cost under 32%, and the reading order rebuilt so the highest contribution margin lands in the upper right third.

Step 5: raise the average check through menu design, not server suggestion

At the Medellín room the check had been stuck at US$15 for fourteen months with two agencies involved, because no agency touches the menu: it is not their job. Moving that check by 8% was worth close to US$2,900 a month with zero additional ad spend. You verify it by comparing the average check of the four prior weeks against the four following ones, broken out by service period. A new channel opens once the previous three levers each have an owner and a number, never before, because a channel amplifies whatever you already have: if repeat business is broken, the channel multiplies the leak. When its turn arrives, go in with a written budget cap and a cutoff threshold. Public benchmarks give you a floor: PPC Chief reports an average CPC of US$2.05 and a 7.6% CTR for restaurants and food in 2026, while Grand View Research projects Latin American online delivery growing 8.6% a year through 2030, which explains why the channel exists but not why it suits you.

Step 6: open the new channel last, and with a written spending ceiling

With delivery commissions at 25% to 30%, a dish at 30% food cost leaves almost nothing. The deliverable: one sheet with monthly ceiling, maximum tolerated CAC and an automatic shutoff date. The mistake I run into again and again is inverting the order: launching delivery or hiring an agency while repeat business still sits at 19%, which means paying for expensive traffic to fill a punctured bucket. Second comes leaving the four numbers without an owner who has a first and last name; a number that belongs to everyone belongs to nobody and reaches the committee with no one to explain it. Third is measuring monthly: a campaign burning cash takes four weeks to surface in an agency report and seven days to surface in a cohort, and those three weeks are worth between US$700 and US$2,200 depending on the size of the room. Fourth is mistaking engagement for cash: 56% of creator campaigns aim at generating user content per Socially Powerful, and content is not a booking.

Closing checklist: how to know the plan is actually built

The plan is finished when it fits on one page and clears seven concrete checks: the four control numbers exist with this week's figure, each one has a named owner, CAC sits below the contribution margin of the first visit, last month's cohort is already loaded with its 30 and 60 day cuts, the new menu has at least four weeks of check measurement behind it, a written spending ceiling exists for any open channel, and a 25-minute meeting is on the calendar every Friday with those four numbers on screen. If a single one fails, the plan is not built: it is merely drafted. And that distinction costs money. Open the POS today, export new guests from the last 90 days and calculate your first cohort before Friday. The channel plan optimizes TRAFFIC; the unit-economics plan optimizes the margin that traffic leaves behind. A restaurant can double visits and lose profit whenever customer acquisition cost exceeds first-visit contribution margin, which happens far more often than the trade admits.

Four differences that change the cash register

Review cadence: monthly against weekly. A campaign burning budget takes four weeks to show up in an agency deck and seven days to show up in a cohort. In restaurant marketing those three weeks are worth between 3 and 9 million pesos depending on the size of the room. How repeat business is treated. The channel plan assumes it; the growth plan measures it by cohort and sets a target: if 19 out of every 100 new guests return and you take that to 31, sales climb 12% without a single extra peso of paid media. Cheapest lever in the whole sales funnel, and almost nobody pulls it first. Who owns the number. In the channel plan the owner is the agency, which controls none of the operation. In the unit-economics plan each of the four numbers carries a name from inside the restaurant, and the person answering for average check is the floor manager, never the community manager.