Marketing metrics that matter: what to measure, what to pay for, and what to cancel in 2026

Which four marketing metrics actually move the cash register?

There are four, and none of them lives on Instagram: customer acquisition cost (CAC), 90-day repeat frequency, average check by channel, and contribution margin after commissions.

That 34-indicator report your agency sends every Monday —reach, impressions, saves, follower growth by time slot— describes how a platform behaves, not how your business does. The difference is an accounting one: reach has no owner inside the restaurant, while CAC forces somebody to sign off on a number. With 72% of people using social media to research restaurants (Restroworks, 2025), visibility does matter; yet visibility turns into money only once you know what you paid for each guest who walked through the door and how many times that guest came back. As of August 2026, tracking those four costs between 0 and 180 USD a month, and the range depends far less on software than on how orderly your data already is.

What each price tier includes, from 0 to 400 USD a month?

The zero-dollar tier is real and it works:

Google Business Profile, your POS's native analytics, and a spreadsheet fed by two weekly exports will give you CAC, check by channel, and 90-day repeat rate for about four hours of work per month. That free tier pays off more than most owners assume —complete Google Business profiles are 7 times more likely to get clicks (WebFX, 2026)— and it charges nothing per data point. Between 40 and 120 USD monthly you get the automatic connector: the tool matches ad spend against POS sales without you exporting anything, campaign attribution usually included. From 120 to 400 USD the loyalty module with guest identification appears, and that is the only thing turning "estimated repeat" into measured repeat. Operators in the 90th percentile pull 37% or more of their transactions from loyalty members (Paytronix, 2024). Above 400 USD you are buying volume and seats, not new capability.

CAC vanishes from the report because joining two systems takes labor, not money

Working out your acquisition cost means pairing ad spend with sales recorded at the point of sale, and no platform hands that pairing over, because exporting two files every Monday is a job with a person's name attached. Meta gives you impressions; your POS gives you tickets; nobody marries them unless you pay for it or do it yourself. Divide the month's total investment —paid media, agency fees, discount promotions— by the new guests you identified, and the uncomfortable figure that comes out usually lands between 9 and 34 USD for an urban independent. Diego F. Parra keeps hammering the same point at Masterestaurant: anchor that number to the contribution margin of your average check. If your margin per guest is 11 USD and your CAC is 19, every campaign you celebrate at the Monday meeting is draining cash. This is not a creative problem. It is arithmetic.

Delivery gets reported gross, and that is where real margin hides

Delivery platforms report conversion against the gross order value, and with commissions running from 15% to 30% depending on the deal you signed, a channel that looks like 22% margin may be handing you 4% in practice. That gap is no reporting nuance: it decides whether you open another shift or shut it down. Nor is the channel marginal —roughly 75% of restaurant traffic now happens off-premise (Circana)— so an 18-point misreading gets multiplied across three quarters of your operation. In the United States, DoorDash closed 2024 with 60.7% of the market, Uber Eats with 26.1%, Grubhub with 6.3% (Earnest Analytics); in Brazil, iFood holds 80% (Grand View Research). Such concentration strips your negotiating power, which is why the 67% who prefer ordering straight from the restaurant's own site (Statista) is worth gold: every point you shift from the aggregator to your own channel stays whole in the till.

Five factors that move the price of measuring, and how much each weighs

Location count rules everything: going from one site to five typically multiplies the bill by 2.5 to 4 times, since almost every vendor charges per active location. Second comes the POS: an open API puts the connection at 30 to 60 USD a month, while a closed or aging system adds a one-time 200 to 600 USD of integration work, or condemns you to manual exports forever. Third, social listening bills by mention volume, and because an independent generates between 8 and 40 mentions weekly, you will nearly always be paying a floor price engineered for chains. Fourth, guest identification —email, phone, or card— adds 40 to 90 USD monthly, and it is the single thing that makes genuine repeat measurement possible. Fifth, running several delivery channels doubles reconciliation work: each aggregator exports in its own format on its own settlement calendar. Start by measuring three months for free before you sign anything, because the vendor quoting you today has no idea how many identified guests you actually have, and that blind spot gets billed to you as the priciest plan.

How to negotiate and cut the bill without losing the data

Once the history is in hand, ask for per-location pricing with a volume discount starting at the third site, an annual contract with a six-month exit, and data migration thrown in —that last line is the one they usually concede when they sense the fight is real. With aggregators, commission gets negotiated by tier: commit to a monthly volume and push for two or three points off the standard 15%-30%, or trade temporary exclusivity for a cut. Then move traffic to your own channel using the cheapest lever available, which is price. Half of the people who stopped dining out would return at lower prices (Circana, 2025), and an 8% discount on direct orders costs you less than a 25% commission. The first is your time, usually the dearest of the lot: four to six hours a month from someone earning 12 USD an hour works out to 60 or 70 USD that never show up on the software invoice.

The three costs nobody puts in the quote

The second is data cleanup, because a menu with inconsistent item names across the POS and the aggregator breaks any attribution, and sorting it takes ten to twenty hours the first time around. The third is the cost of the bad decision, which has no budget line but plenty of consequences —the client billing 61,000 USD a month without knowing his CAC had spent fourteen months bankrolling a channel that lost money on every order. Suppose you kill that channel today: you drop 9% of volume, recover four margin points on the rest, and next quarter your profit climbs even as your revenue falls. No social media dashboard will ever run that calculation for you. If you can only measure one thing this quarter, measure how many of your January guests came back before April, because that figure holds everything else inside it: product, service, price, and the true quality of the audience your ad spend bought.

The 90-day repeat rate separates a business from an experiment

A healthy urban independent runs somewhere between 22% and 38% repeat at 90 days; below 18%, what you have is not a marketing problem but a missing reason to return, and no campaign fixes that. Formal loyalty helps —37% or more of transactions among 90th-percentile operators (Paytronix, 2024)— though the program is the measuring instrument, not the cause. One warning belongs here: high repeat paired with a low check can be discount addiction dressed up as loyalty, and you tell them apart by comparing the average check of returners against everyone else. Export those two numbers from your POS tomorrow. The agency report measures whatever the platform gives away free, and platforms give away what makes them look essential. Reach and impressions cost nothing to compute and have no accounting owner inside the restaurant. CAC requires joining two systems —ad spend and the POS— and nobody does that join for free.

Where the traditional model breaks?

So it vanishes from the report: not because it is hard, but because it forces someone to export two files every Monday. Delivery conversion gets reported on gross order value.

With commissions running 15% to 30% depending on the contract, a channel that looks like 22% margin may be delivering 4% real, and that gap decides whether you open another shift or close it. Social listening prices by mention volume. An independent restaurant generates 8 to 40 weekly mentions; paying 600 USD a month to track that volume works out to more than 3 USD per mention read.