How to read a restaurant marketing report without guessing

A good marketing report should help you make a decision. It should show what happened, how confidently you can connect it to the restaurant, and what you should do next. A large number on a dashboard is useful only when you understand what it measures.

This is especially important for restaurants because a guest's journey often crosses several places. Someone might see a social post, search your name days later, look at the menu, and walk in without making a reservation. Another person may click an ad and never visit. Both journeys can leave digital signals, but they do not represent the same business outcome.

Use this guide to review your own reporting or question a marketing partner. You do not need a perfect attribution system to make better decisions. You need clear definitions, honest gaps, and a consistent review process.

Start with the business question

Before opening a dashboard, write the question the campaign was supposed to answer. “Can we attract more Tuesday dinner parties?” is specific enough to investigate. “Did our marketing work?” is too broad to resolve with one chart.

Then identify the period, location, offer, and audience. A report covering all branches cannot automatically explain performance at one branch. A month with five Fridays should not be compared casually with a month that had four. Holidays, weather, roadworks, price changes, and service capacity can also affect the interpretation.

Record the intended decision. You may be choosing whether to continue an offer, fix a landing page, change a message, or pause spending. Knowing the decision helps you request the evidence that matters instead of collecting every available metric.

Read the report in four layers

The first layer is work completed: pages updated, messages prepared, campaigns launched, or review responses handled. This tells you whether the agreed work happened. It is an accountability measure, not a sales result.

The second layer is attention: impressions, reach, video views, or visits to a page. These measures can reveal whether people had an opportunity to notice the restaurant. Their definitions vary by platform, and totals may include the same person more than once.

The third layer is action: menu clicks, reservation starts, calls, direction requests, or completed signups. These are stronger signals of interest, but they still require careful labels. A direction request does not prove arrival. A reservation start is not a seated party.

The fourth layer is verified restaurant outcomes: completed orders, honored reservations, recorded redemptions, or other actual transaction records. Even here, attribution and causation are different questions. A sale associated with a campaign may have happened without it.

Ask for a definition beside every important number

For each headline metric, ask: where did this come from, what exactly counts, what is excluded, and when was the data last updated? “Customers generated” is not a sufficient definition. Is the number unique contacts, bookings, completed transactions, or estimated visits?

Check denominators when percentages appear. A conversion rate might divide purchases by ad clicks, website sessions, or visitors to a specific landing page. Those rates answer different questions. Record the numerator and denominator so a change can be interpreted correctly next month.

Missing data should remain visibly missing. A disconnected ordering system is not evidence of zero orders. A blank field should be marked unavailable with an explanation and an owner for the repair. Sample data belongs outside real performance totals.

Understand what attribution can and cannot show

An attribution rule assigns credit to an interaction within a defined window. One report may credit the last ad clicked; another may include a view without a click. Different platforms can claim the same transaction. Adding their reported revenue together can therefore overstate the restaurant's total.

Ask for the attribution window, event definition, and method used to connect an interaction to a transaction. Check whether refunds, cancellations, tax, tips, and duplicates are handled consistently. Where possible, reconcile totals against the restaurant's transaction records.

Incremental sales are the sales that occurred because of the campaign. Establishing that usually needs a comparison design, such as a suitable holdout or a carefully interpreted test. A before-and-after change alone cannot isolate marketing from every other change in the business.

This does not make attribution useless. It makes it directional evidence that needs a label. “Recorded orders associated with this campaign under this rule” is more defensible than “Orders we created” when there is no incrementality test.

Read email signals with care

Email opens are an imperfect signal of attention. Apple's Mail Privacy Protection can prevent senders from reliably learning whether a recipient opened an email. Treat open rates as limited diagnostic information, not a count of interested diners. [S06]

Clicks are usually closer to a deliberate action, but they still do not establish a visit. Review useful destinations, completed actions, delivery problems, unsubscribes, and complaints together. A higher click rate is not a win if the landing page disappoints people or the message drives an unprofitable offer.

Avoid automatically sending “we noticed you didn't open” messages. The measurement may be wrong, and the wording can feel intrusive. Build follow-up decisions around information you actually have and permissions the guest actually provided.

Put revenue beside the costs required to earn it

Return on ad spend divides attributed revenue by ad spend. It does not subtract food, packaging, transaction fees, labor, offer costs, creative work, or management fees. A high reported revenue multiple can coexist with a weak commercial result.

For a campaign decision, estimate contribution from the sales in question: net sales minus relevant variable costs. Then subtract the campaign's additional fixed costs. The SBA's break-even framework uses fixed costs divided by per-unit contribution; the same arithmetic can help you plan a bounded campaign, with your own cost assumptions. [S09]

Keep the scope clear. Campaign contribution is not the restaurant's total profit. Fixed rent and ordinary overhead still exist, and some labor costs change only after you cross a capacity threshold. Ask whoever knows the restaurant's costs to check the assumptions.

A worked example: an impressive number with a missing answer

Consider an illustrative campaign with $500 in ad spend and 40 recorded orders attributed under a stated click-based rule. At $35 net sales per order, the attributed revenue is $1,400 and the reported return on ad spend is 2.8. These are fictional numbers for explanation.

Suppose the relevant variable cost is $20 per order. The 40 orders produce $600 in contribution before the ad spend. Subtracting $500 leaves $100 before any additional fixed campaign costs. If some of those orders would have happened anyway, the incremental result is lower. The revenue multiple alone did not reveal this.

The right follow-up is not automatically “stop advertising.” It is to investigate the order mix, incremental demand, repeat behavior, and total costs. You might improve the offer or landing page, or decide the test is not worth extending. The report should make that decision easier.

Compare periods without pretending they are identical

Use comparable service periods where possible. For a Tuesday promotion, examine Tuesdays, not the entire week's sales. Note whether opening hours, pricing, local events, and staffing were similar. Keep a simple campaign log so unusual conditions are not forgotten during the monthly review.

Small samples deserve restrained conclusions. A change from two bookings to four is a doubling, but it may not tell you much about the next month. Show the actual counts beside the percentage. If the evidence is weak, choose a smaller next test instead of declaring a winner.

Where a holdout is practical, agree on its design before launch and avoid treating a tiny difference as proof. If a controlled test is not feasible, explain the limitations of the comparison you used. Honest uncertainty is information the owner needs.

End every review with a decision and an owner

Your closing note should contain one finding, one limitation, one next action, and one review date. For example: “Menu-page visits rose, but ordering completions are not connected. Repair the ordering handoff and tracking before increasing spend. Alex owns the fix; review next Tuesday.”

Use the monthly report worksheet to put definitions and decisions beside the numbers. DineROI's role is to help make restaurant marketing understandable and manageable. A growth plan should begin with the business outcome you need and the evidence available to judge progress.