Restaurant Promotions: Fill Seats Without Giving Away the Margin
Choose, launch, and measure restaurant promotions by verified visits, settled sales, repeat behavior, and contribution—not claims alone.
Restaurant promotions work when they target a visit your restaurant needs, lead to a paid check, and leave money after every applicable promotion cost. Treat the promotion as one part of your restaurant marketing system: choose one business problem, make the guest response trackable, verify the in-person visit and settled check, then watch what those first-time guests do after the offer.
A claim, reservation, click, or coupon download is not a completed visit. Tracked sales are not profit, and a sales increase during a promotion does not prove the promotion caused every extra dollar.
Give each restaurant promotion one job
The right promotion starts with the result your restaurant needs, not a clever discount. A Monday dinner offer, a new-item launch, and a reward meant to bring a first-timer back should not use the same audience, timing, or score.
| The job | A promotion that fits | The result to record | The mistake to avoid |
|---|---|---|---|
| Bring in new guests | A first-visit perk tied to a specific meal period | Unique first-time guests with verified in-person visits and matched paid, non-voided checks | Counting claims as customers |
| Turn interest into visits | A clear, time-bound reason to finish a reservation or claim | Eligible records whose full visit window has elapsed and whose outcome is reconciled and known: completed or did not complete | Mixing open, unknown, or immature records with finished outcomes |
| Fill a weak daypart | A lunch bundle, early-week experience, or daypart-only extra | Completed checks or covers compared with the same day and hours | Discounting a shift that was already full |
| Raise the average check | A bundle or add-on that protects the base item’s margin | Net settled sales and average check for matched transactions | Calling more sales more profit |
| Earn another visit | A milestone reward after the first completed visit | Matured first-time guests who return within one fixed window | Counting guests who have not had enough time to return |
| Learn what to run next | One offer, one audience, and one written success rule | Directly matched results, a like-for-like observed difference, or a forecast, kept separate | Treating a forecast or before-and-after change as proof |
A forecast can show how many new guests, later visits, dollars, and time would be required to hit a goal. It cannot prove that a particular promotion will create that growth.
This choice also protects you from solving the wrong problem. If first visits are healthy but almost nobody comes back, another acquisition offer may produce a busy opening week and the same empty dining room later. If guests return but lunch has too few first-timers, a loyalty reward is not a substitute for giving new people a reason to try lunch.
“We definitely need return business. I want new guests, but ideally I want them to become continuous guests, regulars.” A two-location pizza owner put it plainly.
The offer should make the right visit more attractive
A strong promotion adds a reason to visit now without teaching every guest that your regular price is optional. Start with the occasion, item, or access people already want; use the perk to remove the last bit of hesitation.
“Because the experience is going to attract the user, not the offer.” That was one restaurant owner’s reason for planning a separate promotion for each event night.
That distinction opens up more useful restaurant promotion ideas than a blanket percentage off:
- For a weak midday shift, build around the meal people need at that hour. Match your lunch and daily specials to that daypart rather than discounting the whole week.
- For trivia, live music, or another event, sell the experience and let the perk help someone choose the date. The headliner is the reason to go; the offer is the nudge.
- For a new or unfamiliar item, give the guest a lower-risk first taste through a fixed bundle, add-on, or prepay option. Keep the item, redemption period, and eligible order written on one offer card.
- For a premium restaurant that will not coupon, use access: a limited seating, an off-menu taste, or an existing happy-hour benefit at a different eligible time. The principle behind protecting your brand from discounts is to offer something specific instead of marking down everything.
- For a first visit, attach a small extra to a paid meal. The base check still matters; the extra gives the guest a reason to act.
- For a return visit, make the second reward contingent on a separate completed first visit. Enrollment is not loyalty, and the first check cannot prove the later check happened.
Do not copy an offer merely because another restaurant reported strong results. A tourist location, a neighborhood lunch spot, and a reservation-only dining room have different return opportunities and capacity. Take the offer shape, write your own margin and staffing limits, then test it with your own guests.
Four promotion results that can fool you
Four common results look encouraging before you have actually learned whether the promotion worked: responses without arrivals, arrivals without contribution, sales without a fair comparison, and first visits without a mature return window.
1. Lots of claims, few completed visits
High response tells you the offer earned attention. It does not tell you that people arrived, paid, or will return.
One Feast free-fries campaign produced about 822 signups, yet redemption was poor after a reminder failed, severe weather hit the redemption window, and the restaurant was closed on important redemption days. The response count did not reveal those breaks between signup and service.
Keep every eligible claim or reservation until its full visit window passes, then reconcile the outcome. The denominator for your completed-visit rate is all eligible records whose full window has elapsed and whose outcome is reconciled and known. The numerator is the subset of that same group whose outcome was a verified in-person completed visit. Exclude early completions, open records, and unknown outcomes until the window closes, and report them separately. Keep claims and reservations in separate reports because one record can represent a person while the other may represent a party.
2. A full room, but weak contribution
Revenue from a promotion can rise while the restaurant gives up too much on food, fees, or labor. Check the actual cost of the promoted item and whether you discounted something that already sells at full price.
An owner rejected a proposed half-price reward on his tri-tip because the item cost him $6, sold for $24.95, and already sold out. He changed the idea to half off dinner for four with one unique entree per person, limiting each redemption to one tri-tip.
“I’m giving away something I’m making the best money on right now, and I already run out.” The owner’s objection stopped the wrong reward before launch.
His decision points to a useful rule: do not spend promotion dollars creating demand for the item or service period already constrained by supply. Put the offer where another paid check is useful.
3. More sales during the promotion, but no clean cause
A promotion-period increase is an observed difference until you compare like with like and rule out obvious alternatives. Weather, holidays, events, menu changes, hours, staffing, and local traffic can move sales at the same time.
Record the same location, weekday, daypart, open hours, and capacity before comparing periods. Keep directly matched promotion checks separate from total sales. A matched check supports a direct connection to that transaction; it still does not prove that the guest would never have come without the offer.
One Feast case shows why repeated testing matters. The restaurant’s first offer spent $500 and recorded $800 in tracked sales; the second spent $295 and recorded about $100; the third spent $674 and initially recorded $5,000, later reaching $27,000 in tracked sales. Those are tracked sales from one restaurant, not profit, causal lift, or a promise that the same offer will work elsewhere.
The useful lesson is not “copy the ribeye.” It is that one failed promotion does not settle whether your restaurant can find a workable offer. Compare all three attempts in the full restaurant case instead of remembering only the winner.
4. A strong first check, but no mature repeat result
A promotion can acquire a guest and still fail to create a regular. The return question cannot be answered on launch day.
Choose one fixed return window before launch. The denominator is unique eligible identified first-time guests who completed the initial in-person visit, matched their own paid, non-voided settled check, whose full fixed return window elapsed, and whose return outcome was reconciled and known. The numerator is the subset of that exact denominator with a separately verified return visit and its own paid, non-voided settled check. Keep early returns out until the cohort matures, and report open, immature, ineligible, and unresolved records separately.
If a second offer goes to the same guests before the first window matures, you may still drive visits, but you lose a clean read on which offer changed the behavior. Before sending it, account for what happens when the second offer overlaps the first.
Build the promotion all the way to the check before launch
A promotion is ready only when you can test its full path once: eligible guest, trackable response, identified arrival, paid non-voided settled check, correct report, and correct exclusion after a void or refund.
- Write the business problem and baseline. Name the location, daypart, weekdays, open hours, and capacity. Save completed checks or covers, average check, first-time and repeat identified guests, and settled sales for a comparable period. This record tells you what must change.
- Set the offer boundaries. Write the eligible item, required purchase, dates, hours, location, one-per-guest rule if applicable, and staff exception policy. Add the actual incremental food and beverage, packaging, payment-processing, fulfillment, and labor costs you will use in the final contribution calculation when they apply. Do not mix those costs with the offer’s face value.
- Choose one trackable response. Use a claim, reservation, code, paid order, or another record you can connect to the guest. A response creates an eligible record; it does not create a visit. If the promotion uses text, send the first message to a small internal test list and confirm delivery before the public launch.
- Connect arrival to the POS check. Decide how the host, cashier, or server marks the promotion, identifies the guest when appropriate, and reconciles the record to a paid, non-voided settled check. Keep people, parties, covers, and checks in their own columns.
- Run the positive test before cleanup. Complete one real internal test through in-person arrival and a settled check, then confirm it appears in the correct promotion report. After that success is visible, void or refund a separate test and confirm it no longer qualifies.
- Train the shift on one sentence. Staff should know who is eligible, what they receive, what purchase is required, where to mark it, and who handles an exception. If the explanation changes by server, the results will change with it.
- Launch one controlled version. Keep the offer, audience, location, and service period stable long enough to collect mature known outcomes. There is no universal number of days or dollars that makes a test conclusive.
- Follow the first visit into the return window. Keep the later message or milestone reward tied to the same identified guest, but require a separate verified return and settled check. A reply is not a return.
Capacity is part of launch readiness. One restaurant owner delayed ads because small daily spend had previously filled the restaurant beyond what one wok, four stoves, and four gas lines could handle; she put a trained kitchen assistant ahead of more demand.
Use one scorecard from response through contribution
The scorecard should preserve each event instead of collapsing everything into “redemptions” or “ROI.” Record these counts and dollars for the same location, offer, and dates:
| Result | Exact calculation | What it does and does not mean |
|---|---|---|
| Promotion responses | Unique eligible claims, reservations, purchases, or codes by source | Interest only; keep unlike response types separate |
| Completed-visit rate | Matured eligible records with reconciled, known outcomes that were verified in-person completed visits ÷ all eligible records whose full visit window elapsed and whose outcome is reconciled and known | The numerator is a subset of the denominator; report early completions, open records, and unknown outcomes separately |
| Directly matched settled sales | Net paid sales from non-voided settled checks tied to the promotion record | Sales connected to matched transactions, not profit or causal lift |
| Observed sales difference | Comparable-period settled sales minus baseline-period settled sales | A change worth investigating, not proof of cause |
| Matured repeat-visit rate | The subset of unique eligible identified first-time guests in the denominator with a separately verified return and its own paid, non-voided settled check ÷ unique eligible identified first-time guests who completed the initial in-person visit, matched their own paid, non-voided settled check, whose full fixed return window elapsed, and whose return outcome was reconciled and known | Keep early returns out until cohort maturity; report open, immature, ineligible, and unresolved records separately |
| Promotion contribution | Matched net settled sales minus the serving-cost ledger and full marketing-investment ledger | Contribution after both ledgers, before the restaurant’s full overhead and taxes; not profit |
Use two non-overlapping dollar ledgers. The serving-cost ledger contains the ordinary food and beverage cost, packaging, payment processing, fulfillment, and applicable incremental labor for the qualifying sold items. The full marketing-investment ledger contains media, creative production, food influencer compensation, usage rights, allocated software, management or agency charges, any other campaign-specific marketing expense, and actual reward cost only when that cost is not already in the serving ledger.
Allocate a shared monthly charge once using a written basis you will use again, such as the promotion’s share of the month’s active campaign days. If reward food or labor is in the serving ledger, exclude those same dollars from marketing investment. Do not subtract the face value of a discount again when net settled sales already reflect it.
Promotion contribution is matched net settled sales minus both ledgers. It can be calculated for the matched promotion checks, but it is not the restaurant’s final profit. It also does not prove incrementality: some matched guests might have come without the promotion. Keep the directly matched total beside the like-for-like observed difference so neither number pretends to answer the other question.
Read the loss before choosing the next promotion
The next move comes from the first place the guest path broke.
- Responses are weak: repair the promise, audience, or creative. If you have only tried one offer, the result belongs to that offer—not to all restaurant promotions. A study of offers across tracked campaigns and the patterns shared by stronger restaurant campaigns give you more specific questions for the next attempt.
- Responses are healthy but verified visits are weak: inspect hours, expiration, reminders, booking steps, staff recognition, and weather before buying more reach.
- Visits are healthy but contribution is weak: reduce the perk’s actual cost, protect the required purchase, move the promotion to a weaker service period, or stop it.
- Contribution is acceptable but few matured guests return: the promotion may be acquiring first visits without building a habit. Change the post-visit message, second-visit reason, or experience before repeating the same acquisition offer.
- Demand reaches capacity: stop adding traffic to that shift. Move the offer, limit claims, or fix the service constraint first.
- The records do not join: the result is insufficient evidence. Repair the claim-to-guest-to-check path and rerun the acceptance test.
Do not combine every available offer into one campaign. For a low-cost acquisition test, compare a tracked $1 promotion, a free-item signup offer, a free-soda signup offer, or a low-ticket coffee-shop promotion. For a purchase commitment or menu launch, look at a prepaid dining offer or a new surf-and-turf entree. Carry over the measurement discipline, not the headline number.
The same applies to follow-up. A short text may restart a conversation, but a reply is not a visit; one restaurant text test shows why responses and finished outcomes need separate columns.
Use food influencer videos after the offer survives the math
Food influencer videos help when the promotion is sound but your restaurant cannot produce enough local creative to test it. The owner still chooses the offer, approves the influencers, judges the files, and decides whether the economics justify another run.
Remy handles the repetitive middle: a local recruitment ad presents nearby food influencers with one written offer, interested influencers apply to those terms, and the software manages booking, texts, reminders, and file collection. It does not make a claim, reservation, post view, or delivered video equal a guest visit or restaurant revenue.
The promotion should leave you with more than activity. You should know which visit you wanted, what the guest did, which check settled, whether the guest returned after a full window, and what remained after the costs you chose to count.
If you want one system for the promotion, restaurant follow-up, and available sales matching—not just food-influencer coordination—Book a demo of Feast.