Operators love discounts because they show up as immediate top-line lift. Sales go up in the promotion window, the marketing calendar has something to show, and the deck looks fine at the quarterly review. The problem is that top-line lift is not the number to watch. Margin dollars are. And in almost every discount post-mortem I have run, margin dollars were flat or down while the sales number told a story of success.
This is not a marketing team problem. It is a math problem. And once you see the math clearly, the rules for which discounts to run and which to kill become obvious.
The math nobody runs before hitting go
Take a $16 entree with a 30 percent food cost. That is $4.80 in food, $11.20 in gross margin. Now run 20 percent off.
Original price: $16.00 Food cost: $ 4.80 (30% food cost) Gross margin: $11.20 (70% margin) Discounted price: $12.80 (20% off retail) Food cost: $ 4.80 (food cost does not move) Gross margin: $ 8.00 (62.5% margin) Margin dollars lost: $ 3.20 per cover (29% of original margin dollars)
The guest sees 20 percent off. The P&L takes a 29 percent hit to margin on that cover. Now factor in that food cost on many discounted items runs 32 to 38 percent because they are the popular ones. On a 35 percent food-cost item, the same 20 percent discount takes margin from 65 down to 56, a 14-point drop, which is a 21 percent hit to margin dollars. And that is before you account for the additional labor to serve the incremental cover, which does not scale to zero.
Every discount is a much larger discount than it looks. If you do not know how much larger before you launch, you will find out the hard way at close.
The break-even math you owe the promotion
The right question for any discount is not "will sales go up?" It is "how many additional covers do I need to break even on margin dollars, and is that plausible?"
For the $16 example above, the promotion loses $3.20 in margin on every covered cover. To break even on total margin dollars, you need incremental covers to make up that gap. If baseline was 200 covers on the promoted item and you gave $3.20 back on each of them, that is $640 in margin left on the table. At $8.00 margin per incremental cover, you need 80 incremental covers to break even. That is a 40 percent incremental lift on that item. Very rarely happens.
This is why the standard blanket percent-off promotion loses money almost every time. It cannibalizes covers that would have bought at full price, gives them a discount, and does not drive enough new demand to make it back.
Fig. 1 · A 20% discount is a 29% margin hit. A 30% discount is a 43% margin hit.
The three discount structures that actually work
1. Bundle discounts
Bundle a low-margin item with a high-margin add-on and price the bundle just below the sum of the parts. The guest sees a discount. The operator sees an attach rate lift on the high-margin item that pays for the entree concession.
Example that worked in one of my Bay Area units: $22 entree paired with a $9 wine pour for a $27 combo. Retail sum $31, "discount" $4. Attach rate on wine at full price ran 32 percent. In the bundle it ran 71 percent. The wine margin at 78 percent covered the $4 concession twice over. Contribution per cover on the bundle beat the standalone entree by $1.60.
Bundles work because they change mix, not price. The guest is not being trained to expect a lower entree price. They are being nudged into a bigger check.
2. Time-shift promotions
Move demand from a full daypart into a slow one. Discount only on Tuesday. Only 3 to 5 PM. Only pre-theater 5 to 6 PM. The labor is already scheduled for a shift you are staffed for. The incremental cover carries almost pure contribution because the fixed cost is already sunk.
The math flips when the daypart is slow enough. If your kitchen is running at 40 percent capacity from 3 to 5 PM with a full crew on the line, an incremental cover at even 50 percent margin contributes almost 50 percent of its check to the bottom line, because the labor cost is already committed. Time-shift discounts on peak Friday dinner make no sense. Time-shift discounts into empty Tuesday afternoon almost always do.
3. Category-push promotions
Discount high-margin add-ons to lift attach rate. A $2 side upgrade instead of a $4 upgrade. A free dessert with entree if the dessert margin is 80 percent and it drives a review, a repeat visit, or a photo. The math works when the discounted item has a very high standalone margin.
Beverages are the canonical case. If beer margin is 78 percent and cocktail margin is 82, a $2 off cocktail promo during Wednesday dinner lifts attach rate materially and still holds margin dollars per cover.
The two discount structures that almost never work
Across-the-board percent off
The 20 percent off everything Groupon-style deal. Cannibalizes regulars. Attracts one-time deal seekers who do not return at full price. Trains the market to wait for the next promo. I have seen this destroy margin dollars in every case I have watched it run, including a case where I recommended running it in year one and had to reverse the recommendation by month three.
Standing weekly coupons
The Tuesday coupon that never ends. Two weeks in, it is a promotion. Two months in, it is pricing. The guest budget for that visit is now $2 lower forever. You have permanently repriced the item and given up all promotional lift, because guests planning around a permanent discount are not making an incremental visit, they are just paying less for the visit they were going to make anyway.
Every promotion needs an end date at the start. If you cannot end it, you are not running a promotion, you are repricing.
The third-party delivery problem
Delivery platforms push operators to stack promotions on top of their commission structure. Do the math before saying yes.
$16 entree on delivery: Retail: $16.00 Delivery commission: $ 4.00 (25%) Food cost: $ 4.80 (30%) Packaging: $ 0.60 Contribution: $ 6.60 (41%) Same entree with "spend $15, get $5 off" promo funded by operator: Retail: $16.00 Guest pays: $11.00 Delivery commission: $ 4.00 (still on gross) Food cost: $ 4.80 Packaging: $ 0.60 Contribution: $ 1.60 (10%)
A single stacked promo cuts contribution by roughly 75 percent. If the incremental lift does not more than quadruple, it loses money. Sometimes the guest acquisition play still makes it worth doing, but only if you can measure conversion to non-promo orders within 60 days. If you cannot measure that, do not run stacked delivery promos.
The post-promotion review nobody does
Every promotion needs a 30-day post-mortem with actual numbers, not the top-line story. The template I use is short:
- Baseline period: Same days of week, same daypart, 4 weeks pre-promo.
- Promo period: Same days of week, same daypart, promo window.
- Incremental cover count: Promo covers minus baseline covers.
- Contribution per cover: Promo period vs baseline.
- Total contribution dollars: Promo vs baseline.
- Verdict: Renew, kill, or restructure.
Publish it to the general manager team every quarter. If you do not run this review, every promo gets renewed by inertia, and inertia has never once made anyone money.
Building a promotional calendar with discipline
The way to institutionalize this is a written promotional calendar that runs a quarter ahead. One page. Every promotion listed with the same six columns:
- Promotion name and dates (start and end, both required).
- Structure (bundle, time-shift, category push, or specific).
- Target daypart or channel.
- Baseline covers or check average for the period.
- Break-even incremental covers or check average lift.
- Owner (which manager is responsible for measuring the result).
Post the calendar in the office. Every promotion pitched by the marketing team, the chef, or a franchise field rep has to fit on that page. If it does not have an end date, it is not a promotion. If it does not have a break-even number, it is not approved. If it does not have an owner, nobody will measure it and it will renew forever.
This discipline sounds heavy for a small operator. It is actually lighter than the alternative, which is a promotional calendar of five overlapping deals that nobody can trace to a margin outcome and that quietly costs the P&L 1 to 2 points a year.
Loyalty discounts as a separate category
Loyalty programs deserve their own treatment because the math is different. A loyalty program is not a promotion. It is a permanent structural discount in exchange for repeat visit frequency. The math question is not "did the promotion pay for itself" but "did the incremental visits from loyal guests, over their tenure, cover the ongoing cost."
The right benchmark for a loyalty program is 8 to 12 percent effective discount rate on loyalty-attached sales, offset by a 15 to 25 percent visit frequency lift for enrolled guests. If your effective discount rate is above 15 percent or your frequency lift is below 10 percent, the program is upside down. This is the review nobody runs and it is worth doing annually.
What I got wrong
I lost about $18k in contribution on one campaign because I approved a "buy one get one" on our best-selling entree in a Zareen's location without running the math. The entree was 32 percent food cost, gross margin 68 percent, so the BOGO cut margin per two-cover order by half. Sales for the item went up 62 percent in the promo window. Contribution dollars for the item went down 24 percent. I renewed it for a second month before I saw the number because I was looking at the top line at the daily flash and not the margin.
The lesson: never approve a promo without the break-even cover count written on the same page as the pitch. If the number is not there, the pitch is incomplete.
The point
Discounts are not free marketing. They are a spend, and the spend comes directly out of margin. The operators who run discounts well are the ones who treat every promotion like a small P&L project: baseline, hypothesis, math, launch, measure, kill or renew. The ones who run them badly are the ones who confuse a top-line lift with a bottom-line win.
If you take one thing from this, take the break-even cover count. Compute it before you launch. Say the number out loud. If it does not sound plausible, do not run the promo. That single discipline will save more margin than any pricing model you can buy.