The percent-of-sales P&L is the language every operator learns first, and it is not wrong. Food cost at 30 percent, labor at 32 percent, prime cost at 62 percent. Those numbers do a lot of work. But they are ratios, and ratios can hide the actual thing that is happening in the restaurant, especially when either half of the ratio is moving.
Per-cover unit economics fix that. Instead of dividing a cost line by a dollar figure, you divide it by a guest count. The number that comes out is a dollar amount per real human who walked through the door. It is harder for that number to lie to you.
I started running per-cover numbers next to the percentages at Zareen's about six months into the turnaround, after the percent-of-sales lines told us we had fixed the food cost line and the actual walk-in inventory told us we had not. The percentages were flat because we had raised prices. The per-cover food cost was climbing quietly the whole time. Once we started reading both, the drift stopped hiding.
Why percentages hide drift
A percentage is a ratio: cost divided by sales. Both halves can move. When both halves move together, the percentage stays flat even though something operational changed.
Two examples I have watched play out live:
The price-lift illusion
You raise menu prices 6 percent in January. Food cost dollars stay flat because ingredient costs did not change that month. Food cost as a percent of sales drops from 32 percent to about 30 percent. Everyone on the leadership team calls it a win. But six months later a supplier increase quietly took cost per plate up 8 percent, and by then the percentage crept back to 32 while nobody looked at cost per cover. The price lift bought you nothing. It hid a decay.
The mix-shift blur
Ticket average climbs from $22 to $26 because guests started ordering more cocktails. Labor cost as a percent of sales drops from 32 percent to about 28 percent because the labor dollars did not change and the sales dollars went up. Looks like a labor productivity win. It is not. Labor minutes per cover are exactly the same. You did not get more productive. You just started selling more per guest. The moment cocktail sales normalize, the labor percentage snaps back and nobody understands why.
Percent-of-sales lines tell you where you ended up. Per-cover lines tell you what actually happened. Both are useful. Only one is a diagnostic.
The four per-cover numbers to track
These are the four I put in every Monday P&L review, next to the standard percent-of-sales lines. Not a replacement. A companion set.
Fig. 1 · The per-cover stack, healthy full-service casual.
1. Sales per cover
Net sales divided by the number of guests served in the same period.
Not the same as average ticket. Average ticket is divided by number of checks. In a full-service restaurant a check often covers two or three guests, so a $60 average ticket at a two-top is $30 sales per cover. Average ticket can look strong while sales per cover softens, which is what happens when parties shrink but the operator does not notice because the ticket line looks fine.
Healthy ranges:
- Full-service casual: $20 to $40
- Fast-casual: $12 to $22
- QSR: $8 to $15
2. Food cost per cover
Total food cost dollars divided by covers.
This is the one that catches ingredient inflation, portion drift, and spec deviation. When your cost per plate goes up because the produce distributor pushed through a quiet 4 percent increase, the food cost percentage might not move, but food cost per cover will. Track it weekly and any three-week rising trend is real signal.
Healthy ranges:
- Full-service casual: $6 to $10 per cover
- Fast-casual: $4 to $7 per cover
- QSR: $2.50 to $4.50 per cover
3. Labor minutes per cover
Total labor hours worked multiplied by 60, divided by covers.
This one strips out wage rate. A Bay Area restaurant paying $22 an hour and a Texas restaurant paying $14 an hour cannot be compared on labor cost per cover, because the wage rates are different. They can be compared on labor minutes per cover, because the productivity is the same measurement in both markets.
Healthy ranges:
- Full-service casual: 12 to 20 minutes per cover
- Fast-casual: 6 to 12 minutes per cover
- QSR: 3 to 7 minutes per cover
If labor minutes per cover is rising and labor cost as a percent of sales is flat, the schedule is degrading and only a favorable mix shift is masking it. The mix shift will eventually reverse and the labor line will look terrible in a single week for no obvious reason.
4. Contribution per cover
Sales per cover minus food cost per cover minus labor cost per cover. Not labor minutes here, labor dollars.
Contribution per cover is the dollar amount each guest leaves you with to pay everything else: rent, utilities, insurance, R&M, credit card fees, and profit. It is the number the operator owns most directly, and it is the honest scoreboard.
Healthy ranges:
- Full-service casual: $10 to $16 per cover
- Fast-casual: $6 to $10 per cover
- QSR: $3 to $6 per cover
The critical comparison: contribution per cover against rent per cover. Rent per cover is monthly rent divided by monthly covers. If contribution per cover is lower than rent per cover, the unit is losing money on every guest, and adding volume will not fix it. That is a menu, pricing, or format problem, not a marketing problem.
Where per-cover reveals leaks the percentages miss
Three specific patterns show up in almost every unit that has been running on percentages alone. All three are hard to see any other way.
Portion drift under a price increase
You raised prices. Food cost percentage looks fine. But the line cooks have been over-portioning to keep the plate looking generous relative to the higher price. Food cost per cover is up 8 percent even though food cost percentage is unchanged. This shows up in every unit that raises prices without retraining the line on portion specs. The fix is not the price. The fix is the training.
Labor productivity buried under a cocktail mix shift
Cocktail sales pop for a quarter and labor as a percent of sales drops 3 points. The team celebrates. Labor minutes per cover show that the actual productivity did not change. When cocktails normalize, the labor percentage snaps back to trend and the general manager is confused. They should not be. Labor minutes told them the truth all along.
Menu mix pulling down contribution per cover
You added three new low-margin items and they became the top sellers. Percent-of-sales food cost creeps up 1.5 points, which nobody escalates because it looks like noise. But sales per cover dropped $1.20 (guests substituting into the new items) while food cost per cover held steady, so contribution per cover fell $1.20 per guest. On a unit doing 800 covers a week that is $960 a week in lost contribution, roughly $50K a year, hiding inside a "food cost is slightly up" story.
A worked example
Take a full-service casual unit doing $220K a week in sales at $28 sales per cover. That is about 7,857 covers a week. Food cost is 30 percent, or $66K a week, which is $8.40 per cover. Labor cost is 28 percent, or $61.6K a week, which is $7.84 per cover, roughly 15 labor minutes per cover on a $31/hour blended rate.
Week 1 baseline: Sales per cover: $28.00 Food cost per cover: $8.40 (30.0% of sales) Labor cost per cover: $7.84 (28.0% of sales) Contribution per cover: $11.76 (42.0% of sales) Week 12 (12 weeks later, percentages unchanged): Sales per cover: $28.00 Food cost per cover: $9.05 (32.3% of sales) → drift Labor cost per cover: $7.90 (28.2% of sales) Contribution per cover: $11.05 (39.5% of sales) Per-cover contribution drop: $0.71/cover Weekly impact at 7,857 covers: $5,578 Annualized: ~$290,000
The percent-of-sales food cost line moved by only 2.3 points, which many operators would round to "roughly flat." The per-cover contribution moved by $0.71, which on a group doing 400K covers a year across five units is roughly $280K to $300K in lost annual contribution. Same data. One view sees the leak, the other does not.
The variance conversation this makes possible
The most useful thing about running per-cover next to percent-of-sales is the conversation it opens with the general manager on Monday morning. The percent lines invite defensive answers ("food cost was a little high, we had a bad prep day"). The per-cover lines invite specific ones, because the number is a dollar per guest.
Compare the two conversations:
Percent-only version: "Food cost was 31.5 percent last week, versus 30 percent the week before." "Yeah, we had a big waste day on Thursday." Discussion over. Nothing changes.
Per-cover version: "Food cost per cover was $9.10 last week, versus $8.40 the week before. That is a $0.70 gap on 1,600 covers, so $1,120 for the week." "Where did the $0.70 land?" That is a different conversation, and it always ends in a specific answer. Portion drift on the pasta station, spec change on the salmon, a supplier increase on cheese, or a genuine waste event that will not repeat. Every one of those has a different fix.
Per-cover numbers change the Monday conversation from a defensive review of what went wrong to a shared analysis of where the dollars went. That shift is more valuable than the numbers themselves.
How to run this weekly
The whole system is one row of data pulled from the POS and the accounting system:
- Every Monday morning, pull last week's total net sales, food cost dollars, labor hours, labor cost dollars, and total covers from the POS.
- Divide the four cost lines by covers. Write the four per-cover numbers on the whiteboard next to the four percentages.
- Compare to the last four weeks. Note anything moving more than $0.30 per cover in either direction on food or contribution.
- Root cause the movement before the meeting ends. Do not defer.
It takes ten minutes once the system is running. The first three weeks are slow because you are calibrating what "normal" looks like at your specific unit. After that it is background hygiene, and the drift stops hiding.
The point
Percentages are ratios and ratios blur. Per-cover numbers are dollars per guest and guests are real. Sales per cover, food cost per cover, labor minutes per cover, and contribution per cover give you a diagnostic view that survives price changes, mix shifts, and menu changes without needing recalibration.
Contribution per cover, tracked weekly against rent per cover, is the honest scoreboard for whether the unit economics work. If it is above rent per cover, you have a business. If it is below, you have a hobby. Percentages will not tell you which one you are running. Per-cover numbers will.