Most restaurant dashboards fail for the same reason most checklists do: they are too long. Forty metrics on a screen is an achievement to build and a chore to read, and within a month nobody reads it.
The list below is twelve numbers. Six of them are worth a weekly look, six a monthly one. Every one of them, if it moves, tells you to do something specific — which is the only test a KPI needs to pass.
Why a fixed review beats a live dashboard
The value is not in the numbers. It is in looking at them on a schedule, before they become a problem you can feel. The habit that makes this stick, and how to turn a review into a decision, is covered in restaurant analytics.
Food cost drifting from 30% to 34% over two quarters is invisible on the floor. Covers falling 8% on Wednesdays is invisible when Saturday is packed. Both are obvious in a weekly sheet and both are cheap to fix early.
So: one fixed day, thirty minutes, same format every week. Monday morning works well — the weekend is complete and you have five days to react.
The six weekly numbers
1. Covers
How: count of guests served, by day and by day-part.
The base rate everything else is measured against. Sales without covers is ambiguous — revenue can rise while your restaurant is emptying, if your prices went up.
Look at the pattern, not the total. One weak weekday, repeated three weeks running, is a real signal.
2. Average order value (AOV)
How: total sales ÷ number of bills. Also worth it per head: sales ÷ covers.
Falling AOV with flat covers almost always means beverages, desserts or higher-margin items have stopped selling — often because the floor stopped suggesting them, not because guests stopped wanting them.
Track it separately for lunch and dinner. A single blended figure hides most of what is interesting.
3. Table turnover
How: covers served ÷ seats available, per service. A 40-seat restaurant serving 100 dinner covers turned 2.5 times.
The number that tells you whether a busy-feeling night was actually productive. Turnover falls for specific reasons — slow ordering, slow kitchen, a long wait for the bill — and each has its own fix.
4. Labour cost percentage
How: total labour cost (salaries, wages, overtime, staff meals) ÷ sales × 100.
Most Indian full-service restaurants run 18–28%. Weekly matters because it catches overtime creeping in — a roster that quietly needs two extra shifts a week costs real money and rarely gets noticed until payroll.
5. Sales by day-part
How: sales and covers split into lunch, evening and dinner.
This is where staffing decisions live. A restaurant staffing evenly for a floor that does 30% of its business at lunch and 70% at dinner is overstaffed at noon and underwater at nine.
6. Top and bottom five items
How: units sold per dish for the week.
Weekly, because it moves. A best-seller falling out of the top five is worth a question — did the recipe change, did a cook leave, did it move down the menu? A permanent bottom-five dish is stock you carry, prep you do, and menu space you are spending on nobody.
The six monthly numbers
7. Food cost percentage
How: (opening stock + purchases − closing stock) ÷ food sales × 100, on a real stock count.
The single largest controllable expense in most restaurants, and the one that drifts silently. It has its own full guide, including why the recipe-sheet version of this number is always optimistic.
8. Prime cost
How: (food and beverage cost + total labour cost) ÷ sales × 100.
If you keep one number, keep this one. Under 60–65% is the usual working target for an Indian full-service restaurant.
It matters because the two halves trade against each other. Cutting staff pushes food cost up through wastage and mistakes; cheap ingredients cost you in remakes and lost guests. Prime cost is the only figure that sees both at once.
9. Beverage cost percentage
How: beverage cost ÷ beverage sales × 100. Kept separate from food, always.
Beverage cost runs far below food cost, so blending the two hides a food problem behind a healthy bar. Bar variance is also where over-pouring and unrecorded wastage show up first.
10. Wastage as a share of food cost
How: value of recorded wastage ÷ total food cost × 100, from a wastage log.
Requires a clipboard by the bin and a month of honest entries. Nearly always finds two or three specific items responsible for most of the money — a prep yield assumption that was never true, or one dish that gets sent back.
11. Repeat guest share
How: the proportion of covers from guests who have been before, however roughly you can estimate it.
Hard to measure precisely in a small restaurant and worth a rough figure anyway, because it is the difference between a business and a crowd. What moves it is covered in how to increase repeat customers.
12. Break-even covers
How: total fixed costs ÷ contribution margin per cover, where contribution margin is average spend per head minus variable cost per head.
Calculate it once a quarter. Knowing that you need 62 covers a night to break even changes how a slow Tuesday feels — and turns "should we open for lunch?" into arithmetic instead of an argument.
About the benchmark figures above. The ranges quoted for prime cost, labour cost and food cost are conventional industry working ranges rather than findings from a published survey of Indian restaurants. They are useful as a sanity check and useless as a target — your own trend over eight weeks tells you more than any of them. Anything here bearing on tax or statutory reporting is general information, not financial advice; check it with your accountant. Last reviewed: 31 July 2026.
A worked weekly review
Twenty minutes, one sheet. A 40-seat restaurant, week ending Sunday:
| Metric | This week | Last week | 4-week avg | Read |
|---|---|---|---|---|
| Covers | 742 | 771 | 758 | Flat |
| AOV per head | ₹612 | ₹648 | ₹641 | Down 5% |
| Turnover (dinner) | 2.3 | 2.4 | 2.35 | Flat |
| Labour cost % | 24.1% | 22.8% | 23.0% | Up |
| Lunch share of sales | 28% | 31% | 30% | Slightly down |
| Top item units | 186 | 190 | 188 | Flat |
Two numbers moved. AOV is down 5% on flat covers — check beverage attachment and whether the floor is still suggesting starters. And that is also most of the labour story: covers held up, but spend per head did not, so the week's sales came in around 9% below last week's while the roster stayed the same size. Labour rising 1.3 points is the arithmetic of a smaller denominator, not of overtime — check the roster anyway, but fix the AOV first.
That is the whole exercise. Not forty metrics; two questions, asked while there is still time for the answer to matter.
Building it in one week
- Pick your day and defend it. Same day, same thirty minutes, every week.
- Build one sheet with the six weekly numbers as rows and weeks as columns. A spreadsheet is entirely adequate.
- Backfill four weeks from your existing sales data, so week one already has a comparison.
- Add the monthly six to the same sheet, filled on stock-count day.
- Write your normal ranges down once you have eight weeks. A number is only useful against an expectation.
- Every review, act on the two that moved most — and write down what you did, so next month's review can tell you whether it worked.
Summary
- Twelve numbers is enough: six weekly, six monthly, on a fixed day.
- Prime cost is the single most valuable figure, because food and labour trade against each other and only prime cost sees both.
- Covers and average order value together tell you what sales alone cannot.
- Compare against your own last four weeks and the same week last year; outside benchmarks are context, not targets.
- Act on the two numbers that moved, record what you did, and check it next month.
- A KPI you never act on costs the same time as one you do and buys nothing but the feeling of control.
