Restaurant KPIs are the handful of numbers that tell you whether last week made money, which channel it came from and what is quietly going wrong before it shows up in the bank balance. Most owners look at one number, total sales, and only when it drops. The twelve restaurant KPIs below take about twenty minutes every Monday morning, need nothing more than your order records and last week's purchase bills, and catch the slow leaks that total sales hides for months: a food cost that crept up two points, a repeat rate that is falling, an accept time that doubled on Saturdays.
Each one comes with the formula, what a normal range looks like for a small Indian restaurant or cloud kitchen, and the one action to take when it moves the wrong way.
Why Monday, and why only twelve
Monday is the first morning after your busiest two days, and the only morning when last week is complete.
Twelve is the limit because the point is to look at them every single week. A forty-row dashboard gets opened twice and abandoned. Four groups of three fit on one page, and you notice when one of them moves.
You need three inputs: last week's orders with channel, value and customer phone number; last week's purchase bills; and last week's staff cost. If your orders live in three places (a delivery app panel, a delivery app panel and a WhatsApp chat), the first job is to get them into one sheet or one order management system, because every number below needs all your orders together.
Sales numbers: 1 to 4
1. Revenue by channel
Formula: total order value, split into dine-in, direct online (WhatsApp and your store page), pickup, and each aggregator.
Total sales of ₹2.8 lakh tells you little on its own. ₹2.8 lakh with ₹1.9 lakh from the delivery apps tells you that roughly two-thirds of your revenue sits on platforms that can change their commission or your ranking next week. Watch the direct share week over week. If it is rising, your retention work is paying off; if it is flat for a quarter, nothing you are doing is moving customers off the apps.
2. Order count and average order value
Formula: number of orders; total revenue divided by number of orders.
These two explain each other. Revenue down with order count flat means the average ticket fell, which usually points to discounts or a combo that cannibalised a bigger dish. Revenue flat with orders up means you are working harder for the same money.
3. Orders per day of the week
Formula: order count for each of the seven days.
The shape matters more than the total. A kitchen that does 40 orders on Tuesday and 160 on Saturday has a staffing problem on Saturday and a marketing problem on Tuesday. The first is covered in peak hour management; the second is fixed with a weekday offer sent to your own customer list.
4. Discount given
Formula: total discounts and coupon value divided by gross revenue.
Count every rupee of discount, including the ones aggregators push you to fund. If discounts run above 10 percent of gross sales for a few weeks, you are buying orders rather than earning them, and the margin numbers below will show it.
Cost numbers: 5 to 8
5. Food cost percentage
Formula: cost of ingredients used divided by food revenue, times 100.
Use purchases adjusted for opening and closing stock rather than last week's bills alone. For a small Indian kitchen, a food cost somewhere between the high twenties and the mid thirties percent is the range most consultants work towards, with biryani and thali formats at the higher end and beverages and snacks lower. A two-point rise in one week is almost always a supplier price change or a new cook over-portioning. The food cost formula post has the worked calculation.
6. Labour cost percentage
Formula: total staff cost for the week (salary, overtime, meals) divided by revenue.
Add this to food cost and you have prime cost, the number that decides whether the restaurant survives. Prime cost above roughly 65 percent leaves too little for rent, power, packaging and you. If it is drifting up, the usual cause is staff hired for the Saturday peak and idle the rest of the week.
7. Commission and platform fees paid
Formula: aggregator commission, payment gateway charges, ads and per-order fees, in rupees and as a share of revenue.
Aggregator commission is commonly 18 to 30 percent of order value, and it never appears on a bill you pay, so most owners never add it up. Put the number in rupees next to revenue every Monday. ₹1.9 lakh of aggregator sales at 25 percent is ₹47,500 a week leaving before food cost. The commission versus direct ordering breakdown shows how that compares with a flat per-order fee on a direct channel.
8. Contribution per order by channel
Formula: for each channel, average order value minus food cost, packaging, commission or fees, and delivery cost.
This is the number that settles the "but the apps bring volume" argument. A ₹400 aggregator order might leave ₹90 after commission, food and packaging; a ₹400 direct order with a ₹10 platform fee and a ₹40 rider might leave ₹170. Once you know the contribution of each, you know what a direct customer is worth and how much you can spend to win one.
Customer numbers: 9 to 11
9. Repeat customer rate
Formula: orders from customers who have ordered before, divided by total orders.
Only possible when you have the phone number, which is one reason direct orders matter. A repeat rate that climbs week over week means the food and the experience are good enough to earn the second order, the only conversion that compounds. If it is low, start with the first-to-second-order problem before any loyalty scheme.
10. New customers added
Formula: count of new phone numbers that ordered this week.
This is the top of your funnel, and for direct channels it is also the size of the list you can message next month. If new numbers are flat while aggregator orders grow, the apps are keeping the customer relationship and you are keeping the kitchen work.
11. Rating and complaint count
Formula: average rating this week on each channel; number of complaints by type (late, wrong item, quality, missing item).
The average hides the story. Five complaints about missing items in one week is a packing-station problem you can fix by Tuesday. Keep the complaint types in a simple tally and watch for one that repeats.
Operations number: 12
12. Time to accept and time to ready
Formula: average minutes from order received to accepted; average minutes from accepted to ready, by hour of day.
Accept time is the customer's first impression and a direct driver of cancellations. Anything over two or three minutes during service means orders are not being seen. Ready time by hour shows exactly when the kitchen tips over, which tells you where a second cook or a shorter peak-hour menu would pay for itself.
The one-page Monday sheet
| # | KPI | Formula | Act when |
|---|---|---|---|
| 1 | Revenue by channel | Order value split by channel | Direct share flat for a quarter |
| 2 | Orders and AOV | Count; revenue divided by count | AOV falls two weeks running |
| 3 | Orders by weekday | Count per day | Weekday to weekend gap widens |
| 4 | Discount given | Discounts divided by gross revenue | Above 10 percent for a month |
| 5 | Food cost | Ingredients used divided by food revenue | Up two points in a week |
| 6 | Labour cost | Staff cost divided by revenue | Prime cost (5 plus 6) above about 65 percent |
| 7 | Commission and fees | All platform and gateway charges, in rupees | Growing faster than revenue |
| 8 | Contribution per order | AOV minus food, packaging, fees, delivery | Any channel near zero |
| 9 | Repeat rate | Repeat orders divided by total orders | Falling three weeks running |
| 10 | New customers | New phone numbers this week | Flat while orders grow |
| 11 | Rating and complaints | Average rating; complaint tally by type | Any complaint type repeats |
| 12 | Accept and ready time | Minutes, by hour of day | Accept time above three minutes |
Print it and fill it by hand for the first month if you have to; the habit matters more than the tool. Once your direct orders come through one system, most of these fill themselves. A live dashboard that shows orders, revenue and customer insights, like the one in Tezzu, gives you numbers 1, 2, 3, 9, 10 and 12 without a spreadsheet, and the dashboard guide shows where each sits. Food and labour cost still come from your own bills, and always will.
See which numbers fill themselves
If your direct orders already come through WhatsApp or a store page, a short demo will show which of the twelve a live dashboard can track for you.
Book a demoFrequently asked questions
What are the most important KPIs for a restaurant?
If you only track three, make them prime cost (food cost plus labour cost as a share of revenue), revenue by channel, and repeat customer rate. Prime cost tells you whether the business can survive, channel split tells you how exposed you are to aggregators, and repeat rate tells you whether customers like you enough to come back without a discount.
What is a good food cost percentage for a restaurant in India?
Most consultants work towards a food cost somewhere between the high twenties and the mid thirties percent, depending on format. Biryani, thali and meat-heavy menus sit at the upper end; beverages, snacks and bakery items sit lower. The exact figure matters less than the trend: a two-point jump in a single week nearly always has a specific cause you can find and fix.
How often should a restaurant review its KPIs?
Weekly, on the same morning, for the operational numbers: sales by channel, accept and ready times, complaints, discounts. Monthly for food and labour cost, because stock counts and salaries are monthly events. Daily checking creates noise and panic over one slow Tuesday; monthly checking lets a problem run for four weeks before you see it.
How do I track restaurant KPIs without a POS?
A spreadsheet with one row per order (date, channel, value, phone number, status times) is enough to start. Aggregator panels export order lists. For direct orders, an ordering system that records every WhatsApp and store order with its timestamps gives you most of the operational numbers automatically. Food and labour cost always come from purchase bills and salary records, with or without a POS.