In this article +
The short answer
Restaurant demand forecasting estimates how many of each dish you'll sell on a given day, so you can prep and buy the right amount. The simplest reliable method is the same-weekday average: average the last four same weekdays, then adjust for known events (festivals, promotions, weather, matches). Check your forecast against actual sales every week and track the error. Software helps when you have many dishes, outlets or channels.
Over-prep and you throw food away. Under-prep and you say "sorry, finished" at 8:30 pm, which is how you lose a regular. Forecasting is how you find the middle.
What drives restaurant demand
| Driver | Example | How big an effect |
|---|---|---|
| Day of week | Saturdays busy, Mondays slow | Usually the biggest |
| Time of day | Lunch rush, dinner peak | Shapes prep timing |
| Seasonality | Summer drinks, monsoon pakoras, winter soups | Shifts the dish mix |
| Festivals and holidays | Diwali, Eid, Onam, long weekends | Can double or halve demand |
| Promotions | A combo offer or aggregator discount | Spikes specific dishes |
| Weather | Heavy rain: dine-in down, delivery up | Shifts channels |
| Local events | Cricket match, exams, an office party nearby | Hard to predict, easy to note |
| Pay cycles | First week of the month | Higher average bill |
Method 1: Same-weekday average (start here)
For each dish, take the last four same weekdays and average them.
| Dal makhani, Tuesdays | Plates |
|---|---|
| 4 weeks ago | 38 |
| 3 weeks ago | 42 |
| 2 weeks ago | 35 |
| Last week | 41 |
| Forecast for next Tuesday | 39 |
Illustrative numbers.
That's it. It beats "same as yesterday" because Tuesday is not Monday, and it beats gut feel because guts remember the busy Tuesday, not the average one.
Method 2: Adjust for what you know
Then nudge the forecast for known events:
- Festival or long weekend: check what happened last year on the same occasion.
- Promotion: a combo on a dish typically lifts that dish and its partner.
- Weather: rain forecast tomorrow? Shift prep from dine-in favourites to delivery favourites.
- Local event: a big match at 7:30 pm? Expect an early dinner rush, then silence.
The cricket coefficient
India vs Pakistan at 7:30 pm: dine-in drops, delivery of anything fried rises. Nobody has published the exact multiplier. Every Indian restaurant owner knows it in their bones.
Forecast vs actual: a sample week
How to measure forecast accuracy
Each week, compare forecast to actual for your top dishes and calculate the average percentage error:
Error % = |Actual − Forecast| ÷ Actual × 100
| Day | Forecast | Actual | Error % |
|---|---|---|---|
| Mon | 31 | 28 | 10.7% |
| Tue | 39 | 41 | 4.9% |
| Wed | 40 | 37 | 8.1% |
| Thu | 44 | 46 | 4.3% |
| Fri | 58 | 63 | 7.9% |
| Sat | 72 | 70 | 2.9% |
| Sun | 66 | 61 | 8.2% |
| Average | 6.7% |
Sample data. Track this number over time. If it's getting smaller, your forecasting is improving. If one day is always off, look for a pattern (maybe Sundays are being over-forecast because of a one-off event in the four-week window).
Turning a forecast into prep and purchasing
- Forecast plates per dish for tomorrow.
- Multiply by the recipe: 39 plates × 120 g paneer = 4.7 kg paneer.
- Add a buffer for your highest-variance dishes, maybe 10%.
- Subtract what's already prepped or in stock.
- Prep and order the difference.
This is where forecasting pays: less over-production (the biggest waste category in many kitchens, see our food wastage guide), fewer stock-outs, and smaller, more frequent orders.
When to use software
A notebook works for 10 dishes in one outlet. It breaks when you have:
- 80+ dishes
- Several outlets
- Dine-in, takeaway and delivery with different patterns
- Promotions running regularly
Software can calculate same-weekday averages automatically, adjust for patterns, and show forecast against actual without anyone doing arithmetic.
How Ecomeal helps today
Ecomeal's dashboard shows sales, orders and average bill against the same time yesterday, revenue split by morning, afternoon and evening, your peak hour and most-ordered times, and dine-in versus takeaway. That's the raw material for prep planning. AI Up Sell uses the last seven days of dish sales to find slow movers and put a deal on the POS, which helps clear what was over-prepped.
Dish-level demand forecasting for prep and purchasing is on our roadmap, not live today.
Frequently asked questions
What is demand forecasting in a restaurant?
Estimating how much of each dish you'll sell on a future day, so you can plan prep, staff and purchasing.
What's the easiest way to forecast restaurant sales?
Average the last four same weekdays for each dish, then adjust for known events like festivals, promotions and weather.
How accurate should a restaurant forecast be?
It varies by dish and volume. Track the average percentage error weekly and aim to bring it down over time rather than hitting a fixed number.
Does weather affect restaurant sales in India?
Often, yes. Heavy rain commonly shifts demand from dine-in to delivery. Your own sales history is the best guide to how much.
See it in your Kitchen.
Ecomeal Lite starts at ₹333/month + GST, billed yearly. A short demo on WhatsApp shows you billing, KOTs and AI Up Sell on your own menu.