Key takeaways
- A sales forecast is an estimate of covers and sales for each day and meal period. You use it to decide staff hours, prep quantities and orders.
- Start with a day-of-week baseline from your POS history, the average of the same weekday over the last four to eight weeks.
- Adjust for public holidays, school holidays, local events, weather and promotions, and write down why you adjusted.
- Compare forecast to actual every week. A simple method you check and improve beats a complex one you never review.
Every restaurant forecasts, even if it does not call it that. The chef who preps extra rendang on Friday, or the manager who adds a server on Saturday night, is making a forecast. The problem is that gut-feel forecasts are hard to check and hard to hand over. When the owner is away, the kitchen either over-preps and throws food away, or runs out at 8pm. This guide shows a simple, repeatable way to forecast covers and sales from your POS history, how to adjust for Singapore holidays, events and weather, and how to turn the forecast into rosters and prep lists. You need no special software, only clean POS data and a spreadsheet.
What a sales forecast is for
A forecast is useful only if it changes a decision. In a restaurant, three decisions depend on it:
- Staff hours. How many servers, cooks and dishwashers you schedule for each shift. Labour is one of your largest costs, and our labour cost guide shows why getting it right matters.
- Prep quantities. How much of each item the kitchen prepares before service. Too much becomes waste; too little means 86’d dishes and unhappy customers.
- Ordering. How much you buy from suppliers for the next delivery, which ties up cash and chiller space.
Forecast at the level those decisions need. For rosters, forecast covers by meal period (lunch, afternoon, dinner, supper). For prep, forecast sales of your top items. You rarely need a forecast by the hour unless you run a queue-heavy lunch trade.
The data you need from your POS
Your POS already has most of what you need. Pull these reports for at least the last eight weeks, and ideally a full year:
- Sales and covers by day and meal period. If you do not record covers, use the number of bills as a proxy.
- Item sales by day. At least for your top 20 items and any item with a long prep time.
- Channel split. Dine-in, takeaway and delivery, because they peak at different times.
- Sales by hour. Useful for splitting shifts.
Our POS reports guide explains what each report shows. ChaChaCha includes reports, and the ERP gives head office a view across outlets. Clean the data before you use it: mark days when you closed early, had a private event or ran a big promotion, so they do not distort the average.
A simple method: the day-of-week baseline
Most restaurants have a strong weekly pattern. Monday lunch looks like other Monday lunches more than it looks like Saturday lunch. So the simplest useful forecast is:
- Take the same weekday and meal period for the last four to eight weeks (for example, the last six Friday dinners).
- Remove unusual days, such as a public holiday or a night you closed early.
- Average the rest. That is your baseline for next Friday dinner.
- If sales are clearly rising or falling, weight recent weeks more (for example, count the last two weeks twice).
Here is how a weekly forecast sheet might look. The numbers are an illustration, not benchmarks.
| Day | Baseline dinner covers | Adjustment | Forecast | Reason |
|---|---|---|---|---|
| Monday | 60 | 0 | 60 | Normal week |
| Wednesday | 75 | +15 | 90 | Eve of public holiday |
| Thursday | 80 | -20 | 60 | Public holiday, office trade away |
| Saturday | 140 | +10 | 150 | Concert nearby |
Always write the reason for an adjustment. After a few months, those notes tell you how big each effect really is at your outlet.
Adjusting for public holidays and events
Holidays change trade a lot, and not always in the same direction. A CBD lunch spot loses most of its trade on a public holiday, while a family restaurant in a heartland mall may have its busiest day. Look up how your outlet traded on the same holiday last year, and on the eve of it.
MOM publishes the public holiday dates. For 2027, it lists New Year’s Day on Friday 1 January, Chinese New Year on 6 and 7 February (a Saturday and Sunday), Hari Raya Puasa on 10 March, Good Friday on 26 March, Labour Day on Saturday 1 May, Hari Raya Haji on 17 May, Vesak Day on 20 May, National Day on Monday 9 August, Deepavali on 28 October and Christmas Day on Saturday 25 December. MOM notes that Monday 8 February 2027 will be a public holiday for employees whose rest day falls on 7 February. Put these into your forecast sheet now, along with long weekends.
Other events to note:
- School holidays, which change family dining and weekday lunch patterns.
- Concerts, sports and exhibitions near your outlet.
- Festive seasons, such as Chinese New Year reunion dinners and year-end parties. Our festive season guide covers planning for them.
- Your own promotions and any delivery platform campaigns.
- Paydays, if your outlet serves office workers.
Weather and seasonal effects
Rain affects trade in Singapore, especially for outlets with outdoor seating, open-air hawker centres and delivery. Heavy rain can push dine-in down and delivery up. The Meteorological Service Singapore weather portal publishes a 2-hour nowcast, a 24-hour forecast, a 4-day outlook, a fortnightly outlook and monsoon updates, plus historical daily records.
Use weather in two ways:
- Short term. Check the 24-hour forecast and 4-day outlook when you finalise the next day’s roster and prep. If heavy rain is expected, move a server from the terrace to packing delivery orders.
- Looking back. Note the weather in your forecast sheet each day. After a few months, compare wet days with dry days of the same weekday to see how much weather moves your sales.
Keep it simple. Weather is one adjustment among several, not a model in itself.
From covers to rosters
Once you have forecast covers for each meal period, convert them to staff hours using a ratio that fits your outlet, such as covers per server per hour. Work it out from your own good nights: the nights when service was smooth and staff were busy but not overwhelmed. Then:
- Set a minimum crew for each shift, however quiet (for example, one cook, one server and one cashier).
- Add staff in steps as forecast covers pass each threshold.
- Stagger start times so staff arrive just before the rush, not at opening.
- Keep one or two part-timers on call for busy days you could not predict.
Rosters must also follow the law. For employees covered by Part IV of the Employment Act, MOM’s hours of work page sets a maximum of 12 hours a day, overtime capped at 72 hours a month and paid at least 1.5 times the hourly basic rate, and one rest day a week. Our F&B roster guide goes through building the roster itself.
From covers to prep lists and ordering
For the kitchen, forecast covers turn into prep quantities through item mix. If 20% of dinner guests usually order your signature curry, and you forecast 150 covers, plan for about 30 portions plus a small buffer.
- Work out each key item’s share of covers from the last four weeks of item sales.
- Multiply by forecast covers to get portions needed.
- Subtract what is already prepped and in date.
- Add a buffer for items that are slow to prepare, and keep it small for items that spoil quickly.
Our prep list template shows how to lay this out for each station. Food safety limits apply too. SFA’s December 2024 advisory says ready-to-eat food should not be kept between 5°C and 60°C for more than four hours, and reminds operators to manage their business so they can cope with customer orders. Better forecasts help on both counts.
For ordering, ChaChaCha’s recipe-level stock deduction deducts ingredients as each dish sells, so stock levels are closer to real. Combine current stock, forecast sales and supplier lead times to set order quantities. See inventory management.
Checking accuracy and improving
A forecast you never check does not get better. Once a week:
- Compare forecast and actual covers for each meal period.
- Work out the percentage difference, and look for patterns. Are you always too high on Mondays? Too low when it rains?
- Update your adjustments. If eve-of-holiday dinners are always 30% above baseline, use that next time.
- Look at the cost of errors. Over-forecasting shows up as waste and idle staff hours; under-forecasting shows up as stock-outs, long waits and overtime.
Most restaurants find that a simple method, reviewed weekly, gets them close enough to plan rosters and prep with confidence.
Tools: spreadsheets, POS reports and AI
A spreadsheet is enough for a single outlet. Export sales by day and meal period from your POS each week, paste them into the sheet, and let formulas calculate the baseline. Multi-outlet groups benefit from pulling data centrally so each outlet uses the same method.
Forecasting sales for a new restaurant without history
A new outlet has no POS history, so build the projection from capacity and test it against what the site can realistically deliver.
- Start with seats and meal periods. Count seats and the services you will open for, such as lunch and dinner on weekdays and weekends.
- Estimate turns per service. Use a conservative figure, such as how many times each seat fills during lunch, based on your concept and opening hours.
- Set an average spend per cover from your menu prices and a realistic order mix, excluding GST and service charge.
- Multiply: seats × turns × average spend gives sales per service. Add takeaway and delivery as separate lines.
- Apply a ramp-up. Most outlets take months to build regulars, so plan lower figures for the first months rather than assuming full trade from day one.
- Sense-check. Count footfall at the site at your intended trading times and watch how busy similar outlets nearby are.
Replace the projection with real POS data as soon as you have four weeks of trading, and switch to the day-of-week baseline method above. Use the projection to test the lease too: our rent-to-sales ratio guide shows how to check whether the sales you need are realistic.
AI tools can help summarise trends and draft forecasts, but they still depend on clean data and your judgement about local events. Our AI tools guide covers what is available. ChaChaCha includes the AppsPOS AI agent on WhatsApp for back-office questions. If you want forecasting or rostering features in the POS itself, ask us to confirm what is available for your setup. For anything you rely on for planning, get it in writing from each vendor, including us.
To talk through the reports your outlet needs, contact ChaChaCha.