Restaurant Sales Forecast: Formula, Template and Example

A restaurant sales forecast estimates future order volume and net sales for a specific day, shift, channel or period. A useful forecast is not a target chosen by the owner; it is a documented estimate based on comparable history, current demand signals and capacity.

Forecasting will never remove uncertainty. Its value is giving purchasing, prep, staffing, cash flow and marketing one reasonable operating assumption. Begin with a simple order-count model, measure the error and add complexity only when it improves decisions.

The basic restaurant sales forecast formula

Forecast sales = forecast orders × forecast average order value. For a full restaurant, calculate that by material channel or daypart and add the results. Net sales should use a consistent treatment of discounts, refunds, VAT or service charges.

Example: forecast 120 dine-in orders at ৳850, 55 direct takeaway orders at ৳600 and 35 marketplace orders at ৳700. Estimated sales are ৳102,000 + ৳33,000 + ৳24,500 = ৳159,500 before any adjustments defined in the model.

• Forecast orders from comparable demand
• Forecast average bill from recent mix and planned prices
• Calculate each major channel separately
• Apply known closure, event or capacity adjustments
• Document assumptions and version date

Choose comparable historical data

Compare like with like. Last Friday is usually more useful for next Friday than yesterday; the same Ramadan service period is more useful than an ordinary month; a rainy delivery day may differ from a clear dine-in day. Use several comparable periods so one unusual event does not dominate.

Clean known distortions. Record closures, stock-outs, major campaigns, private events, system downtime and days when capacity was constrained. Low sales during a power failure do not measure normal demand.

Add current demand signals

Adjust the historical baseline for reservations and deposits, group bookings, holidays, salary-cycle patterns, school or office calendars, weather, traffic disruption, promotions, influencer activity, competitor openings or closures and local events. Use evidence, not excitement.

Separate confirmed signals from speculative ones. A paid booking is stronger than an expected walk-in crowd; a repeat promotion has a better response estimate than an untested campaign. Note the adjustment and its reason so accuracy can be reviewed later.

Forecast by daypart and channel

A daily total is too broad for staff and prep. Split breakfast, lunch, afternoon and dinner where patterns differ. Then separate dine-in, counter takeaway, direct online ordering, marketplace and catering if their demand and economics are material.

Granularity has a cost. Do not create 40 rows nobody maintains. Start with the segments that change staffing, prep, fees or capacity. Forecast menu mix only for high-impact categories or ingredients.

Convert the forecast into an operating plan

Orders and sales are not the end product. Convert forecast orders into covers, tables, delivery pickups, prep quantities, critical ingredient needs and staff-hours by role. Check whether counter, kitchen, seating or rider capacity makes the forecast impossible.

Use a prep yield and safety buffer based on item volatility and replenishment time, not one blanket overproduction percentage. A slow, expensive protein needs a different buffer from rice or packaging.

• Expected orders by 30- or 60-minute block
• Staff required at floor, counter and kitchen bottlenecks
• Prep by category and critical ingredient
• Supplier order or transfer needed
• Table, kitchen and delivery capacity
• Trigger for calling extra staff or stopping a promotion

Create base, low and high scenarios

The base case is most likely, not desired. A low case tests weak demand or disruption; a high case tests capacity and stock. Define actions for each: reduce prep or delay a purchase in the low case, open another station or call approved cover in the high case.

For cash planning, use the base case and inspect the downside. For sales targets, keep the goal separate from the forecast. Comparing actuals with both answers different questions: forecast error measures estimation, while target variance measures performance against ambition.

Measure forecast accuracy

At the end of each day or week, record actual orders and sales beside forecast. Error = actual − forecast. Absolute percentage error = absolute error ÷ actual, when actual is not zero. Review bias: repeated over-forecasting creates waste and excess labour; repeated under-forecasting causes stock-outs and delays.

Do not punish a manager for every error. Ask whether the input was available and whether the response was sensible. Improve the model when errors follow a pattern by weekday, weather, channel or event.

A simple weekly forecast template

Use rows for date, daypart, channel, comparable historical orders, baseline average bill, demand adjustments, forecast orders, forecast average bill, forecast sales, actual orders, actual sales, error, reason and next action. Add owner and update time.

Lock definitions so one branch does not report gross billed amount while another reports net sales. Keep a notes column for stock-outs and capacity constraints that affected actual demand.

Use restaurant software as the evidence layer

Rosuii keeps orders, channels, payments, menu and branch reports together, providing cleaner historical inputs than manually totaling separate notebooks. Export or review consistent periods, then combine that evidence with reservations, weather and local knowledge.

Forecasting remains a management decision. The system can show history; it cannot know an unrecorded road closure, private event or equipment limitation. Record those adjustments so the next forecast learns from them.

Approve and hand off one forecast version

Before ordering or publishing a roster, have the manager, kitchen or purchasing lead and roster owner review the same version. Confirm the order and sales estimate, risky assumptions, critical ingredients, capacity limit, staffing trigger and downside response. A revenue target alone does not tell the operating team what to prepare.

Lock the approved version and name the next refresh time. If a booking cancels, weather changes, equipment fails or a supplier cannot deliver, record a revision instead of silently replacing the forecast. Compare original, revised and actual results after close so the team learns whether the input, timing or response needs improvement.

A useful forecast is simple enough to update, detailed enough to change a decision and honest enough to show uncertainty. Start with orders × average order value by major channel, translate it into staff and prep, and improve the model from measured error every week.

Related guides

• Restaurant Reporting Software
• Restaurant KPIs Every Owner Should Track
• Restaurant Peak-Hour Management

See this workflow in Rosuii: Turn Rosuii order reports into operating forecasts

Start using Rosuii for free