How Restaurants Can Forecast Demand to Cut Food Waste

Written by: JJ Tan, Founder, Jelly | Last updated: 20 July 2026

Key Takeaways for UK Restaurants

• UK restaurants lose 4–10% of food spend to waste when ordering relies on intuition instead of data-driven forecasting.

• A 12-week POS baseline plus UK variables such as weather, school holidays, and events can cut waste by 18–50% within 90 days.

• Converting menu-item forecasts into precise ingredient quantities using recipes, yield factors, and safety buffers prevents over-ordering.

• Daily forecast updates from reservations and real-time sales, plus weekly waste audits, keep accuracy above 90% and reduce bin waste.

• Jelly automates invoice capture, recipe costing, and sales-mix reporting so the forecasting system stays accurate without extra admin hours — see how it works for your operation.

Before You Begin: What You Need in Place

This guide needs three foundations before you start: access to supplier invoices (paper or digital), at least 12 weeks of item-level POS sales history, and a set of standardised recipes with portion weights. The head chef or operations manager should own the process, with the finance manager or owner setting cost targets and margin thresholds. Without these inputs, any forecast you build will rely on guesswork instead of data.

1. Build a Baseline from Historical Sales by Day, Hour and Menu Item

Objective: Establish a reliable sales baseline per menu item, per day of week, and per daypart that replaces gut-feel ordering with structured data.

Exact action: Export item-level sales data from your POS covering the most recent 12 weeks. Reliable forecasting requires at least 12 months of daily item-level sales data for full seasonal coverage, but 12 weeks is enough to begin. Sort by dish name, quantity sold, date, and daypart. Separate weekday from weekend figures. Calculate average daily portions per dish for each day of the week.

Inputs: POS sales history, invoices to confirm which ingredients map to which dishes, and standardised recipes.

The following simple template turns raw POS exports into a clear weekly baseline that highlights both volume drivers and slow-moving waste risks.

Forecasting template — weekly baseline per menu item:

1. List every active menu item in column A.

2. In columns B–H, enter average portions sold per day of week (Mon–Sun) based on the last 12 weeks.

3. In column I, note the primary ingredients and portion weight per dish.

4. In column J, flag items selling fewer than 8 portions per week — slow movers frequently generate more waste cost than profit contribution.

Successful output: A weekly sales baseline per dish that shows which items drive volume on which days. This baseline becomes the foundation for every downstream purchasing and prep decision.

2. Adjust for UK Weather, School Holidays and Local Events

Objective: Adjust the baseline forecast to reflect the demand shifts that UK operators experience but rarely quantify.

Exact action: Before each week, review three variables and apply a percentage adjustment to your baseline cover count.

• Weather: England received 70% more rainfall than the long-term average in February 2026, which directly suppressed footfall at pubs, bars, and restaurants. When your forecast week includes wet days, apply a −5% to −10% cover adjustment to reflect this drop in traffic. The reverse applies during heatwaves, as venues with meaningful outdoor seating can see sales increases, so adjust upward if you have a terrace or garden that attracts guests in good weather.

• School holidays: Half-terms and summer holidays shift both cover counts and sales mix. Family-friendly sites typically see a 10–20% uplift in lunch covers, while city-centre sites serving office workers often see a dip.

• Local events: A nearby festival, sporting fixture, or concert can shift both volume and sales mix significantly. Granular per-shift adjustments enabled by context variables can reduce protein waste, which is usually the most expensive input.

Inputs: Met Office forecasts, local council event calendars, school term dates, and your own historical notes on event-day performance.

Successful output: An adjusted weekly cover forecast that reflects real UK demand drivers. This reduces the risk of over-prepping on slow days and running short on event days.

3. Turn Menu-Item Forecasts into Ingredient Order Quantities

Objective: Translate cover forecasts into precise raw-ingredient purchasing quantities for each supplier order.

Exact action: For each forecasted menu item, apply the recipe portion weight and a yield factor to calculate the raw order quantity. A 90g recipe card weight becomes a 115g purchase requirement once a 0.78 yield factor is applied. Use this formula for every ingredient.

Order quantity = (Forecasted portions × Portion weight × Yield factor × Safety buffer) − On-hand stock

Set the safety buffer by ingredient type so the formula reflects real risk. Fresh proteins usually need a higher buffer than dry goods and ambient ingredients. Keep invoice-level price data current, because recipe costs built on last month’s prices will distort GP before service even starts.

Inputs: Standardised recipes with portion weights and yield factors, current invoice prices, and on-hand stock counts.

Successful output: A supplier-ready purchase list derived from the forecast instead of habit or last week’s order.

4. Refresh Forecasts Daily with Reservations and Live Sales

Objective: Refine the weekly forecast each morning so prep lists reflect today’s actual demand signal rather than a seven-day-old estimate.

Exact action: Each morning, pull the current reservation count from your booking system and compare it with the baseline forecast for that day. If reservations run 15% above forecast, increase prep quantities for high-velocity items by a similar percentage. Forecasts should be refreshed daily rather than quarterly, with intra-shift adjustments for high-velocity perishables based on real-time sales trajectory. Where your POS integrates with your back-of-house system, live sales data during service supports mid-shift adjustments, such as holding back a second batch of a slow-moving special instead of committing all prep at the start of service.

Inputs: Reservation system data, live POS sales, and the prior day’s closing stock figures.

Successful output: A daily prep list that tracks closely with the morning’s reservation count and live demand, which reduces end-of-service overproduction on perishable items.

5. Connect Forecasts Directly to Purchasing and Prep

Objective: Ensure the forecast output drives purchasing decisions directly, closing the gap between what the data shows and what actually gets ordered.

Exact action: Use the ingredient quantities calculated in Step 3 to generate a draft purchase order per supplier. The reorder point formula is: Reorder Point = Average Daily Usage × Vendor Lead Time + Safety Stock, which gives a clear trigger for each order. Separate your inventory into three categories so each group has a sensible ordering rhythm.

• High-velocity items such as proteins, bread, and fries work best on daily or every-other-day orders that follow the rolling forecast closely.

• Perishable items such as dairy, fresh vegetables, and seafood suit a dynamic par driven by the weekly forecast instead of a fixed standing order.

• Low-velocity items such as ambient and dry goods can be ordered weekly or fortnightly with a modest 3–7% buffer.

A forecast that stays in a spreadsheet and needs manual translation into purchase orders will be used inconsistently, while teams that connect forecast output directly to purchasing can generate draft orders automatically for buyer review.

Inputs: Ingredient-level forecast quantities, current stock counts, supplier lead times, and minimum order quantities.

Successful output: Purchase orders sized to forecasted demand rather than habit, with standing orders replaced by dynamic quantities each cycle.

See how Jelly connects invoice data to real-time dish costs and purchasing decisions in a live walkthrough.

6. Run Waste Audits and Track Forecast Accuracy

Objective: Quantify how accurate your forecasts are and pinpoint which items and days drive the most waste.

Exact action: Run a weekly waste audit using a simple tracking framework. Record waste at the moment of disposal, because delayed logging misses a substantial portion of the data needed for accurate pattern detection.

Track these four metrics weekly so you can see how volume, accuracy, and waste interact.

• Waste KPI %: (Value of discarded ingredients ÷ Total ingredient purchases) × 100. Restaurants track total waste as a percentage of food purchases to monitor performance.

• Forecast Accuracy %: (Actual covers ÷ Forecasted covers) × 100. Target 90% or above as a weekly-reviewed KPI.

• MAPE (Mean Absolute Percentage Error): The average percentage difference between forecasted and actual demand per item. Manual spreadsheet-based forecasting typically produces a MAPE in the 25–50% range. Lower values indicate higher performance.

• Waste per cover: Total waste value ÷ Number of covers served. This normalises waste for changes in business volume and supports comparison across weeks and sites.

Well-calibrated purchasing forecast models can predict and reduce waste while keeping stockouts under control. Sites that use continuous daily waste measurement usually cut waste faster than those relying on monthly or quarterly audits.

Inputs: Daily waste logs listing product, quantity, and reason, plus invoice purchase values and POS cover counts.

Successful output: A weekly one-page dashboard showing Waste KPI %, Forecast Accuracy %, MAPE, and waste per cover. This gives the head chef and operations manager a clear signal on whether the system is improving.

7. Decide When to Move from Spreadsheets to Automation

Objective: Spot the point at which manual forecasting creates more admin than it saves, then move to an automated platform that keeps the system running without extra hours.

Exact action: Treat manual spreadsheet forecasting as a workable option for one to three sites with clean POS data. After 5 locations, manual spreadsheet forecasting typically breaks down due to data volume and multi-site variability. The real trigger for automation arrives when the time spent maintaining the system outweighs the savings, or when invoice price changes slip through and recipe costs no longer reflect current prices.

Jelly supports this transition. It scans every line item of every supplier invoice, captured by photo or email, and updates ingredient costs in real time across every dish in your recipe book. When a supplier raises the price of a protein, Jelly flags it immediately through its Price Alert feature and recalculates the GP margin for every dish that uses that ingredient. Its Sales Mix report, powered by live POS integrations, shows which dishes are selling and which are most profitable, giving you the data needed to keep forecasts accurate without manual data entry. The Flash Report delivers a daily, weekly, or monthly view of gross profit margin calculated from live invoice costs and POS sales, which replaces the month-end accountant report with a same-day signal.

Inputs: Supplier invoices by photo or email and POS sales data via API integration.

Successful output: A fully automated data layer that keeps recipe costs, sales mix, and GP margins current without extra admin hours. This makes the forecasting system in Steps 1–6 sustainable at scale.

Common Forecasting Mistakes to Avoid

Three recurring errors undermine restaurant demand forecasting systems.

• Inconsistent data capture: Forecasts built on incomplete POS data, where some transactions are voided, some channels are excluded, or some days are missing, produce baselines that are structurally wrong. Every POS transaction must be captured at item level, every day, for the baseline to stay reliable.

• Ignored price changes: A forecast that predicts volume accurately but uses last month’s ingredient prices will drive purchasing decisions that erode GP quietly until the month-end report arrives. Forecasts and ingredient prices should be reviewed regularly, with adjustments for significant price shifts.

• Failure to adjust for events: Using a four-week rolling average without flagging bank holidays, local events, or school holidays produces forecasts that are structurally wrong on the days that matter most. Traditional rolling averages can create notable forecast errors because they ignore these variables.

How to Measure Forecasting Success

Three metrics show whether the forecasting system is working.

• Reduced bin weight and Waste KPI %: Track total waste value as a percentage of purchases weekly. A well-functioning system should move this from the 6–10% range toward the 2–4% range within 90 days. Restaurants that track waste regularly often reduce their food costs within the first few months.

• Improved forecast accuracy %: Maintain the 90% accuracy threshold established in your weekly audits. MAPE should fall from the manual baseline as the system matures.

• Lower food cost %: Food cost percentage should show a clear reduction over the first quarter.

Advanced Practices for Faster Results

Once the core system is running, two practices can accelerate results.

• Daily rolling forecasts: Instead of setting a weekly forecast on Monday and leaving it unchanged, update the remaining days of the week each morning using the prior day’s actual sales and that morning’s reservation count. Dynamic wave production, such as 50% initial batch, 30% mid-service, and 20% on-demand, reduces overproduction risk compared with single large-batch forecasting.

• Integration with accounting software: Connecting invoice data directly to accounting software such as Xero removes the reconciliation step between kitchen purchasing records and financial reports. Jelly’s one-click push to Xero sends invoice line items captured in the kitchen straight into the accounts, giving the finance manager or owner real-time cost visibility without waiting for manual data entry.

Frequently Asked Questions

How often should a restaurant update its demand forecast?

The baseline forecast should be reviewed and updated weekly, with daily adjustments made each morning using that day’s reservation count and the prior day’s actual sales. For high-velocity perishables such as proteins and fresh produce, intra-shift adjustments are appropriate when real-time POS data shows sales running significantly above or below the morning forecast. Monthly reviews of the underlying baseline are necessary to capture shifts in menu mix, seasonal patterns, and changes in customer behaviour. A full seasonal recalibration, comparing year-over-year data for the same period, should be completed at least twice a year.

Can this forecasting process work across multiple sites?

Yes, but the data infrastructure requirements increase with each additional site. For one to three sites, a well-maintained spreadsheet system running a weighted moving average can produce reliable forecasts. Beyond three sites, the volume of invoice data, POS transactions, and recipe updates makes manual maintenance impractical, so forecast accuracy degrades and the admin burden grows faster than the savings. At this point, a platform that automates invoice capture and integrates with POS systems across all locations becomes necessary to maintain forecast quality without adding headcount. Jelly is designed specifically for operators at this growth stage, providing a single dashboard across all sites at a flat rate of £129 per location per month.

How do supplier price changes affect the forecast, and how should operators respond?

Supplier price changes affect the forecast in two ways. First, they change the cost of the ingredients underpinning each dish’s GP margin, so a dish that was profitable at last week’s prices may be loss-making at this week’s prices. Second, significant price increases on a key ingredient may justify a change in menu mix, such as substituting a lower-cost protein or adjusting portion weights, which then changes the ingredient quantities the forecast needs to generate. Operators should treat any price change exceeding 10% as a trigger for immediate recipe cost recalculation and a review of the affected dish’s menu price. Tracking price changes at invoice line-item level, rather than waiting for a supplier statement, is the only way to catch these shifts in time to act. Jelly’s Price Alert feature flags every price increase or decrease the moment a new invoice is scanned, giving chefs and operations managers the data to negotiate credits, switch suppliers, or reprice dishes before the margin damage compounds.

Conclusion: Turning Forecasting into a Repeatable System

A repeatable demand-forecasting system built from historical POS sales, UK-specific variables, and accurate invoice-level ingredient costs gives a direct route to cutting food waste and protecting gross margins in a UK restaurant, pub, or boutique hotel. The seven steps above provide the process. The missing piece for many operators is not the method but the data infrastructure to sustain it as the business grows. Jelly automates the invoice capture, recipe costing, and real-time sales mix reporting that keep this system working continuously, not just when someone has time to update a spreadsheet. Operators can cut food costs once they gain live visibility into dish costs and supplier price changes.

Start cutting food waste and protecting margins this week — see how Jelly works for UK restaurants.