Why does theoretical food cost never match the actual figure?

Theoretical cost models the plates sold; actual cost measures stock used

Theoretical food cost starts with each recipe’s current portion cost. Multiply that cost by the number of dishes sold, then add the results.

Actual food cost starts with stock and purchases for the same period.

Use the same dates, sites, categories and VAT basis on both sides. For the full actual-cost method, read https://involiqo.com/blog/food-cost-percentage-formula-uk.

A £2,000 gap becomes useful only when it is bridged

The following Riverside Kitchen example is illustrative. During one month, it records £48,000 of ex-VAT food sales.

• Theoretical cost: £14,400, or 30.0% of sales.
• Opening stock: £8,200.
• Purchases: £19,350.
• Closing stock: £11,150.
• Actual cost: £8,200 + £19,350 − £11,150 = £16,400, or 34.2%.

The variance bridge gives five testable parts:

Supplier price drift can create variance before service starts

Riverside Kitchen’s old recipe file costs chicken breast at £6.20/kg. Its current invoice shows £7.00/kg. At 500kg used, that stale price understates theoretical cost by £400.

Two more current invoice differences add £120 on oil and £100 on cheese. The supplier-price bridge is therefore £620.

Involiqo item analytics can place the latest recurring invoice cost beside the previous item cost. It can help expose a unit-price change, changed pack or duplicate description. The operator still decides which invoice item belongs to each recipe and what yield applies.

Portion drift should be weighed, not guessed

The manager weighs a sample during three services. Chicken portions average 20g above the 180g recipe on 2,000 monthly dishes.

Cheese portions average 12g above specification on 1,000 dishes at £6.90/kg.

Together, the measured portion variance rounds to £363. Check the highest-cost, highest-volume ingredients first.

Waste and staff food need records before they can explain the gap

WRAP reported in November 2024 that UK hospitality and food service throws away 1.1 million tonnes of food each year. It said 75% could have been eaten, at a sector cost of £3.2 billion.

The Food Standards Agency’s published research describes operators using daily waste records, stock rotation and separate food-waste bins to make waste visible.

At Riverside Kitchen, logged preparation, spoilage and plate waste costs £420 at current invoice prices. Staff food adds £280, based on 140 meals at £2 product cost. Recording staff meals stops the cost being mislabelled as unexplained loss.

The final £317 is a measurement question before it is a theft claim

The residual may come from a missed stock item, a case counted as one unit, a delivery posted outside the period, an unrecorded transfer or a POS modifier with no recipe mapping.

Recount high-value lines first. Confirm that sites use the same stock unit. Match delivery dates to the stock period. Check voids, comps, modifiers and staff meals. Then repeat the bridge next week.

Theft is possible, but it is not a plug figure. Escalate only after reliable counts, current costs and operational records fail to explain a repeated item-level pattern.

Multi-site comparisons need one definition and local causes

Use one recipe version, stock calendar, unit dictionary and variance formula across every site. Then keep the cause evidence local.

A 3-point gap at one site may be stale recipe costs. The same gap elsewhere may be over-production. Ranking sites by the percentage alone rewards the team with the most convenient counting method.

1. Theoretical food cost using current invoice prices.
2. Actual stock usage for the identical period.
3. Variance pounds and percentage points.
4. Share explained by price, portion, waste and staff food.
5. Residual pounds by item and site.

Use https://involiqo.com/blog/good-food-cost-percentage-uk to judge the headline food-cost percentage. Use https://involiqo.com/blog/pub-gp-targets-by-category to see how sales mix changes venue GP.

Frequently asked questions

What is a good theoretical-versus-actual food-cost variance?

There is no universal percentage for every restaurant. Set a tolerance from a reliable period with current recipes, consistent stock counts and recorded waste. Investigate any repeated rise above that baseline. A stable small gap can reflect measured yield and normal usage; a changing unexplained pound gap needs action.

Can actual food cost be lower than theoretical cost?

Yes. Check whether closing stock was overcounted, deliveries were omitted, portions were smaller, recipes were costed too high or the sales mix was mapped incorrectly. A favourable-looking variance is still a reconciliation question. Do not treat it as a saving until the units, dates and recipe versions agree.

How often should a restaurant calculate food-cost variance?

Run a weekly bridge for high-value or unstable items and a full monthly reconciliation. Use the same cut-off time and stock units each period. A long interval makes causes harder to reconstruct because staff, prices, promotions and prep patterns have already changed.

Key takeaways

1. Compare both costs for the same period and sales basis.
2. Size the gap in pounds before percentage points.
3. Update recipes with current supplier invoice costs first.
4. Support portions, waste and staff food with measured records.
5. Treat the residual as an investigation queue, not proof of theft.

Sources

1. WRAP, November 2024: https://www.wrap.ngo/media-centre/press-releases/wrap-urges-hospitality-and-food-service-ceos-and-leaders-take-stand
2. Food Standards Agency: https://www.food.gov.uk/sites/default/files/media/document/Creating%20food%20waste.pdf
3. Meez, updated June 2026: https://www.getmeez.com/blog/actual-vs-theoretical-food-costs

Build one variance bridge this Monday

Choose one week and one food category. Refresh the recipe costs from current invoices, reconcile stock usage, then add price, portion, waste and staff-food evidence until the residual is clear enough to investigate.

See how Involiqo connects supplier invoice lines with item-level cost evidence at https://involiqo.com/#bento-overview.