Stock Variances in Food Service: Reducing Them for Good
Operations · September 26, 2025 · By Alex Dondelinger · 10 min read
Stock variance is the difference between theoretical consumption (calculated from sales + recipe cards) and actual consumption (from stock counts, deliveries and outflows).
Definition: Stock variance (food service)
Formula: Variance = Actual consumption – Theoretical consumption
What does it look like in practice?
You sold 120 burgers. The recipe card calls for 150 g of patty per portion → 18 kg in theory.
Stock counts + deliveries show 19.2 kg actually consumed → variance = +1.2 kg (+6.7%).
Why is it critical for margin?
Every +1% of variance on expensive categories (meat, cheese, alcohol) immediately eats into margin: at equal sales, ingredients cost more for the same plate.
What a variance reveals (operations)
- Approximate portioning / recipe cards not followed
- Deliveries poorly recorded (quantities, prices, substitutions)
- Undeclared losses that add up
Customer impact
- Excess → overproduction, waste and stockouts elsewhere
- Shortfall → inconsistent portion sizes, uneven quality, falling reviews
Key takeaway
Variance is a signal, not a fatality. With standardized recipes, clean deliveries, a loss log and regular stock counts, it becomes a direct lever for margin and waste reduction.
Measuring Stock Variances Properly: KPIs, Methods & Routines
Before correcting a variance, you first have to make it measurable, comparable and regular. The backbone is your KPIs and the routines that feed them.
Start by Calculating Stock Variance
Formula (%):
Variance % = ((Actual stock – Theoretical stock) / Theoretical stock) × 100
Formula (€):
Variance € = Σ(Actual qty × Unit cost) – Σ(Theoretical qty × Unit cost)
Track it: as a percentage and in value, by period (week/month), by site, by category (meat, cheese, alcohol…) and by dish.
A Simple Example (a Single Ingredient)
- Theoretical: 100 kg
- Actual: 92 kg
- Unit cost: €6/kg
- Variance % = ((92 − 100) / 100) × 100 = −8%
- Variance € = (92 − 100) × 6 = −€48 (loss)
Second Pillar: Unknown Shrinkage.
It covers untraced losses (theft, tastings/tests, unit errors…). Give it a measured existence, in € and as a % of food cost, and aim for < 0.5–1.0%. The lower this ratio goes, the sharper your diagnosis gets. Alongside it, track your categorized losses (use-by date, breakage, staff meals, comps) in a mandatory log filled in at the end of every service: without daily categorization, unknown shrinkage inevitably balloons.
Counting Cadence Creates Consistency.
Roll out rotating stock counts: weekly on sensitive items (meat, alcohol, cheese), twice a week on intermediate categories, monthly on the rest, in addition to a full monthly stock count. In the kitchen, check the accuracy of recipe cards with spot weigh-ins of best-sellers: the goal is 95–97% compliance on portion weights.
On the process side, a three-step receiving routine avoids false variances: first quantities, then prices & substitutions, finally immediate data entry (don’t put it off until the next day, or the variance “shifts” and becomes undetectable). Keep your price lists up to date at least weekly to make valuation reliable. And block a 30-minute weekly review in the calendar: go through the Top 10 variances of each site and assign named corrective actions (who does what, by when).
Reliable Data
All of this rests on data prerequisites: harmonized units & equivalences (case / bag / piece / g), a single multi-site item list linked to the POS, versioned recipe cards with locked portion weights and a ±3–5% tolerance at most, and a standardized loss log (four reasons, no more). Conversely, three mistakes are costly: counting stock before unrecorded deliveries have been entered, patching together approximate conversions (“1 case = ? pieces”), or leaving prices out of date (poor valuation produces a good diagnosis… on false numbers).
Key takeaway: what isn’t measured doesn’t exist. In most groups, the sheer regularity of stock counts and delivery-receiving discipline explain most of the drop in variances from the first month.
Diagnosing Stock Variances: Causes, Symptoms and Corrective Actions
Cause Observable symptoms Checks to run Corrective actions Kitchen / Portioning
- Recurring positive variances on “rush” dishes.
- Portion weights vary from one chef to another.
- “Cold” weigh-ins of 3 best-seller plates.
- Recipe cards (RC) up to date vs. size/supplier.
- Scales & dosing tools at stations + ±3–5% tolerance.
- Quick briefing & recipe cards displayed in the kitchen.
Receiving & Purchasing
- Variance spikes after delivery days.
- Invoices ≠ purchase orders (PO).
- Check quantities, prices & substitutions on the latest delivery.
- Units entered (kg vs. piece, L vs. cL).
- Three-step receiving + immediate data entry.
- Refuse/categorize substitutions and issue a credit note if needed.
Stock counts & Units
- “Unexplained” variances at month-end.
- Inconsistent multi-format products.
- Trio per item: storage unit / recipe unit / conversion.
- Is the stock count frequency (A/B/C) respected?
- Harmonize units & equivalences (case/bag/piece/g).
- Rotating stock counts + cross-checks.
Use-by dates / FEFO & Overstock
- Uncategorized losses, heavier trash.
- Paradoxical stockouts despite overstock.
- FEFO respected, dates marked, linear rotation.
- Loss log filled in daily.
- Prep in FEFO order + standard labeling.
- Anti-waste plan (rescue recipes, donations).
“Known” & “unknown” losses
- Unknown shrinkage on the rise (>1% of food cost).
- Variances concentrated at the end of service.
- Loss log (breakage, use-by date, staff, comps) kept every day?
- Traceability of sensitive products (alcohol/meat).
- Categorize 100% of losses (4 reasons max).
- Securing sensitive areas/products.
Data & POS
- Same dishes with opposite variances depending on the site.
- Theory ≠ actual with no operational reason.
- Single multi-site item list linked to PLUs?
- Do sales properly deplete ingredients?
- Harmonize labels & ingredient↔PLU mappings.
- POS depletion audit (test sales).
Organization & Routines
- Variances that repeat, no actions tracked.
- Unclear responsibilities.
- 30-minute weekly review (Top 10 variances) actually held?
- Actions assigned (who does what, by when)?
- Set up the weekly review & an escalation plan.
- Tracking board: variances → actions → results.
Investigation method (common thread): start from the Top 5 variances by product, then work back up the chain: execution (weigh 3 plates) → receiving (invoice & delivery note) → stock count (unit & conversion) → losses (service log). You’ll almost always end up with a specific, fixable cause.
Fullsoon • AI & Data
Cut Your Stock Variances Without Weighing Down Operations
Forecast right, buy right, trace right, with measurable impact from the first month.
≈ 94%Dish-level forecast accuracy −33%Food waste +6 to +8%Gross margin −20h/monthStock/purchasing management time
1) Fine-grained forecasts ➜ less overstock
Dish-level forecasts every 30 minutes over ≥ 30 days, factoring in weather & events.
- Production & purchasing aligned → fewer use-by date losses and unsold items.
- Stockouts avoided on sensitive best-sellers.
2) Guided purchasing ➜ more accurate theory
Purchase recommendations by ingredient with lead time and up-to-date prices.
- Orders sent to your preferred suppliers from the platform.
- Reliable valuation = robust theoretical consumption.
3) Actuals traced ➜ variance explained
Invoice OCR, loss log (use-by date, breakage, staff, comps) & guided rotating stock counts.
- “Unknown” shrinkage reduced to a minimum.
- Alerts on products/categories above the threshold.
4) Clean data ➜ actionable diagnosis
Harmonized multi-site item lists & versioned recipe cards (±3–5% tolerance).
- Recipes ↔ PLUs linked: ingredient depletion guaranteed.
- “Variance %/€” dashboard by site, category, dish.
Stock Variances: Reading and Acting With the Green / Orange / Red System
🟢 Green light (≤ 1.5%): “We’re on track.”
When variance stays under 1.5%, it means your basics are in place: the portion weights stated in the recipe cards resemble what actually leaves the kitchen, deliveries are recorded correctly, and rotating stock counts are done regularly. In this zone, the right reflex is not to do more, but to preserve the rhythm: a weekly check-in to review the biggest variances, a loss log filled in every day (use-by date, breakage, staff, comps) and a particular eye on the expensive categories (meat, cheese, alcohol). If one of them nears 1%, correct it locally right away to avoid sliding toward orange.
🟠 Orange light (1.5–2.5%): “It’s drifting, correct quickly.”
Beyond 1.5%, you’re rarely facing a “mystery”: the cause lies most often in receiving (quantities, prices, substitutions poorly recorded) or in badly converted units (case/bag/piece ↔ grams/kilos). The goal is to get back to green within a week. Day 1: revisit the latest delivery and straighten out the conversions product by product. Day 2: weigh three plates of the best-sellers to check the actual portion weight and recalibrate the recipe card if needed (±3–5% tolerance). The correction is considered lasting when variance falls back below 1.5% for two weeks.
🔴 Red light (> 2.5%): “We trigger the escalation plan.”
Beyond 2.5%, there isn’t one cause but several adding up: irregular portioning and uncategorized losses and sometimes poorly mapped POS data (recipe ↔ PLU). Here, we act within 48 hours. Day 1: tight stock count on sensitive items (meat/alcohol), update of the loss log, securing access. Day 2: cold weigh-ins to objectify actual portion weight, and a POS test to make sure sales properly deplete ingredients. The following week, we step up rotating stock counts, apply FEFO strictly and update the price lists. The target is clear: back under 2.5% in 14 days, then under 1.5% in 30 days.
An example that speaks for itself.
You sell 120 burgers with 150 g of beef planned: theory says 18.0 kg. If the stock count/deliveries show 19.2 kg, the variance is +1.2 kg, or +6.7%. You go straight into red: the priority isn’t to “search everywhere”, but to follow the chain in order: receiving → portioning → stock count → losses. In practice, it’s often just an undeclared substitution or an actual portion weight that has drifted by 10–15 g during the rush that creates this kind of variance.
Where Fullsoon helps in concrete terms.
Dish-level forecasts every 30 minutes over more than a month prevent overstock, and therefore uncategorized use-by date losses; purchase recommendations and invoice OCR make the theory (costs/quantities) reliable, and the “Variance %/€” dashboard by site, category and dish highlights drifts in the right place at the right time. The result: you spend more time fixing the real causes and less time guessing.
Frequently Asked Questions About Stock Variances in Food Service
Frequently Asked Questions
What is a stock variance and how do you calculate it?
Stock variance is the difference between actual consumption (stock counts, deliveries, outflows) and theoretical consumption (sales × recipe cards).
Formula: (Actual − Theoretical) ÷ Theoretical. Track it in % and in €.
What threshold should you aim for to be “in the green”?
In food service, a good benchmark is ≤ 1.5% overall variance. Between 1.5–2.5%, you monitor and correct; beyond 2.5%, you trigger an escalation plan. For sensitive categories (meat, cheese, alcohol), aim for a tolerance cut in half.
“Known” vs. “unknown” shrinkage: what’s the difference?
Known shrinkage covers declared losses (use-by date, breakage, staff, comps). Unknown shrinkage corresponds to untraced losses (theft/spillage, pilfering, unit errors). Goal: categorize 100% of known losses daily and bring unknown shrinkage down to < 0.5–1% of food cost.
How often should you count stock to make variances reliable?
Set up rotating stock counts: A (sensitive items) weekly, B twice a week, C monthly, + a monthly full stock count. Regularity explains most of the drop in variances.
What are the most common causes of stock variances?
Portioning (recipe cards not followed, lack of tools), receiving (quantities, prices, substitutions), inconsistent units/conversions (case/bag/piece ↔ g/kg), poorly managed use-by dates/FEFO and the absence of a loss log. In multi-site setups, a misaligned item list is often added to the mix.
How do you correct quickly when variance goes “orange” or “red”?
D+1: check the latest delivery (quantities, prices, substitutions) and the conversions.
D+2: weigh 3 plates of the best-sellers and recalibrate the recipe cards (±3–5%).
In “red”, add an A-category stock count on sensitive items and secure access (alcohol/meat).
What role does the recipe card play in reducing variances?
The recipe card defines the theory. It must be versioned, linked to the POS (PLU), and its portion weight held within a clear tolerance (±3–5%). Spot weigh-ins (≥ 95–97% compliant) keep the theory from drifting away from reality.
How do FEFO and use-by dates affect stock variance?
Without FEFO and date marking, expired products end up in unknown shrinkage. Prepping in FEFO order + clear labeling mechanically reduces use-by date losses and stabilizes variance.
In multi-site setups, why harmonize product item lists?
Aligned labels and units across sites avoid different theories for the same product. The result: comparable variances and targeted actions, rather than data-driven “false alarms”.
How exactly does Fullsoon help reduce stock variances?
Through dish-level forecasts every 30 minutes (≥ 30 days) to avoid overstock, purchase recommendations with up-to-date prices, invoice OCR and a built-in loss log. The “Variance %/€” dashboard by site, category and dish lets you detect drifts early and assign quick corrective actions.