Full inventory counts on Sunday night are how most operators were taught to do this. Full counts take 4–6 hours, exhaust a senior team member, and produce numbers that are stale before the new week starts. A focused weekly audit takes 90 minutes and catches more variance.
The 80/20 of inventory
Most restaurants have 200+ SKUs and 20 SKUs that account for 80% of the food cost. The 20 are usually proteins, premium produce, premium spirits, and a handful of high-cost specialty ingredients. Audit those 20 every week. The other 180 get a full count once a quarter.
That is the framework. Everything else is execution detail.
The 90-minute weekly audit
Step 1: Print the top-20 list (5 min)
Pull a sales-by-item report for the week. Identify the 20 highest-cost ingredients used during the week (cost-of-goods-used, not invoiced quantity — those differ if you didn't sell what you bought). Print the list.
Step 2: Physical count of just those 20 (40 min)
Count what is on the shelf, in the walk-in, in the freezer. Convert to consistent units. This is the part that takes practice; the senior cook or sous chef should own it and not delegate to a new hire.
Step 3: Compare to theoretical (15 min)
Theoretical use = sum of (recipe quantity × portions sold) across the menu items that contain that ingredient. Subtract from beginning inventory + week's purchases. The result is what should be left. Variance from physical count = your audit signal.
Step 4: Investigate per-SKU variances over 5% (30 min)
Industry-wide inventory variance benchmarks sit at 1–2% of total sales (anything sustained above 10% triggers immediate audit), but those are top-of-house numbers. At the per-SKU level over a single week, 5% variance on a major protein is your investigation threshold — high enough to filter out legitimate recipe-yield noise, low enough to catch real theft, vendor shorts, and recipe drift. Investigate: are recipes drifting (cooks plating heavier than spec)? Is there theft (rare but real)? Did a vendor short the delivery and nobody caught it? Is the recipe yield wrong because of a process change?
The 30-minute investigation budget is what makes the audit useful. Without it, the variance numbers sit in a spreadsheet doing nothing.
Where the apps fit
VendorWatch handles the invoice-vs-delivery layer: scan every invoice, compare against prior weeks, flag missing items or duplicate charges. That removes one entire class of variance from the audit (vendor mistakes) before it shows up as inventory shortage.
ChefScale + MyCookbook hold the recipe data that backs the theoretical-use math. Without accurate recipe quantities, your theoretical inventory is fiction and the audit catches nothing useful.
The Restaurant Consultant ties this into the weekly P&L review so the variance numbers translate to dollar terms in the same screen as labor and prime cost.
The full quarterly count still matters
Once a quarter, count everything — including the 180 long-tail SKUs. The point is to catch slow-moving items that have spoiled, theft of low-cost items at scale, or outright SKU drift (you stopped using an ingredient six months ago but it is still on the shelf taking up space). Quarterly is enough; weekly for those 180 is paperwork-for-paperwork's-sake.
See the apps or read the food cost automation guide.
Sources
Last updated: .
This article draws on industry-standard operational data plus 14 years of operating experience at Mouton's Bistro & Bar (Cedar Park, TX) and Mouton's Southern Bistro (Leander, TX).