Restaurant inventory management: the practical playbook

Inventory is the least glamorous hour of your week and one of the most profitable, which is exactly why it's the first thing operators skip when they get busy — and skipping it is precisely how food cost drifts up two points before anyone notices. Counting shelves isn't the fun part of running a restaurant. It's the part that tells you the truth about where your money went. Here's the working playbook, minus the theory.

Why it's the truth serum for food cost

Your food cost percentage is only as honest as your inventory counts. The formula makes it obvious:

COGS = beginning inventory + purchases − ending inventory

Skip the counts and divide purchases by sales instead, and you're measuring your buying, not your usage — the number spikes when you stock up and magically improves when you run the shelves down. Only real beginning and ending counts turn food cost into a metric you can act on, which is the whole foundation of the food cost guide. Get inventory right and you control COGS; get it wrong and every number downstream is fiction.

Par levels: how much to keep

A par level is the minimum you keep on hand before reordering. The formula:

Par = (average daily usage × days between orders) + safety stock

If you order a protein twice a week and use 10 lb a day, par is roughly 35–40 lb — about five days of usage plus a buffer. Par levels do two jobs: they stop you running out mid-service, and they stop you over-buying perishables that spoil into the trash. Set them per item, and revisit them seasonally as usage shifts.

Count cadence: tiered, not uniform

You do not count everything every day — that's how counting becomes the task nobody does. Tier it:

• High-value, high-volume items (premium proteins, top sellers): count daily or weekly. This is where the money and the shrink are.
• Stable, low-value items (dry goods, disposables): weekly or monthly is fine.

The discipline that matters more than frequency: count consistently — same people, same units, same time relative to deliveries — or your period-over-period numbers become incomparable and your variance turns to noise.

FIFO and the variance that catches theft

Two practices separate a controlled kitchen from a leaky one:

FIFO — first in, first out. Use older stock before newer. It's mundane and it directly cuts spoilage, keeps costing accurate during inflation, and is the single easiest waste win on this list.

Theoretical vs. actual variance. Theoretical usage is what your recipes say you should have used; actual is what the inventory math says you did. The gap is the diagnosis. A variance over roughly 5% is a clear signal that something — waste, over-portioning, theft, or bad purchasing — needs attention. The catch: you can't calculate theoretical usage without standardized recipes specifying gram weights per ingredient. No recipe costing, no variance, no diagnosis.

The 80/20 rule for where to spend your attention

Not all SKUs deserve equal effort. The Pareto (ABC) principle holds in kitchens: roughly 20% of your items — the "A" items — account for about 80% of your consumption value. Those get the tightest control and the most frequent counts; the low-value "C" items (napkins, disposables) can be counted loosely to save time. Focusing on the A-items is also where the real savings hide — bulk negotiation, supplier bids, and portion control on your top proteins move food cost far more than fussing over the cheap stuff.

What software does (with the numbers kept honest)

Inventory software genuinely helps, but the vendor stats are vendor stats — real direction, optimistic magnitude:

• Reduces food cost by roughly 3–8% within a few months (vendor-reported).
• Cuts waste 25–35% without changing recipes or portions (vendor-reported).
• Recovers 8–12 hours a week of manager time from manual counts (vendor-reported).

Take the exact figures with salt, but the mechanism is real: counts that take an afternoon by hand happen weekly; counts that take a day happen never. The point of the tooling is to make the discipline survivable — a theme I covered in AI-powered inventory management.

The mistakes that quietly cost you

Nearly every inventory problem traces to one of these:

1. Inconsistent counting — different people, units, or timing.
2. Ignoring the waste log — unrecorded spoilage makes variance look like theft and hides fixable process problems.
3. No recipe costing — kills variance analysis before it starts.
4. No FIFO — avoidable spoilage straight into COGS.
5. Weak par discipline — over-ordering high-value perishables ties up cash and feeds the dumpster.

Disclosure: I work at Katalyst, and the version of this that scales is one where item sales, recipes, and inventory live in one system, so theoretical usage and variance fall out as a weekly report instead of a spreadsheet marathon — that's the analytics we build. But the playbook doesn't need us: set pars, count the A-items consistently, run FIFO, cost your recipes, and watch the variance. Inventory is the most controllable big lever on your food cost. It just refuses to be controlled from memory.

Related Katalyst products

• See Katalyst analytics