PAR level: a QSR operator's guide

Most operators learn the word "PAR" without a clear definition of it. It gets used to mean "how much we usually order," "how much to have on the shelf," and "how much we need for the week" — sometimes in the same conversation. This article nails down what PAR is, what it's for, and why the number you wrote on a clipboard last summer is probably lying to you by now.

What PAR actually is

PAR is the target quantity of an item you want on hand at the moment you count. If your PAR for vanilla oat milk is 45 half-gallons, that means: when I walk to the walk-in with a clipboard, I want to see 45 half-gallons — no more, no less. If I see fewer, I order up to 45. If I see more, I ordered too much last time.

That's it. It's a number, set per item, and it answers one question: how much should be on the shelf right now?

It does not answer: how much to order, how much you'll use this week, or how much to prep. Those are downstream calculations that use PAR as an input.

Why PAR exists

The whole point of inventory work is to avoid two bad outcomes:

1. You run out. You 86 an item mid-shift. You lose the sale, you frustrate the customer, and you put your staff in a bad spot.
2. You sit on dead stock. The product expires or goes stale before you sell it. You threw money in the dumpster.

For a restaurant that sells mostly non-perishables — a bar with kegs that last months, a steakhouse with items that freeze — the downside of overordering is small, so operators lean toward high PARs. The cost of a 86 is higher than the cost of sitting on two extra kegs.

For QSR with a 3–7 day shelf life on most SKUs — bakeries, coffee shops, juice bars — the cost of overordering is real. Oat milk expires, dough proofing gets out of sync, croissants get sold at half-off or thrown out. You can't just set PAR high and sleep well.

The job of PAR is to land in the middle: enough buffer to handle a normal bad day, not so much that you're eating spoilage.

How PAR relates to ordering cadence

PAR depends on how often you order. The basic mental model:

PAR ≈ (daily usage × days between orders) + safety buffer

If you order once a week and sell 6 oat milk units a day on average, your baseline PAR is ~42 (6 × 7). Add 3–5 for the bad days and you're at a PAR of 45–47.

If you order twice a week, you only need to cover 3–4 days of usage, so PAR drops to ~20–24.

This is why operators who switch from weekly to twice-weekly ordering suddenly have "too much inventory" complaints — the PARs never got updated to match the new cadence.

Why static PARs go stale

The dirty secret of PAR is that it's not a static number. Sales patterns drift:

• Menu changes. Launch a new oat milk latte and your oat milk usage doubles. The PAR you set when oat milk was only in drip coffee is wrong within two weeks.
• Seasonality. A coffee shop near a college goes dead in July. A bagel shop near an office district tanks in August. A bakery next to a Little League field triples on Saturdays in spring.
• Neighborhood shifts. A new competitor opens across the street. A nearby restaurant closes and sends their regulars to you. A construction project eats your foot traffic for six months.
• Your own ops. A new register layout that prompts the upsell. A new social post that went viral for a specific item. A staff change that shifts which items get suggested.

All of these are normal. None of them get reflected in a PAR sheet that was set in January and never touched.

The minimum viable PAR workflow

If you have nothing today, here's the smallest version that works:

1. Pick a counting cadence. Once a week is enough for most QSRs. Count Monday morning before the truck arrives.
2. Set an initial PAR. Use rough daily usage × days between orders + 25% buffer. Write it on the same sheet you count on.
3. Order up to PAR. Count, subtract from PAR, round up to the next whole case, order that much.
4. Revisit every 4–6 weeks. If you're running out of something on day 5 consistently, PAR is too low. If you're throwing out more than 2–3 units, PAR is too high.

This is the same work every inventory SaaS does, minus the software. The software's job is to do steps 2 and 4 for you using real sales data instead of memory.

Where Par Inventory fits

Par Inventory reads your Toast menu-item sales for the trailing 60 days, maps each menu item to the inventory items it consumes (one oat milk half-gallon feeds four lattes and two matchas), and recomputes PAR every time you run the agent. You still count the shelf. You still confirm the order. The PAR number is the only thing we're automating — because it's the only piece that requires looking at 60 days of data and doing pattern-matching.