Blog

Do You Need Inventory Counts to Forecast Prep?

Sep 11
decor image

By Matt Wampler, CEO of ClearCOGS

Quick Answer: No. An inventory count tells you what is on your shelves right now, which is a record of what already happened. A prep forecast predicts how much of each item you will sell tomorrow, which is what determines how much to make. Counting and forecasting answer different questions. Counts are valuable for measuring variance and valuing what you own, but they are not a prerequisite for producing an accurate prep number.

Why this question comes up

Most operators learned inventory the same way. You count everything on a set schedule, you enter the numbers into a system, and the system returns usage, food cost, and a suggested order. The prep sheet, if you got one at all, came out the far end of that process. Nothing worked unless the counts were entered, and entered correctly.

So when a brand starts looking at forecasting, the instinct is to treat it as the same machine with a nicer interface. The question underneath is usually practical rather than theoretical: are we about to ask every general manager in the system to start counting again, and will the numbers be any good if we do?

It is a fair question, and the answer changes the shape of the rollout.

What does a restaurant inventory count actually measure?

An inventory count measures the position of your stock at a single moment. Combined with your purchase records, it produces a usage figure for the period between counts.

That is genuinely useful. It tells you what you own, what you spent, and how much product moved. It lets you compare what your recipes say you should have used against what you actually used, which is the only reliable way to find over-portioning, yield loss, spoilage, and theft.

What it does not tell you is what tomorrow looks like. A count is a rear-facing measurement. It describes a period that has already closed.

Why can’t an inventory count tell you what to prep tomorrow?

Because demand does not come from your shelves. It comes from your guests.

The amount of dough you need on Thursday is a function of how many pizzas you will sell on Thursday. That number depends on the day of the week, the season, the weather, a local event, a limited time offer you are running, and the specific trade area around that one location. None of that information lives in a walk-in.

Counting harder does not fix this. If you count twice as often, you get a more precise picture of the past and exactly the same amount of information about the future. Operators feel this as a familiar frustration: the numbers are tighter, the reports are cleaner, and the Thursday prep decision is still a judgment call made by whoever happens to be on shift.

What is the difference between inventory counting and prep forecasting?

Inventory counting compared with prep forecasting
Inventory countingPrep forecasting
Question answeredWhat do we have and what did we use?How much will we sell and how much should we make?
DirectionBackward lookingForward looking
Primary inputPhysical counts and invoicesHistorical item-level sales, seasonality, events, local conditions
Primary outputUsage, variance, food cost, inventory valueProduction quantities by item and by day
Manager workloadRecurring manual counting and data entryReviewing a number that arrives before the shift
Fails whenCounts are skipped, rushed, or inconsistentSales history is thin or the menu changes without notice

The two are complements, not substitutes. The mistake is assuming one is a gateway to the other.

Are inventory counts accurate enough to build decisions on?

This is where the sequencing argument gets sharper, and it is worth being honest about the evidence.

In a study of nearly 370,000 inventory records across 37 stores of a single retailer, researchers at the University of Chicago and Harvard Business School found that 65 percent of those records were inaccurate. The same work found that auditing practices reduced inaccuracy, while complexity in the store environment made it worse (DeHoratius and Raman, Management Science, 2008).

That study is retail, not restaurants, and the comparison is not exact. But a restaurant is a more complex counting environment than a retail shelf, not a simpler one. Product gets portioned, batched into sub recipes, moved between stations, and transformed into something that no longer resembles what arrived on the truck. Half a pan of cheese sauce is a judgment call. A case of brisket is not the same weight as the last case of brisket.

The practical conclusion is not that counting is worthless. It is that a count is a measurement with error in it, and building your daily production decision on top of a measurement with error in it puts the error into every shift.

A forecast built on point of sale transactions has a different failure mode. It reads what was actually sold, item by item, which is the one thing a restaurant records precisely because a guest paid for it.

What should you use inventory counts for instead?

Counting earns its keep when you point it at the questions it is actually good at.

  • Variance. Compare theoretical usage against actual usage to find portioning drift, yield problems, and shrink. This is the highest-value use of a count and it does not need to happen daily.
  • Valuation. Month-end inventory value feeds your financial statements. You need it regardless of how you plan prep.
  • Spot checks on the exceptions. Count the twenty items that carry most of your cost, more often than you count the other two hundred.
  • Confirming a fix. When you change a recipe, a portion tool, or a storage practice, a count tells you whether it worked.

Notice that none of these are daily production decisions. Counting is a diagnostic. Forecasting is an instruction.

How should a multi-unit brand sequence the two?

A useful order of operations for a growing group:

  1. Start with the forecast. Connect your point of sale, map your recipes down to the ingredient level, and get production numbers in front of managers before their shift. This requires no new counting from the stores.
  2. Watch adoption before you watch accuracy. A forecast that is 95 percent accurate and ignored is worth less than one that is 88 percent accurate and followed. Find out whether the number arrives in a format the closing manager will actually use, on the device they actually have.
  3. Layer counting in where it answers a question. Once production is planned, a count tells you something it could not tell you before: whether the gap between what you planned to make and what you actually used is a forecasting problem or an execution problem.
  4. Use the spread across locations. In any group of scale, a handful of operators are running noticeably tighter than the rest. Forecast accuracy reporting shows you which ones, and a targeted count shows you why.

This order matters most in franchise systems, where you often cannot mandate a new process. Asking a franchisee to start counting is asking for labor. Handing a franchisee a number that saves the closing manager fifteen minutes and reduces Sunday waste is offering something. Those are very different conversations, and they tend to produce very different adoption rates.

Frequently Asked Questions

Does forecasting replace inventory management?

No. Forecasting replaces the guesswork in daily production and ordering. You still need inventory records for financial reporting, variance analysis, and vendor reconciliation.

How often should a restaurant count inventory?

It depends on what you want the count to do. Full counts monthly are common for valuation. High-cost or high-movement items are often worth a weekly or twice-weekly spot check. Counting every item every week is usually more effort than the information justifies.

Can you forecast prep without a back-office system?

Yes. Forecasting needs item-level sales history and recipes. Recipes can come from a back-office system, a shared drive, or a spreadsheet. The format matters less than whether the recipes reflect what the kitchen actually does.

What if a large catering order is not in the point of sale?

Known orders that bypass the point of sale should be added on top of the forecast rather than predicted by it. The forecast covers expected walk-in and delivery demand. A confirmed large-format order is not a prediction problem, it is an input.

Will forecasting work if our data is messy?

Usually. Sales transaction data is typically the cleanest data a restaurant has, because it is tied to payment. Recipe data is where the cleanup effort tends to land, and it is a one-time effort rather than a recurring one.

Let’s Talk

Sources

  • DeHoratius, Nicole, and Ananth Raman. Inventory Record Inaccuracy: An Empirical Analysis. Management Science, 54(4), 627–641. April 2008. pubsonline.informs.org