Menu prices up 3.5%, wholesale food down 2.6%: Test Whether Your Food Cost Percentage Actually Fell

September 2026 CPI figures put menu prices up 3.5% year over year, while the National Restaurant Association reports wholesale food prices 2.6% below September 2025. If you took price with the market, your food cost percentage should have fallen about 1.8 points on a 30 percent base. Whatever did not fall is usage, and usage has a dollar value.

Why this matters now

The Bureau of Labor Statistics released the September 2026 Consumer Price Index on 2026-10-14. All items rose 0.6 percent in the month and 3.7 percent over twelve months. The food away from home index, which is the government's measure of menu prices, rose 0.2 percent in September and 3.5 percent over the year, with full service meals up 0.3 percent and limited service meals up 0.2 percent in the month. Food at home rose 1.8 percent over the year and fell 0.1 percent in September. Energy rose 20.4 percent over the year and gasoline 34.2 percent.

The National Restaurant Association's menu prices indicator, updated 2026-10-14, splits the 3.5 percent into 3.8 percent for full service and 3.2 percent for limited service, and notes that grocery prices rose at roughly half the pace of menu prices. Its food costs indicator, updated 2026-10-15, reports that the Producer Price Index for All Foods rose 0.6 percent between August and September after falling 1.3 percent in July and 0.6 percent in August, and that average wholesale food prices remained 2.6 percent below their September 2025 level. The same page shows how uneven the month was: finfish up 59.1 percent, fresh vegetables up 36.9 percent, fresh fruit up 8.6 percent, beef and veal up 2.6 percent, wheat flour up 2.1 percent.

The BLS Producer Price Index release of 2026-10-15 adds the other direction. Its final demand foods index rose 1.2 percent in September after declines of 0.3 percent in August and 1.0 percent in July, while chicken egg prices dropped 39.7 percent and raw milk fell 12.3 percent. A breakfast concept and a seafood concept lived through very different Septembers.

Two industry reads close the loop. Nation's Restaurant News published a Restaurant Business analysis by Jonathan Maze on 2026-10-14 showing hourly wages up 3.0 percent against the 3.5 percent menu increase, with quick service traffic down 2.6 percent in the third quarter. MarginEdge's September sales report, published 2026-10-12, put food costs at an average of 29 percent of sales in September, a slight increase from August, with full service sales down 2.23 percent year over year. Read together: menus went up, wholesale food came down, and the average food cost percentage still rose. That combination is the signal this article is about.

October is when the test is cheapest to run. Operators on a 13 period calendar close Period 10 at the end of the month, calendar month operators close October, and both are deciding holiday menu prices and 2027 budgets in the next few weeks. The published coverage reports the indexes. It does not show an operator how to turn two indexes into an expected food cost percentage for one location and split the difference into price, cost, and usage. That is the gap this post fills.

What the two indexes say your food cost percentage should have done

Food cost percentage is food cost divided by food sales for the same period. Menu price changes move the denominator. Purchase price changes move the numerator. If covers and usage per cover are unchanged, the expected percentage after both moves is last year's percentage multiplied by one plus the purchase price change, divided by one plus the menu price change.

With the September national figures, that factor is 0.974 divided by 1.035, or 0.941. A location that ran 30.0 percent in the same period last year should run about 28.2 percent now. A 28 percent location should be near 26.3, a 32 percent location near 30.1, and a 35 percent location near 32.9. The drop is close to 1.8 points at a 30 percent base, and it happens with no operational improvement at all. It is the market handing you margin.

Here is a worked example for one full service location, comparing Period 10 of 2025 with Period 10 of 2026, both four weeks. In 2025 the location served 10,000 covers at an average food check of $20.00, so food sales were $200,000, and food cost was $60,000, or 30.00 percent. In 2026 covers fell 2 percent to 9,800, the average food check rose 3.5 percent to $20.70, food sales came in at $202,860, and food cost was $60,250, or 29.70 percent. The percentage fell by 0.30 points, which looks like a good month. The decomposition says otherwise.

In dollars: 2025 food cost per cover was $6.00. At 2.6 percent lower purchase prices, expected cost per cover is $5.84, and expected food cost for 9,800 covers is $57,271. Actual food cost was $60,250. The gap is $2,979 for the period, or about 30 cents per cover, and it is usage: waste, portioning, theft, a count error, or a mix shift toward higher cost items. At one location across 13 periods that is roughly $38,700 a year. The headline percentage improved and still hid a five figure leak.

Volume deserves one sentence. When covers fall and usage per cover holds, numerator and denominator shrink together and the ratio does not move. If your percentage worsened on lower covers, the cause is not the lower covers. It is prep levels, par stocks, and ordering that did not fall with them, which the test records as usage.

Build your own two indexes from the POS and the invoices

The national figures are the right starting benchmark and the wrong final answer. Your menu did not move exactly 3.5 percent and your basket did not move exactly 2.6 percent. Replace both with numbers from your own systems before anyone acts on the residual.

Your menu price index

Export item sales by quantity and price for both periods from the POS. Take the top 20 to 30 items by quantity, which usually cover most food revenue. For each item, multiply this period's quantity by this period's price and by last period's price, then divide the two totals. The result is the menu price change weighted by what you actually sold. An item whose portion changed is a new item, not a price change; drop it from the index and let it show up in mix. Comps, voids, and discounts come out of the denominator before the ratio runs, and so do sales tax, tips, and gift card loads. The POS reconciliation guide covers how to get a clean net food sales figure from Toast or Square into the books.

Your purchase price index

Pull invoice lines for the top 30 SKUs by spend from your distributor portals or your invoice processing tool. Match each SKU across the two periods on pack size, and compute the same quantity weighted ratio: this period's quantity times this period's unit price, divided by this period's quantity times last period's unit price. Items that changed pack size get converted to a common unit first. Delivery and fuel surcharges deserve a decision. With gasoline up 34.2 percent over the year, surcharge lines grew in most markets; whether you code them to food cost or to occupancy, do it the same way in both periods.

Your actual food cost

Food cost for the period is beginning inventory plus food purchases less ending inventory, using counts taken at the period boundaries. The flow of consumption is what you compare with sales for the same dates. Employee meals and promotional comps are a policy question: if you reclassify them out of food cost to a benefits or marketing line, do it in both periods. A four week period compared with a five week calendar month is not a comparison, so align the dates before anything else.

The residual, and what to do with it

Run the decomposition with your own two indexes. The residual that remains is usage and mix. Split mix out if you can: a theoretical food cost built from recipes and the POS mix tells you what food cost should have been at this period's mix and this period's prices, and the distance from theoretical to actual is pure usage. Without recipe costing, read the residual by category. September's wholesale moves were uneven, with eggs and dairy down sharply and seafood and produce up, so a category view will often show the proteins line carrying the leak while produce is explained by price. The multi-location restaurant P&L guide covers how to lay this out by location so that one store's residual does not hide inside a group average.

Accounting controls for the food cost percentage test

Source data: net food sales by period and location from the POS after comps, voids, tax, and tips are removed; item sales quantities and prices for both periods; invoice lines by SKU with quantity, pack size, and unit price; beginning and ending inventory counts with count dates and the names of the counters; the published CPI and PPI releases and the Association indicators with their dates; and the recipe file if theoretical cost is in use. Each export is saved with the date it was pulled, because POS and distributor data get restated.

Calculation: a versioned worksheet with formula cells only. One tab computes the menu price index and the purchase price index from the exports. One tab computes period food cost from the inventory roll forward. One tab runs the four step decomposition and the per cover dollar gap. Nothing is typed over. The debit and credit side matters here: purchases are debits to inventory, an asset measured at a point in time, with the credit to accounts payable; consumption is a debit to food cost, an expense flow for the period, with the credit to inventory. Sales are a credit flow. The percentage divides a debit flow by a credit flow for identical dates, and an overstated ending count understates the debit flow and flatters the percentage.

Review: someone other than the person who built the worksheet confirms that the sales denominator excludes tax, tips, and gift card loads, traces the inventory counts to the signed count sheets and checks the dates, re-performs the two index calculations for five items each, confirms that surcharge and comp treatment match across periods, and reconciles period food cost in the worksheet to the general ledger food cost accounts. The review also compares each location's residual to the prior period's and asks for a written explanation of any location whose usage residual exceeds one point.

Decision: the operator decides what the residual gets. Options include a portion audit on the top ten items, a tighter par and prep sheet for lower covers, a waste log for two weeks, a camera review on receiving, a recipe update where the mix has shifted, or a menu price move where the index shows the location lagged the market. Each decision gets a date, an owner, and a re-test period. A residual that is discussed and not assigned will be the same size next period.

Limits and common mistakes

First, national averages are not your basket. The 2.6 percent decline covers all foods. A seafood or produce heavy concept may have an expected food cost percentage that rose in September, and running the test with the national factor would blame the kitchen for a market move. Build your own index before you act.

Second, the tailwind may be ending. The PPI for All Foods rose 0.6 percent between August and September, the BLS final demand foods index rose 1.2 percent, and energy prices up 20.4 percent over the year feed into freight and packaging. Treat the 2.6 percent as a description of the past twelve months, not a forecast for the next three.

Third, period mismatch. Comparing a 4 week period with a calendar month, or a period with a holiday against one without, produces a residual that is calendar, not usage. Align dates first and keep trading day counts in the file.

Fourth, the denominator. Gross sales from the POS include sales tax, tips on card, and gift card activations. A denominator that includes any of them understates the percentage and hides usage.

Fifth, the count. A late or generous ending inventory count moves food cost into the next period. Two consecutive periods that swing in opposite directions usually point to the count, not the kitchen.

Sixth, food cost alone is half of prime cost. Labor moved too. The Florida minimum wage re-baseline covers the labor side of the same period comparison, and both belong on the same page before a menu price decision. Food cost and prime cost trackers that follow this layout are in the FinBoard templates library.

Finally, this article covers the arithmetic and the accounting, not vendor contracts or legal matters. Check contract pricing and rebate terms with your distributor before treating a purchase price index as final.

Frequently Asked Questions

How much should my food cost percentage have fallen if menu prices rose 3.5 percent and wholesale food fell 2.6 percent?

Multiply last year's percentage by 0.974 and divide by 1.035, a combined factor of about 0.941. A 30.0 percent food cost should be near 28.2 percent with the same covers and usage, a drop of about 1.8 points. Any shortfall against that expected figure is usage or mix.

What is the formula for expected food cost percentage after price changes?

Expected food cost percentage equals last period's percentage multiplied by one plus the purchase price change, divided by one plus the menu price change. Menu prices move the sales denominator and purchase prices move the food cost numerator, while equal changes in covers cancel out when usage per cover is unchanged.

Should I use the national CPI and PPI figures or my own numbers for the test?

Use the national figures as the first benchmark, then replace them with a menu price index from your POS item sales and a purchase price index from your invoice lines. Your basket and your pricing differ from the national average, and the residual is only meaningful once both indexes are your own.

Why did my food cost percentage go up when covers went down?

Lower covers alone do not move the percentage, because food cost and sales fall together when usage per cover holds. A higher percentage on lower covers means prep levels, par stocks, or ordering did not fall with volume, and the decomposition records that as a usage residual with a dollar value.

FinBoard brings each location's QuickBooks file into one view with food cost percent, prime cost, and period over period comparison on the same calendar, with drill-down to the purchase and sales entries behind every number, so the test above is a report you run each period rather than a spreadsheet you rebuild. See how FinBoard supports restaurant groups.