Restaurant finance and labour, defined with their formulas

Prime cost

Cost of goods sold plus total labour cost, divided by net sales.

Formula. (COGS + labour cost) ÷ net sales × 100

Why it matters. Food and labour are the two costs a restaurant can actually control week to week. Rent does not move; the rota does. Prime cost is the single ratio that most reliably predicts whether a location makes money, which is why operators quote it rather than net margin.

Worked example. In the pilot estate, August 2026: $221,125.29 COGS + $209,459.71 labour = $431,585.00, over $857,985.42 net sales = 50.2%.

Target. Under 65% for most formats. Above 70% a location is usually losing money on volume it is working hard to produce.

Where it goes wrong. Operating expenses are excluded. A prime cost that quietly includes rent and utilities is not the same number, and comparing the two across systems produces a spread that does not exist.

Labour cost percentage

Reported labour cost divided by net sales.

Formula. labour cost ÷ net sales × 100

Why it matters. It is the fastest-moving half of prime cost and the one a manager can change this week by publishing a different rota.

Worked example. $209,459.71 labour over $857,985.42 net sales = 24.4%.

Target. 25–30% for most full-service and fast-casual formats.

Where it goes wrong. It only covers hours that carry a cost. If some of your staff exist on the POS but not on the roster, their hours reach no report — and your labour percentage reads lower than reality.

Labour cost coverage

The share of hours actually worked that carry a cost in your reporting.

Formula. costed hours ÷ total hours worked × 100

Why it matters. This is the number that decides whether every other labour figure is worth reading. It is almost never published by reporting tools, which is precisely why labour percentages across the industry are quietly understated.

Worked example. In the pilot estate, 12,784 costed hours out of 18,826 worked = 67.9%. 6,043 hours carried no cost at all.

Target. 100%. Anything below it means unmapped POS employees.

Where it goes wrong. Missing hours make labour look lower, never higher — which is what makes it dangerous. Nobody investigates a good number.

Unmapped POS employee

Someone the point of sale knows about who is not linked to a roster record.

Why it matters. Staff get added at the till during a rush. Names differ between systems. Old logins linger after someone leaves. The POS list and the roster are never identical, and everyone on the POS list without a roster link has their hours and cost reaching no report.

Worked example. 257 unmapped employees in the pilot estate, carrying 1,253 unattributed hours in 90 days.

Target. Zero.

Where it goes wrong. Linking one is a one-time action that applies retrospectively to every hour they have ever worked. It is the highest-value setup task in any system of this kind.

Food cost percentage

Cost of goods sold divided by net sales.

Formula. COGS ÷ net sales × 100

Why it matters. The slower-moving half of prime cost. It is driven by purchasing, portioning, waste and menu mix rather than by this week's rota.

Worked example. $221,125.29 COGS over $857,985.42 net sales = 25.8%.

Target. Varies by format — a pizza operation and a bar do not share a target.

Where it goes wrong. COGS is what you bought, not what you used, unless you are counting stock. An in-hand figure derived from purchases and sales is an estimate and should be labelled as one.

Sales per labour hour

Net sales divided by hours worked. Often written SPLH.

Formula. net sales ÷ hours worked

Why it matters. It is the fairest way to compare a high-volume site with a low average check against a low-volume site with a high one. Labour percentage alone flatters whichever has the higher check.

Worked example. $857,985.42 over 18,826 hours = $45.57 per labour hour. Across the estate, individual stores ranged from $23.69 to $267.70.

Target. Higher is better; the useful comparison is against your own best store, not an industry figure.

Where it goes wrong. A store showing an implausibly high SPLH usually has unmapped employees rather than exceptional productivity.

Net sales

Gross sales less sales tax, discounts and comps.

Formula. gross sales − tax − discounts − comps

Why it matters. It is the denominator of nearly every ratio a restaurant runs on. If two of your locations define it differently, every percentage you compare between them is wrong.

Worked example. The P&L statement runs gross sales through sales tax, discounts and COGS to gross profit, then through operating expenses to EBITDA and net income.

Where it goes wrong. The most common cause of an unexplainable store-versus-store gap is a definition difference, not a performance difference.

COGS (cost of goods sold)

The cost of the food and beverage sold in the period.

Formula. opening stock + purchases − closing stock

Why it matters. Together with labour it forms prime cost. Without a counted closing stock, most systems approximate it from purchases.

Worked example. Purchasing across the pilot estate: $4,177,783.00 of spend, 91,286 units received, 287 items across 15 categories.

Target. By format.

Where it goes wrong. COGS adjustments belong in their own ledger. Counting a supplier invoice once as COGS and again as an operating expense is the most common way a restaurant P&L quietly stops reconciling.

Overtime exposure

The projected cost of overtime hours in a scheduled period, before the period closes.

Formula. projected OT hours × the applicable premium rate

Why it matters. Overtime is almost never scheduled deliberately. It accumulates across a week nobody is totalling, and surfaces at payroll after every hour has been worked.

Worked example. A twelve-store pilot estate carried a material, avoidable overrun in a single period — priced and named before the rota was even published.

Target. Under 5% of scheduled hours for most groups.

Where it goes wrong. Hours accumulate across locations. A person working four shifts at one store and two at another is over 40 hours in the eyes of the law and under 40 in each store's own report.

Schedule health

A single graded score for a scheduled period, with deductions shown per store.

Method. Weighted toward overtime risk, with the deduction shown per store rather than folded into a single estate-wide number.

Why it matters. It turns an abstract scheduling problem into a number a district manager can act on, and shows which store cost them the grade rather than just the total.

Worked example. In a twelve-store pilot graded C overall, the score identified one location as responsible for most of the shortfall within minutes, rather than an average hiding it for a month.

Target. Grade A.

Where it goes wrong. The cap matters: without it, one catastrophic store would hide every other problem behind a score of zero.

Planned versus actual labour

The scheduled rota priced out, compared against what the POS reported.

Formula. planned labour cost − actual labour cost

Why it matters. A rota is a forecast. The variance between what you scheduled and what was worked is where overtime, no-shows, extended shifts and unmapped employees all become visible in one place.

Worked example. The variance table compares scheduled shifts against POS and timeclock records per employee.

Target. Within budget.

Where it goes wrong. Actual labour cost excludes uncosted hours rather than estimating them, so a large favourable variance can mean unmapped employees rather than efficiency.

Labour budget

A weekly cash allowance for labour, allocated per store.

Formula. forecast sales × target labour %

Why it matters. Without a budget there is no variance, and without a variance nobody knows on Wednesday that the week is already lost.

Worked example. The allocator forecasts the week from actual POS history — the average of the three preceding weeks and the same week last year, with empty weeks left out rather than counted as zero — then sets labour at 27% of that forecast.

Target. Inside the 25–30% band most stores run at.

Where it goes wrong. Forecasting from an average that counts a closed week as zero sales will under-budget the following week and cause the exact overtime it was meant to prevent.

Budget pace

Spend to date against the share of the period elapsed.

Formula. spend to date ÷ (budget × % of period elapsed)

Why it matters. It answers the only budget question that matters mid-month: is this going to be fine, or is it already not?

Worked example. Pilot estate, August: $52,902.92 over budget at 81% of the period elapsed.

Target. 100% or below.

Where it goes wrong. A store with no budget set for the month has no pace, and the dashboard should say so rather than showing a comforting blank.

ACT and PLAN

Labels distinguishing what actually happened from what was scheduled or budgeted.

Why it matters. Most reporting tools blend the two and produce a figure that is neither. Every tile in Nexora states the basis it was calculated on.

Worked example. Average hourly wage $16.38 (ACT, over 12,784 costed hours). Overtime share of scheduled hours, tracked against plan across the estate.

Where it goes wrong. A number without its basis is not checkable, and an operator who cannot check a number will not act on it.

Store comparison

Every location ranked on identical measures.

Why it matters. The gap that costs the most in a restaurant group is between stores, not between months. A group average hides the location losing money.

Worked example. Pilot estate, August 2026: prime cost from 23.60% at Parkview to 75.84% at Millbrook — a 52.2-point spread on the same brand, menu and suppliers.

Where it goes wrong. Stores showing 0.00% labour have unmapped POS employees, which is why their prime cost reads low. A comparison view that fills that gap with an assumption ranks your data quality, not your operations.

EBITDA

Earnings before interest, tax, depreciation and amortisation.

Formula. net sales − COGS − labour − operating expenses

Why it matters. It is the figure a buyer, a lender or a franchisor will ask for, and the one a store-level P&L should reach without a spreadsheet in between.

Worked example. The statement runs gross sales through sales tax, discounts, COGS, gross profit and operating expenses to EBITDA and net income, with the period beside the year to date.

Where it goes wrong. EBITDA at store level is only as honest as the operating expenses recorded against that store. Group costs allocated arbitrarily make every store's figure arguable.

Tip credit

An allowance letting an employer pay tipped staff a lower cash wage, with tips making up the difference.

Formula. minimum wage − tipped cash wage

Why it matters. It changes labour cost structurally, not marginally. A location in a no-tip-credit state will always show a higher labour percentage than one in a tip-credit state, whatever the management.

Worked example. Federal: $7.25 minimum, up to $5.12 tip credit, $2.13 minimum cash wage. Alaska, California, Minnesota, Montana, Nevada, Oregon and Washington permit no tip credit at all.

Where it goes wrong. Your true labour cost is the cash wage you pay; your exposure if tips fall short is the full minimum. Both numbers have to exist in your reporting.

Daily overtime

Overtime owed after a set number of hours in a single day rather than only after 40 in a week.

Why it matters. A scheduling system that only totals weekly hours will miss it entirely, and it surfaces at payroll.

Worked example. Alaska, California, Colorado and Nevada apply daily overtime rules.

Where it goes wrong. A double shift covering a call-out can trigger daily overtime in a week that never reaches 40 hours.

Predictive scheduling

Laws requiring advance notice of schedules and premium pay for late changes.

Why it matters. They turn publishing the rota late into a direct cost rather than an inconvenience.

Worked example. In force at state or major-city level in California, Illinois, Minnesota, New York, Oregon, Pennsylvania and Washington among others.

Where it goes wrong. Requirements vary by city as well as state. Confirm for each location with counsel rather than assuming the state rule covers you.

Role-based access

Access defined by what a person may do and where, rather than by which buttons are hidden.

Model. Access is a function of who you are, what role you hold, and which stores that role covers — not which buttons happen to be visible.

Why it matters. A franchisee will not accept their numbers sitting in the same database as a competitor's unless the separation is real. Hiding a button is not real.

Worked example. Three altitudes across the pilot estate: 7 super admins, 5 admins and 23 store managers, 35 people with access in total.

Where it goes wrong. The only version that survives someone typing a URL directly is one where the query never asks for the data. Interface-level hiding does not.

Average check

Net sales divided by the number of orders.

Formula. net sales ÷ orders

Why it matters. It sets the ceiling on what a labour hour can produce, which is why comparing labour percentage between formats without it is meaningless.

Worked example. Pilot estate: $19.34 average check across 44,353 orders. Individual stores ranged from $16.75 to $22.16.

Target. By format.

Where it goes wrong. A rising average check with falling orders is a different business problem from a rising average check with steady orders, and the ratio alone will not tell you which you have.

Cost per order

Labour cost divided by the number of orders.

Formula. labour cost ÷ orders

Why it matters. It converts an abstract labour percentage into a per-transaction figure a general manager can hold in their head against the average check.

Worked example. Pilot estate: $9.73 cost per order against a $19.34 average check.

Where it goes wrong. It is a labour figure, not a full unit cost. It does not include the food in the order.

Labour minutes per order

Total minutes worked divided by orders.

Formula. (hours worked × 60) ÷ orders

Why it matters. The one labour measure that is independent of both wage rates and menu price, which makes it the fairest comparison between locations in different states.

Worked example. Pilot estate: 25.5 minutes per order across 44,353 orders.

Where it goes wrong. It rewards throughput, so a site with a long-dwell format will look worse without deserving to. Compare like with like.

Why definitions are the whole game

Operators compare across systems, and a prime cost that quietly includes operating expenses is not the same number as one that does not. When two of your locations define net sales differently, every percentage you compare between them is wrong — and the league table you build from them ranks your accounting rather than your operations.

That is why every ratio in Nexora is stated with its formula on the screen it appears on, and why every tile states the basis it was calculated on — ACT for measured, PLAN for scheduled or budgeted. A number without its basis is not checkable, and an operator who cannot check a number will not act on it.

How Nexora defines each headline figure → · Work out your own prime cost →