Sales Per Labor Hour: The Scheduling Metric That Beats Labor Cost %

The problem with labor cost percentage

Almost every restaurant reviews labor as a percentage of sales, and almost every restaurant then struggles to turn that review into a schedule. The reason is structural rather than cultural: labor cost percentage is a blended number, and blended numbers do not decompose into decisions.

Two things move it. The hours you deployed, and the cost of those hours. A shift that lands at 34% could be four hours overstaffed at ordinary wage rates, or perfectly staffed with three of your most expensive people on the floor at the same time. It could also be neither, and simply reflect a wage increase you agreed to two months ago. The percentage looks the same in all three cases, and only one of them is a scheduling problem.

There is a second issue. Percentage is a ratio of two totals, and totals need a period. By the time you have a defensible weekly figure the week is over. It is genuinely useful for the P&L review, for budgeting, and for comparing this quarter to last. It is close to useless at four o'clock on a Tuesday when you are deciding whether to send someone home.

What sales per labor hour does instead

Sales per labor hour takes net sales for a period and divides by the total hours worked in that period — front of house, back of house and management combined. The output is a dollar figure: this is what an hour of staffed time produced.

Because wage rates do not appear anywhere in the calculation, the number moves for exactly one reason: the relationship between sales and hours. That is the relationship a manager building a rota actually controls, which is what makes it a scheduling metric rather than a reporting one.

Consider a mid-sized casual restaurant. Tuesday dinner: $4,200 in net sales, 46 front of house hours, 34 kitchen hours, 8 management hours. Eighty-eight hours, $47.73 per hour. Saturday dinner: $11,800 across 148 hours, $79.73 per hour. The Saturday shift is running 67% more productively than the Tuesday one, in the same building, with the same menu and the same pay rates. That gap will not appear in any weekly labor percentage, because the week averages the two together into a figure that describes neither.

Once you have that comparison, the conversation changes. It is no longer "labor was high this week" — a statement nobody can act on — but "Tuesday dinner needs eight fewer hours and Saturday might justify four more."

Turning a percentage target into an hours target

The two metrics are not rivals; they are the same information arranged differently. The link between them is a small piece of algebra that most operators never see written down, and it is the most useful thing in this article.

Labor cost percentage equals your average hourly labor cost divided by your sales per labor hour. Rearranged, your required sales per labor hour equals your average hourly cost divided by your target labor percentage.

Suppose your fully loaded average hourly cost — wages plus payroll taxes plus benefits, divided by hours worked — is $19, and you have decided labor should run at 28% of sales. Divide $19 by 0.28 and you get $68. That is your target. Every hour you roster needs to be backed by $68 of forecast sales.

Now the rota builds itself. A Friday forecast at $9,500 supports roughly 140 hours. A Monday forecast at $3,600 supports 53. You are no longer scheduling to a feeling and checking the percentage afterwards; you are scheduling to a number derived from the percentage you already agreed. The labor cost percentage calculator handles the other direction when you need to check the result after the fact.

This also settles a common argument. When wage rates rise, labor percentage rises and sales per labor hour does not move. The scheduler did nothing wrong. What has to change is the target: at $21 an hour and the same 28% ambition, you now need $75 per labor hour, which means either a higher check average or a genuinely tighter rota. Separating the two makes that a strategic decision rather than a blame conversation.

Realistic targets by format

Benchmarks are worth less here than in most metrics, because the number is so sensitive to service model and to local wage structure. As orientation: coffee and bakery counters tend to run $40 to $60, quick service and fast casual $50 to $75, casual full service $50 to $70, bar-led venues $65 to $95, and fine dining $70 to $110. Delivery-heavy pizza operations sit lower, often $45 to $65, because driver hours generate revenue slowly.

None of that should override the calculation in the previous section. A restaurant paying $24 an hour in a high-wage city and a restaurant paying $15 an hour elsewhere are not aiming at the same figure, and pretending otherwise produces either complacency or a rota nobody can staff.

Use the ranges for one thing only: a sanity check. If your own arithmetic says you need $110 per labor hour to hit your target percentage and you run a casual dining room, the problem is not your schedule. It is your check average, your wage structure or your target — and no amount of rota tinkering will close a gap that large.

How to implement it without a project

This does not need software, a consultant, or a new POS module. It needs two numbers per shift, and most operations already have both.

• Pull net sales per shift from the POS, excluding tax. Define the shift boundaries once and never move them.
• Pull hours worked per shift from clock-in data, not from the published rota. The difference between the two is where most overruns hide.
• Include management hours spent on site. Excluding them inflates the figure by ten to twenty per cent and makes any comparison between sites worthless.
• Record it on the same sheet as covers so that a fall in the metric can immediately be attributed to hours or to volume.
• Give it eight weeks before you act on anything. One unusual Tuesday is not a pattern, and reacting to noise is how good schedules get broken.

The near-universal finding after eight weeks is not that the restaurant is overstaffed overall. It is that it is overstaffed for the first hour and the last hour of most shifts — everybody starting at the same time, everybody finishing at the same time. That is a shape problem, and staggering start times fixes it without removing a single person from the floor at peak.

Where the metric points at a demand problem rather than a rota problem, the fix usually sits in throughput: more covers through the same staffed hours. Working on table turnover raises the numerator without touching the denominator, which is the only kind of improvement that is unambiguously good.

The failure mode to watch for

Any ratio can be improved by damaging its denominator, and this one is no exception. Cut hours hard enough and sales per labor hour climbs immediately — right up to the point where service degrades, covers fall, and the numerator follows the denominator down.

The warning sign is easy to read. If the metric is improving while covers, check average and guest feedback are all deteriorating, you have not found productivity. You have found the edge of your capacity and gone past it. Genuine improvement shows up as flat or rising sales on fewer hours, never as falling sales on far fewer hours.

The other limitation is that revenue per hour says nothing about what that revenue cost to produce. A menu pushing expensive, low-margin dishes will post a flattering figure while the operation makes less money than a site with half the check average. Read it next to prime cost, keep labor percentage in the period review for the wage-rate story it tells, and use sales per labor hour for what it is genuinely best at: deciding, before the shift rather than after it, how many people should be there.

Frequently asked questions

Is sales per labor hour better than labor cost percentage?

Better for scheduling, not better overall. It isolates the hours decision from wage rates, which is what a manager building a rota controls. Labor cost percentage remains the right number for the P&L review, because it is the only one of the two that reflects wage inflation.

What counts as an hour for this calculation?

Every paid hour worked on site — front of house, kitchen, and salaried management time genuinely spent in the building. Use clock-in data rather than the published rota, since the gap between scheduled and worked hours is often where the overrun sits.

How do I set my own target?

Divide your fully loaded average hourly labor cost by the labor percentage you intend to run. At $19 an hour and a 28% target, you need $68 per labor hour. Recalculate it whenever wage rates change materially, because the target moves even though the metric does not.

Does the metric work for kitchen staff on their own?

Yes, and the split is informative. Kitchen hours are far less elastic than service hours because a station must be covered whether it sells six covers or sixty. Tracking both separately tells you whether a falling figure is a rota problem or a volume problem.

What if my sales are seasonal?

Compare like with like: the same day of the week across consecutive weeks, and the same month year on year. Seasonality is exactly why a single benchmark number is unhelpful and why the shape of your own trend matters more than anyone else's average.

Should delivery sales be included?

Include them only if the hours that produced them are also included, and be aware that delivery revenue arrives with commission already deducted from your margin. If delivery is a meaningful share of sales, track it as a separate line so that a shift in channel mix does not look like a change in productivity.

How long before improvements show up?

Start-time staggering shows in the next scheduled week, because the change is immediate and entirely within your control. Check average improvements take a menu cycle. Anything that depends on hiring or training moves over a quarter.

Run the numbers

Use the free Sales Per Labor Hour Calculator to apply everything above to your own figures.