How Many Employees Should I Schedule Each Shift at My Ghost Kitchen?

This vs. the common alternatives

Most ghost kitchen operators arrive at headcount one of four ways, and only one of them survives contact with delivery-only economics.

The habit schedule. Somebody decided in month two that dinner runs six people, and it has run six people ever since. Nobody remembers why. Order volume doubled, then a brand got cut, then a new one launched, and the six never moved. This is the most common approach and the most expensive one, because it fails in both directions simultaneously — you're overstaffed on the Tuesday afternoon flat and underspeed on the Friday crest, and the payroll cost of the first exactly funds the fix for the second. The tell is that nobody on your team can articulate why the number is what it is.

The labor percentage cap. "Keep labor under 28% of sales." This is the standard restaurant control and it's genuinely useful as a guardrail, but it's a lagging report card, not a scheduling instruction. It tells you after the fact whether you overspent; it doesn't tell you on Wednesday how many cooks to put on Saturday. Worse, in a delivery-only house the denominator lies. Gross sales include the 15–30% marketplace commission you never see, so a schedule that looks fine at 26% of gross sales can be underwater at 34% of what actually hits your account. If you use a labor percentage at all, compute it against commission-adjusted net revenue.

Orders-per-labor-hour throughput. This one is real and worth running alongside the gross-profit method. You measure that a cook on a defined station can push, say, 12–18 tickets an hour depending on menu complexity, then divide forecast ticket volume by that rate. It's excellent for capacity planning and for knowing when your line physically breaks. Its weakness is that it's blind to margin. A $9 single-item order and a $46 family bundle both count as one ticket, so a shift heavy on cheap singles will show high throughput and terrible profitability, and the throughput model will happily tell you to staff up for the unprofitable rush.

Gross profit divided by a per-employee target. This is the method. It fixes the labor-percentage problem because gross profit is already net of food cost and, if you compute it correctly, net of marketplace commission. It fixes the throughput problem because it's denominated in dollars, not tickets. And it fixes the habit problem because it produces a defensible number anyone in the room can check.

The honest trade-off: the gross-profit method is a *demand* model, not a *physics* model. It tells you what the shift can afford to pay for; it doesn't know that your fryer has four baskets. Run it as the primary and let throughput act as the floor check — if the division says two people and your ticket forecast says one station can't physically clear the queue, throughput wins and you take the margin hit for that shift. In practice these disagree maybe one shift in ten, and the disagreements are informative on their own.

The same logic transfers to any single-channel, no-front-of-house operation — a commissary bakery on wholesale accounts, a meal-prep subscription kitchen batching Sunday for the week, a catering-only shop. Wherever there's no upsell cushion and no table turn to absorb a soft hire, headcount has to answer directly to gross profit.

How to choose between them

Pick the model by what actually constrains you, not by what's fashionable. The decision usually resolves in three questions.

Do you have at least three months of clean, brand-blended gross profit by shift? If yes, run the gross-profit method as primary. If no — you're a new kitchen, or you just launched two brands and your history is meaningless — start with the throughput model on menu-tested prep times, then transition once you've banked a trailing quarter. Don't try to run the gross-profit division on six weeks of data; you'll be scheduling against launch noise.

Is your binding constraint dollars or physical line capacity? If your fryers, ovens, and expo shelf are the bottleneck — you're turning down orders or throttling on the apps during peaks — throughput planning governs and the gross-profit number becomes the budget check. If you have headroom on equipment and the question is purely "can this shift afford another body," the division governs.

How much variance is there between your best and worst shift? Compute the ratio of your highest-gross-profit shift to your lowest. Under 2×, a static weekly template with two staffing tiers is fine and the sophistication isn't worth the overhead. Over 3× — which is typical for delivery-only, where Friday dinner routinely runs four to six times a Monday lunch — you need per-shift math and you need it recalculated monthly, because a static template is guaranteed wrong on both ends.

Setting the per-employee target itself is the one judgment call the math can't make for you. Get your kitchen managers in a room and name the gross profit a solid employee should throw off on a routine shift — not a peak shift, a routine one. Work backward from wage: if a cook costs you roughly $17 an hour fully loaded across an eight-hour shift, that's about $136 in direct labor, and a target of $180 gives you a thin but real contribution margin per body. Set it too low and you overstaff by construction; set it too high and you burn your crew and start losing tickets to slow tickets, which the apps punish with reduced visibility.

Say the number out loud to the crew, plainly: "In this kitchen, if you clock in, own your station, and move tickets at a fair clip, you're carrying at least $180 a shift in gross profit." That transparency is most of the value. It hands you, your leads, and every line cook the identical measuring stick, which kills the "we've always run six" argument and kills the lead who quietly schedules friends onto the easy nights.

Costs, timelines, and expected impact

What it costs to implement. The method itself is free — it's arithmetic on data you already have. The real cost is analyst time to build the shift-level gross profit table the first time: budget 6–10 hours if your delivery sales live in three separate marketplace dashboards and your food cost lives in a spreadsheet. After that it's 30–60 minutes a month to refresh.

If you want tooling, the market splits by pricing model, and the split matters more than the feature lists. Per-location pricing suits a multi-brand kitchen running one shared roster; per-user pricing suits a lean, stable crew. Restaurant-native scheduling platforms typically run in the $25–$80 per location per month range at the small-business tiers, with free single-location tiers available from several vendors. Per-user tools generally land in the $2–$8 per user per month band depending on whether you add time-and-attendance. Enterprise workforce-management platforms sell by custom quote and are genuinely more machinery than one or two kitchens need. Confirm current pricing directly with each vendor before you budget — these tiers move.

Whatever you buy, the tool executes; it doesn't decide. You show up with the per-employee target and the shift gross-profit table already built, and the software handles publishing, swaps, clock-in, and compliance. Nothing on the market will pick your target number for you.

Timeline to a working schedule. Week one, pull three to six months of gross profit by shift and day of week, blended across every brand on the line. Week one also, hold the target-setting conversation with your leads. Week two, publish the first math-derived schedule and warn the crew that headcount is moving. Weeks three through six, run it and log every shift where the number was visibly wrong in either direction. Week seven, adjust the target based on that log — this is the step most operators skip, and it's the one that turns a formula into a system.

Expected impact. Be realistic. The gain comes from three places and they're unequal in size.

The largest is eliminating structural overstaffing on flat shifts. If you've been running four on a Monday lunch that the math says supports two, you're recovering two shifts of wage every Monday — and the same is usually true of Tuesday and Wednesday lunch. Across a week that's frequently five to eight redundant shifts.

The second is fewer late tickets on peaks, which shows up indirectly. Marketplace algorithms weight prep-time accuracy and late-order rates; a kitchen that stops blowing its quoted times generally sees better placement and fewer refund chargebacks. This gain is real but slow and hard to isolate.

The third is smaller and often negative in month one: better cross-brand utilization. When headcount is set by blended profit rather than per-brand habit, you stop staffing a station for a brand that's only doing four tickets a shift. This one requires cross-training, and cross-training costs you speed before it pays you back.

The counter-cost nobody budgets for: cutting a shift is a real event for the person losing it. If your schedule now swings between two and eight bodies, you need a bench of cross-trained part-timers who understand the model, or you will lose your good people to kitchens offering steadier hours. Communicate the model before you cut anyone's hours with it.

When it doesn't pay off. If you run a single brand, single daypart, with under 2× variance between shifts, the method will confirm what you already do and you'll have spent ten hours to learn that. Run the variance check first.

Implementation and handoff details

Getting this from a spreadsheet into a running weekly rhythm is where most implementations die. The handoff has four moving parts.

Data pull. Export delivery order data by hour, not by day — the daily total is useless for placing bodies. You need the hourly ticket curve, because ghost kitchens don't trickle; they spike violently at the midday app rush, detonate again at dinner, and go flat for long stretches between. Pull gross profit separately: revenue net of marketplace commission, minus food cost, by shift and day of week, blended across every virtual brand sharing the line. Whoever owns this pull should own it permanently — a rotating owner means a stale table within two months.

The division. Shift gross profit ÷ per-employee target = headcount, rounded to the nearest whole body. Round down when the throughput check says the line can absorb it; round up when the shift is a hard peak. Part-timers scale linearly: a four-hour body against an eight-hour, $180 target carries a $90 target, so two of them fill one slot.

Placement against the curve. The division tells you how many; the hourly export tells you when. Stack cooks and a dedicated expo into the two crests, strip to a skeleton through mid-afternoon, hold a tight two-body crew for the late-night courier tail. In a delivery-only house the expo role is non-negotiable at peak — someone has to bag, seal, label, and stage orders for couriers stacking up at the pickup shelf, and a cook trying to do both will drop both. Budget one expo per peak shift before you count cooks.

Weekly variance review. Every Monday, compare last week's actual gross profit per shift against the forecast the schedule was built on. Anything off by more than 20% either direction gets a note. Three consecutive misses in the same direction on the same shift means the trailing average is stale and that shift needs its own recalculation.

Who owns what. The owner or GM sets the per-employee target and revisits it quarterly. A kitchen lead owns the weekly placement against the curve — they know who can run which station. Whoever handles payroll owns the variance review, because they're already in the numbers. Write these three ownerships down; ambiguity here is why schedules revert to habit within a quarter.

Common failure modes. Using gross sales instead of commission-adjusted revenue is the number one error and it inflates every headcount by roughly the marketplace take rate. Second is forgetting prep labor — the person filling low pans mid-service is real headcount and belongs in the count. Third is letting the trailing window go stale after a brand launch or a menu price change; both invalidate the average immediately and you should reset the window rather than wait for it to age out.

Where this connects to the rest of the operation. Headcount math is a RevOps problem wearing an apron. The same discipline that sets a sales rep's quota from territory potential sets a cook's shift target from station output — you define a unit of productive capacity, you measure what it actually produces, and you size the roster from demand rather than from tradition. Multi-unit retail, showroom floors, and pizza kitchens all run variants of the same division; what changes is the denominator and the shape of the demand curve, not the logic. If you already run capacity planning on a sales team, you have the muscle for this.

Related questions

How do I handle a shift the math says needs 2.4 people?

Round to the nearest whole body, then use the throughput check as the tiebreaker. If the ticket forecast says one station clears the queue comfortably, round down to 2 and take the margin. If it's a hard peak, round up to 3. Alternatively, schedule a four-hour part-timer to cover the fractional half.

Should the expo count toward the headcount the formula produces?

Yes. The formula produces total bodies on the shift, not cooks. Budget one expo per peak shift first, then fill remaining slots with cooks and prep. Excluding the expo from the count is a common error that quietly overstaffs every peak by one.

What if two virtual brands peak at different hours?

That's an advantage, not a problem — staggered peaks flatten your curve and improve utilization. Blend the gross profit as usual, but build the hourly placement off the *combined* ticket curve rather than either brand's alone. Cross-train so a cook can flip between brand recipes mid-service.

How does this change if I sublease space in a shared cloud-kitchen facility?

The math is identical, but your gross profit must net out the facility fee and any shared-equipment charges before you divide. Shared facilities also cap how much you can flex physically — you can't stack six cooks into a suite built for three, so the throughput ceiling binds sooner.

Does this work for a catering or meal-prep kitchen instead of delivery?

Yes, with a longer window. Batch operations have predictable, order-ahead demand, so you divide gross profit per production block rather than per service shift. The per-employee target logic is unchanged; the hourly placement step gets replaced by production sequencing.

FAQ

What if my gross profit per shift is inconsistent week to week?

Use a trailing three-to-six-month average for each shift and day of week — that smooths seasonal spikes and slow stretches into a usable baseline. If a single shift still swings wildly after averaging, split it into two sub-shifts with different staffing levels rather than forcing one number onto a bimodal demand pattern.

How do I determine the right gross-profit-per-employee target for my kitchen?

Start by working backward from your fully loaded hourly wage times shift length, then add a real contribution margin on top. Publish that figure to the crew and adjust after four to six weeks: if employees are visibly idle, raise the target; if tickets are running late, lower it. The goal is a sustainable pace, not maximum throughput.

Can I use this formula if I have multiple virtual brands running from one line?

Yes. Aggregate the gross profit from all brands for each shift, then divide by your per-employee target. Staffing follows total order volume, not brand count. The prerequisite is that your stations are flexible enough — and your cooks cross-trained enough — to switch between brand recipes without a reset.

What if I have part-time employees who only work four-hour shifts?

Treat a four-hour shift as half a full shift. If your target is $180 per eight-hour shift, a part-timer carries roughly $90, so two of them fill one full slot. Schedule part-time bodies so their combined targets sum to the shift's total requirement.

How often should I recalculate my staffing numbers?

Refresh the trailing averages monthly and revisit the per-employee target quarterly. Recalculate immediately — don't wait for the cycle — whenever you add or drop a brand, change menu prices materially, or see a sustained shift in order volume. Stale data is the single most common cause of a formula-driven schedule drifting back into over- or understaffing.

What if I can't afford the number of employees the formula suggests?

The formula gives a target, not a mandate. If it calls for more bodies than your cash position supports, raise the per-employee target to reduce headcount, and cross-train so one person covers multiple stations during the crest. You'll sacrifice some ticket speed, and you should watch your late-order rate closely, but you won't burn out the crew you have.

Sources

• U.S. Bureau of Labor Statistics, Occupational Outlook Handbook — Food Preparation and Serving Occupations: https://www.bls.gov/ooh/food-preparation-and-serving/
• U.S. Bureau of Labor Statistics, Occupational Employment and Wage Statistics: https://www.bls.gov/oes/
• U.S. Department of Labor, Wage and Hour Division — Fair Labor Standards Act: https://www.dol.gov/agencies/whd/flsa
• U.S. Small Business Administration — Manage your finances: https://www.sba.gov/business-guide/manage-your-business/manage-your-finances
• National Restaurant Association — Research and insights: https://restaurant.org/research-and-media/research/
• Harvard Business Review — Operations management topic hub: https://hbr.org/topic/subject/operations-management
• IRS — Restaurant and food service tax topics for small business: https://www.irs.gov/businesses/small-businesses-self-employed
• U.S. Food and Drug Administration — Retail Food Protection and the Food Code: https://www.fda.gov/food/retail-food-protection

Related on PULSE

• [How Do I Know Where, When, and How Many People to Schedule at Each of My Multi-Unit Retail Locations?](/knowledge/tl0001)
• [How Do I Figure Out How Many People to Schedule Each Day and at What Times for My Single Store?](/knowledge/tl0002)
• [How Do I Know How Many Cooks and Servers to Schedule Each Shift at My Pizza Restaurant?](/knowledge/tl0003)
• [How Many Salespeople Should I Schedule Each Day on My Furniture Store Floor?](/knowledge/tl0004)
• [How Do I Decide How Many Reps to Schedule at Each Store in My Mattress Retail Chain?](/knowledge/tl0005)