Restaurant working capital: before vs after with Masterestaurant

Finding 1 — How much working capital does a restaurant really need?

A full-service restaurant needs to cover 3 to 6 months of OpEx as working capital. The reason is arithmetic:

payroll eats 36.5% of sales (median, NRA 2024) and food cost another 32.0% (NRA 2024), so ~68.5% of every dollar is committed before you even touch rent. And healthy occupancy shouldn't exceed 6-10% of gross sales (Toast). Add utilities for power and water at 2-5% of revenue (Toast, 2025), and the format's operating margin is a thread.

This is the mistake I see over and over in the register: owners who budget the opening down to the cent and leave the operating cushion to improvisation. Working capital isn't the money to open the door; it's what pays Friday's payroll when the banquet ticket lands at 45 days.

Finding 2 — CapEx is not OpEx: the number-one capital leak

Confusing CapEx with OpEx is the leak that kills the most restaurants that "sold well." CapEx is the one-time investment: kitchen equipment of 50,000 to 150,000 USD for a mid-size venue (Rezku, 2025) and construction of 100 to 800 USD per sq ft (Rezku, 2025). The full opening runs from 175,500 USD in the bottom quartile to 750,500 USD in the top (Rezku, 2025). But that only opens the door.

OpEx is what repeats every month: payroll at 36.5%, food cost at 32.0% (both NRA 2024) and occupancy at 6-10% (Toast). An owner can raise the opening capital and still fail in the first low-season quarter because they provisioned zero to operate while the register cools off. The opening is a snapshot; working capital is the movie.

Finding 3 — Why 3 to 6 months and not two weeks?

The 3-to-6-month OpEx range exists because a restaurant pays in advance and collects late.

You buy inventory today —food cost at 32.0% of sales (NRA 2024)—, pay payroll every two weeks —36.5% of sales (NRA 2024)— and rent on day one, but revenue arrives staggered: cash lands fast, card at 2-3 days, and the banquet or corporate account at 30-45. In low season that gap widens exactly when sales fall.

With occupancy fixed at 6-10% (Toast) that doesn't drop even when no one walks in, whoever only saved two weeks of cash can't ride out the dip. Diego F. Parra sums it up in the Masterestaurant costing framework: working capital is measured in available days of cash, not in an absolute number that sounds good on paper.

Finding 4 — Third-party delivery looks like revenue and is a leak

Third-party delivery drains working capital because a significant share of the order goes to fees, promos and refunds. The owner watches volume climb on the screen and celebrates, but the real contribution margin evaporates in the commission. Worse: the platform holds and settles in cycles of days or weeks, so today's order cash isn't in your account when Friday's payroll hits —those 36.5% of sales (NRA 2024) must be paid regardless—.

Without separating the channel in the P&L, it's impossible to know whether delivery adds cash or consumes it. In the Masterestaurant framework we insist on separate accounting by channel: dine-in, delivery and take-away carry different margins, and lumping it all into one line is how a restaurant reaches month-end with record sales and nothing to pay suppliers with.

Finding 5 — Staff turnover as a hidden capital cost

Staff turnover is an invisible tax on working capital that almost no one provisions for. Replacing an hourly employee costs hard costs of separation, recruiting and training; replacing a general manager pushes the figure far higher. In a format where payroll already weighs 36.5% of sales (NRA 2024) and labor cost rose +6.3% in QSR during 2024 due to the minimum-wage hike (NRA 2024), each uncovered resignation is a direct hit to the register.

On top of that comes workers' comp insurance at 1.06 USD per 100 USD of payroll (Kickstand, 2025). The mistake I see over and over: budgeting payroll as a flat number and forgetting that turnover inflates it month after month without warning.

Finding 6 — How to size your cushion: the scorecard by segment

To size your working capital, start from real monthly OpEx and multiply it by 3 to 6 depending on your seasonality. Add the sector's verified weights: food cost 32.0% in full service or 32.4% in limited service (NRA 2024), payroll 36.5% (NRA 2024), occupancy 6-10% (Toast), utilities 2-5% (Toast, 2025).

Total sector labor cost falls between 25 and 35% by format (BLS; Toast), so a QSR with less service provisions differently than a fine-dining venue. A full-service spot with sharp seasonality aims for 6 months; a limited-service one with steady flow can hold with 3-4. This is a Masterestaurant Analysis 2026: an expert synthesis of real public data read by Diego F. Parra, not a proprietary sample.

Every scorecard figure carries an organization and year; the contribution is the consultant's reading and the breakdown by size.

Finding 7 — Before measures sales, after measures days of cash

A fragile restaurant's "before" measures sales; a solid one's "after" measures available days of cash. A full-service venue with occupancy at 6-10% (Toast) that doesn't provision 3-6 months of OpEx fails in the first low quarter not for selling too little, but for having nothing to wait with until flow returns. The difference between the two is working capital.

With power at 2.90 USD per sq ft and gas at 0.85 USD per sq ft a year (Toast, 2025), the fixed costs run even with an empty dining room. The metric Diego F. Parra puts on the Masterestaurant dashboard isn't the month's sales, but how many days the business can pay payroll, suppliers and rent with cash on hand.

That number, not the average ticket, is what separates the one who holds on from the one who closes while selling well.

Finding 8 — What separates a cash box that holds from one that folds

The error I see again and again: owners who calculate the opening (CapEx) to the penny and leave working capital (OpEx) to improvisation. With food cost at 32% and labor at 36.5% (NRA, 2024), ~68.5% of every sale is already committed before rent. The "before" measures sales; the "after" measures days of cash.

A full-service restaurant with occupancy at 6-10% (Toast, 2025) that does not provision 3-6 months of OpEx folds in the first low-season quarter, not from selling little but from having nothing to wait with. Third-party delivery looks like revenue and is leakage: a significant share of the order goes to fees. Without isolating that channel in the P&L, the owner celebrates volume while the real contribution margin evaporates.