The weekly P&L review is the most abused meeting in restaurant operations. Somewhere between the 90-line P&L report the general manager gets on Monday morning and the actual decisions they can make about the next seven days, the entire cadence usually falls apart. Either the review runs 75 minutes and produces zero committed actions, or it takes 12 minutes and skips the numbers that actually matter.
The fix is not more discipline. The fix is a reading pattern. There are six numbers on a weekly restaurant P&L that carry roughly 90 percent of the operating signal. The other 90 lines are either downstream from those six or move on a cadence too slow for weekly attention. Read the six, in this order, at this cadence, and the review shrinks to ten minutes plus a conversation about what to do next.
Below is the pattern I have used for years across roughly $54M in combined multi-unit scope, from a 21-unit franchise portfolio inside Walmart and Sam's Club sublocations to a $30M Michelin-recognized Bay Area group. It works. It is boring. Boring is what makes it work.
The six numbers, in order
Order matters because how the top number reads changes how you interpret the ones below it. A sales miss reframes every cost percent that follows.
1. Sales versus forecast
Weekly net sales in dollars, delta to forecast in dollars, delta to forecast in percent. Compared to trailing four week average sales and to same week prior year. Three data points, one line.
What the number tells you: whether the operating week you are about to interpret was a normal week or an outlier. A sales miss of more than 5 percent means every cost percent below reads differently. A 34 percent labor line on a 5 percent sales miss is not a labor problem, it is a demand problem the schedule could not adjust to fast enough. A 34 percent labor line on a sales hit is a labor problem.
2. Labor as a percent of sales
Total labor including taxes and benefits, divided by net sales. Compared to trailing four week average and to segment target. Trailing four week average matters more than the single week.
What the number tells you: how well the schedule followed demand last week. A 2-point deviation from the trailing four week average is a fire. A 1-point deviation is a conversation. Anything less is noise. Look at the day-part or day-of-week detail on the second screen only if the top line is red.
3. Food cost as a percent of sales
Purchases plus or minus inventory movement, divided by net sales. Trailing four week matters more than single week because a single week's food cost is dominated by inventory count timing. Four weeks of purchase and count data is where the real signal lives.
What the number tells you: whether the counting discipline is real and whether the theoretical food cost from menu mix matches what is actually happening in the walk-in. If theoretical food cost is 28 percent and actual is 33 percent, you have a 5-point counting problem, not a menu problem.
4. Prime cost total
Food plus labor as a percent of sales. This is the single most important number on the whole P&L. Everything else is either upstream of prime cost or slower moving than a weekly cadence can act on.
If prime cost is inside the segment target range, most other cost issues can be absorbed for a quarter or two without a crisis. If prime cost is 3 or more points above target, no amount of skill on operating expenses will save the unit. Prime is the number that decides whether the restaurant is operating or bleeding.
5. Comps and voids as a percent of sales
Total dollar value of comps plus voids for the week, divided by gross sales. Anything above 2.5 percent needs a look. Anything above 4 percent is a serious problem. This is the line that most often hides the largest recoverable margin in an underperforming unit, because comps and voids are the easiest place for undisciplined operations to lose money without anyone noticing.
6. Top variance line
The single biggest dollar variance on any P&L line versus its trailing four week average, with a one-line label. Delivery fees, third-party marketplace commissions, R&M spike, one-time supply order, whatever it is. This is where the general manager's eye should go for next week's action.
Fig. 1 · Six numbers, one page, ten minutes. The rest is a conversation.
The ten minute clock
Ten minutes is not a target. It is a constraint. If reading the P&L takes longer than ten minutes, either the report has too many lines on the front page or the reader is looking at lines that do not move on a weekly cadence. Both are fixable.
Approximate time budget across the six numbers:
- Sales vs forecast: 60 seconds
- Labor as percent of sales: 90 seconds
- Food cost as percent of sales: 90 seconds
- Prime cost total: 60 seconds
- Comps and voids as percent of sales: 60 seconds
- Top variance line: 4 minutes
The variance line is 40 percent of the reading time on purpose. The first five numbers tell you the shape of the week. The variance line tells you what to do about it next week.
The first five numbers are diagnosis. The sixth is intervention. Spend your reading time on intervention.
The trailing four week average versus the single week
Every restaurant P&L has volatility. Weather, local events, staffing anomalies, timing of large delivery orders. A single week's percent on any cost line can swing 2 or 3 points off trailing average for reasons that have nothing to do with operations.
The trailing four week average smooths this out. It also lags reality by about two weeks in the middle of a trend, which is fine for weekly review because your daily flash report is what catches the real-time drift. The weekly review is for pattern recognition, not for real-time intervention.
Reading discipline: put the trailing four week average next to the single week for every one of the six numbers. If both are red, act. If only the single week is red, note it and see if it repeats. If only the four week average is red, you already have a drift and you probably already know why.
What is not on the six-number list, and why
The reading pattern deliberately excludes a lot of what P&Ls contain. Not because those lines do not matter, but because they do not respond to a weekly cadence.
Rent, insurance, corporate allocations
Structural cost. Fixed inside the year. Reviewing them weekly teaches the general manager that the meeting is theater. Move them to a quarterly conversation with corporate finance.
Supplies, small wares, uniforms
These are controllable but not weekly. Order timing produces noise. Vendor consolidation and standardization decisions produce signal. Both are monthly conversations, not weekly ones. Put them in the first-Monday-of-the-month controllable cost review.
Marketing spend, catering revenue, third-party fees
These often deserve their own reviews with different attendees. Marketing is a monthly with the marketing lead. Catering is often a separate P&L. Third-party marketplace fees are pricing and channel decisions above the general manager's decision rights.
Depreciation, amortization, interest
Non-cash and non-operating. They belong on the P&L for financial reporting reasons and belong nowhere near the weekly operating review.
What the review conversation should produce
Ten minutes to read is only half of the meeting. The other 25 to 35 minutes is the conversation, and the conversation should end with one committed action for next week, owned by name, with a measurable outcome.
One action per week. Not five. Not three. One. If the general manager and the area director try to leave the review with five commitments, none of them will happen. This is a discipline I have repeatedly had to enforce with new area directors who feel like they are underperforming if they do not walk out with a long list.
Week 1 action: Rebuild Fri and Sat schedule off demand curve (GM, by Wed) Week 2 action: Portion audit on the two pasta SKUs (Chef, Thu) Week 3 action: Roll out the new comp code discipline (GM, Mon pre-shift) Week 4 action: Cross-train two servers to reduce Sun close labor (GM, ongoing) 52 committed actions per year × 3 units = 156 corrections Cadence beats intensity. Every time.
The mistakes I made building this cadence
Three specific errors, so you do not repeat them.
I let the report have 90 lines on the front page
The first version I tried to install used the standard accounting-generated P&L. Every meeting drifted into questions about line 47 or line 68, and the six numbers never got the attention they needed. The fix was a one-page operating summary generated separately, with the accounting P&L available in an appendix. Almost nobody asked for the appendix.
I made the review 60 minutes
Fell into Parkinson's law. Sixty minutes was the calendar block, so the review always found a way to fill 60 minutes. Shortened to 35 minutes. Same outcomes. Half the meeting fatigue. General managers started staying after voluntarily to discuss specific issues, which was a better use of everyone's time.
I did not commit to one action per week
For the first quarter I let the review produce whatever action list emerged. The general managers dutifully wrote them down. Almost nothing got done. When I moved to one committed action per week with a name and a due date, the actions started shipping. Same team. Same units. Different constraint.
What a general manager should know from memory
The reading pattern is a technique, but the underlying discipline is that the general manager has to be able to talk about the number without the report in front of them. A general manager who can name their prime cost, their trailing four week labor, and their top variance line without looking is a general manager who is actually managing the P&L. A general manager who has to open the report to answer the question is a general manager who has been treating the P&L as reporting rather than as an operating tool.
This is a coaching bar I hold on every general manager I work with. Not because I want to catch them out, but because the memory is the proof that the number is real to them. When the number is real, the schedule reflects it, the prep list reflects it, the pre-shift reflects it. When the number is not real, none of those artifacts move even if the P&L is being reviewed weekly.
The test is simple. Ask three questions in a hallway conversation. What was your prime cost last week. What is your trailing four week labor. What is the one thing you are doing this week to move it. If all three come back cleanly, the review cadence is working. If any of them require looking something up, the cadence is theater and needs to be reset.
The point
The weekly P&L review is not a comprehension test. It is an operating tool. Six numbers, ten minutes to read, 25 minutes to talk, one action to ship. Everything else is either downstream from those six numbers or belongs in a different cadence.
When the review works, the general manager can name their prime cost from memory, their trailing four week average without looking, and the one thing they are doing this week to move it. When the review does not work, none of those three things are true.
The pattern is the diagnostic. Read the six numbers. Watch what the general manager says without prompting. If they lead with the top variance line, the cadence is working. If they lead with an explanation about rent, the cadence has already broken and no amount of P&L detail will fix it.