What daily does that weekly can’t
The weekly prime cost worksheet is a verdict. It tells you whether food and labor were in band over the last seven days. What it can’t tell you is which of those seven days produced the verdict, or which server, or which shift, or which dish. The week is too coarse to act on directly — you need the trail of evidence behind it.
The daily recap is that trail. Net sales, voids, comps, and cash drawer over/short, captured at close every shift, in five minutes. None of those numbers individually mean anything; together, over two weeks, they make the kitchen and the floor legible. The Tuesday that always runs light is a Tuesday you can do something about. The server who voids three times more than anyone else is a 90-second pre-shift conversation. The Saturday-night drawer that drifts by $30 every other week is a process problem you can name.
The weekly worksheet says the engine is or isn’t in the band. The daily recap says where to look.
The four lines that matter most
The recap has more fields than this, but four lines do almost all the work. Get those four right and the rest is paperwork.
Net sales. What the POS reports, after sales tax. The number you compare day-over-day, week-over-week, year-over-year. The trend line is what tells you whether last week’s soft Tuesday was an anomaly or the new normal. Two weeks of light Tuesdays in a row is a signal; one is noise.
Voids. An order rung in and then removed before it printed to the kitchen. A void should be the rare correction of a true mistake — the wrong table number, the wrong modifier. Healthy voids run under 1% of sales. Above 2% is yellow band; above 3% is the kitchen losing track of orders or the floor over-correcting at the POS. Either way, somebody needs re-training.
Comps. An order that printed, was made, and then was removed from the check — the meal the manager comped because something went wrong, the dessert offered when a regular had a complaint. Comps are part of running a hospitable restaurant; the question is whether they’re intentional or reflexive. Healthy comps run 1–2% of sales. Above 3% means the manager is comping their way out of every problem instead of solving it; that’s a margin leak you can’t plug from the recipe cost card.
Cash drawer over/short. The difference between what the POS says is in the drawer and what’s actually there at close. Within $5 either way is the noise floor — counting errors, a customer who tipped in cash that wasn’t logged. Above $10 is yellow; above $20 is the band where the recap stops being a number and starts being a conversation. The conversation isn’t accusatory; the conversation is “what process broke today?”
The patterns that surface over two weeks
One day’s recap is a snapshot. Two weeks of recaps stacked side by side is the operating system of your restaurant becoming visible.
The patterns the stack reveals are almost never new information; they’re information the operator already half-knew, surfaced into something they can act on. The pattern that “Tuesdays are light” becomes “Tuesdays produced 60% of the gross of Wednesdays in three of the last four weeks, and the labor cost was 92% of Wednesday’s.” That second sentence is what gets you to a one-cook-down Tuesday or a closed Tuesday or a Tuesday tasting menu. The first sentence has been the same for years.
The same shape applies to the people lines. Three weeks of recaps with the same name on the void column three days running is not a coincidence and is not a punishment; it is a training gap that has a name. The pre-shift conversation Wednesday morning is short, specific, and free.
The four lies operators tell themselves about the daily ledger
The daily recap collects fewer numbers than the financial sheets, so the lies are smaller. They are also the most consequential because they compound day after day.
“Skipping a day doesn’t matter.” One missed day breaks the pattern detection. Three missed days in a month makes the dataset useless. The five-minute close-out has to be a non-negotiable part of the shift, not an aspirational habit. Operators who run this for six months without missing a day produce a dataset that survives staff turnover, manager changes, and POS migrations.
“The POS is good enough.” The POS is the source of the numbers, but the POS doesn’t do reconciliation. The drawer count, the cash deposit, the credit-card batch — these have to be verified at close, not assumed from the POS report. Most operators who skip the reconciliation discover, six months in, that they have been over-counting sales and under-counting drawer drift the whole time.
“Drawer-short is random.” It almost never is. A drawer that’s short by $3 every fifth day is a counting error; a drawer that’s short by $20 every other Saturday night is a process. Random doesn’t cluster. If the over/short column has a pattern, the pattern is the answer.
“The numbers are private.” The recap is internal, but the patterns are part of how you train the team. A server who sees their void rate compared (without naming) to the team average changes their behavior; a manager who sees their comp rate against last month’s adjusts their approach. Operators who keep the recap private to themselves miss the leverage of the recap as a coaching tool.
Anomalies versus patterns — when to act
Not every day’s recap needs intervention. Most days are noise; the work is filtering the noise out so the signal lands.
An anomaly is one day’s number outside the band. A drawer that’s short by $40 today, after months of hovering around zero, is an anomaly. The action is to find the explanation today — the till counted twice, the deposit short by exactly that amount, a refund issued in cash that wasn’t logged. Anomalies resolve. Pattern formation is the failure of resolution.
A pattern is a number that drifts and stays drifted. Three Saturday-nights in a row with drawer-short over $20 is a pattern. Four weeks of voids over 2% is a pattern. The action on a pattern is structural — the closing checklist, the void-vs-comp training, the deposit-handoff protocol. Anomalies are conversations; patterns are processes.
The recap surfaces both, but it doesn’t separate them for you. The two-week stack is what makes the difference visible.
The five-minute close-out protocol
The recap is meant to take five minutes. If it takes more, the protocol is wrong, not the operator.
The protocol: at close, before anyone leaves, the manager pulls the POS report, counts the drawer, reconciles cash deposit, and types the four lines into the recap. Net sales is one click. Voids and comps are two more. Drawer over/short is the difference between the POS-expected drawer and the actual count. Total: four numbers, three minutes if the POS exports cleanly, five minutes if the math is being done by hand.
Save it to your Workshop or print it. The next morning, the prior day’s recap is on the bar with coffee — you spend thirty seconds reading it, and ninety seconds deciding whether anything needs a Wednesday-morning conversation. The cadence is what produces the dataset; the dataset is what makes the weekly worksheet’s verdict actionable. Without the recap, the worksheet just tells you something happened. With the recap, it tells you what.
Open the sheet: Daily Sales Recap →
Five minutes per shift. Save to your Workshop and the patterns surface over two weeks. The math runs in your browser; your numbers never leave the page.