How to Read a Weekly GP Report as a Head Chef

A weekly GP report comes down to four lines: sales ex-VAT, purchases, stock movement and the GP% they produce. Read the lines before the percentage, treat one bad week as noise until the count is checked, and treat three weeks moving the same way as a trend that needs a change to spec, portion, price or waste.

The four lines that matter

A GP report may carry a dozen rows. Four of them decide the number.

• Sales ex-VAT. The till total with VAT taken out. For standard-rated food at the 20% standard rate, that is gross takings ÷ 1.2. If the report is built on gross takings, every figure below it is wrong before you start.
• Purchases. Every food invoice dated in the week, net of credits for shorts, returns and wrong items, and taken at the net (ex-VAT) invoice figure.
• Stock movement. Opening stock minus closing stock. If closing stock is higher than opening, some of the week's purchases are still on the shelf.
• GP%. What is left of each pound of ex-VAT sales after the food it used.

The arithmetic linking them is short:

• Cost of sales = opening stock + purchases − closing stock (adjusted for credits and any transfers to or from the bar or another site)
• GP% = (sales ex-VAT − cost of sales) ÷ sales ex-VAT × 100

If you want the full method for counting stock, it is in how to do a food stock take and calculate your actual food cost.

A worked week, line by line

An illustrative week for a small kitchen:

• Gross takings: £14,400, so sales ex-VAT = £14,400 ÷ 1.2 = £12,000
• Opening stock: £3,200
• Purchases on invoice: £4,100, with a £120 credit owed for a short on the meat order
• Closing stock: £3,500

Before the credit: cost of sales = £3,200 + £4,100 − £3,500 = £3,800. GP = £12,000 − £3,800 = £8,200, which is £8,200 ÷ £12,000 = 68.3%.

With the credit applied, purchases are £3,980 and cost of sales is £3,680. GP = £8,320, which is 69.3%.

One missing credit note is a full point of GP. That is why you read the lines before the percentage: a percentage cannot tell you which line moved, and the fix for a bad purchases line is a phone call, not a menu change.

One bad week is noise, three is a trend

Here is an illustrative run of seven weeks from the same kitchen, sales ex-VAT and cost of sales as reported.

• Week 1: sales £12,000, cost £3,800, GP £8,200 = 68.3%
• Week 2: sales £11,500, cost £3,650, GP £7,850 = 68.3%
• Week 3: sales £12,200, cost £4,470, GP £7,730 = 63.4%
• Week 4: sales £11,800, cost £3,150, GP £8,650 = 73.3%
• Week 5: sales £12,100, cost £3,990, GP £8,110 = 67.0%
• Week 6: sales £11,900, cost £4,000, GP £7,900 = 66.4%
• Week 7: sales £12,300, cost £4,220, GP £8,080 = 65.7%

Weeks 3 and 4 look like a crisis followed by a miracle. Neither happened. A £600 delivery was invoiced in week 3 but arrived after the Sunday count, so it sat in week 3's purchases without being in week 3's closing stock. Week 3's cost of sales was overstated by £600 and week 4's understated by the same amount, because the food was used in week 4 with no purchase against it.

Put the two weeks back together and the swing disappears: £24,000 of sales, £7,620 of cost, £16,380 of GP, which is 68.25%, bang in line with weeks 1 and 2. The fix is procedural, not culinary: count at the same point every week, after that morning's delivery is put away, and date purchases by when the goods arrived.

Weeks 5, 6 and 7 are the real story. No single week is dramatic, but GP has stepped down three times in a row, from 68.3% to 65.7%. On £12,300 of sales, 2.6 points is about £320 a week walking out of the kitchen. That is a trend, and it will not correct itself.

The rule of thumb: a single out-of-line week gets its count and invoices checked before anyone changes a dish. Three weeks drifting the same way, with the count checked, need an operational answer.

The questions to ask on Monday

Before looking at the menu, work down the report.

• Were sales divided by 1.2? Check the sales figure against the till Z-reads. It sounds obvious, but a report built on gross takings flatters GP by several points, and fixing it looks like a collapse.
• Is every invoice in, and only this week's? Look for deliveries that arrived either side of the count, and for credits still owed.
• Was the count done at the same point in the week, by the same method? A count taken after a delivery one week and before it the next will swing the number on its own.
• Did anything move between departments? Lemons, limes, cream and fruit going to the bar, or staff meals, should be transferred out, not left in food cost.
• What changed in the week? A new supplier price, a new dish, a function, a change of chef on the section, a fridge failure.
• Which dishes sold more? If the sales mix tilted towards lower-margin plates, GP can fall with every dish costed correctly. The menu engineering guide covers how to find them.

If the answers are clean and the gap is still there, compare what the kitchen should have used with what it did use. That method is in ideal vs actual food cost variance.

What to change when the trend is real

Four levers, pulled roughly in this order, because the first three cost the customer nothing.

• Spec. Has a supplier substituted a product, changed a pack size or raised a price without anyone re-costing the dish? One ingredient on a best-seller moving a few pence a portion shows up across the whole week.
• Portion. Weigh ten plates of your three best-sellers during service. Portions creep up a few grams at a time, usually on the proteins and the sides that go on by hand. Put the gram weights on the pass, not in a folder.
• Waste. Look at the bin and the prep list together. Over-prepping for a busy day that did not come, trim nobody is using, and short-dated stock binned all land in cost of sales.
• Price. If spec, portion and waste are all where they should be and the cost of the ingredients has simply gone up, the menu price is the fix. Re-cost the affected dishes and move the prices that need moving, rather than a flat increase across the board.

Change one thing at a time where you can, then watch the next three weeks.

Mistakes that make the report lie

• Reading GP% without reading the lines. The percentage is the conclusion. The cause is in sales, purchases or stock.
• Reacting to one week. Rewriting a menu because of a count-timing error costs more than the error did.
• Letting two bad weeks average out. Timing errors should be corrected in the weeks they belong to. If you only look at rolling averages, a flattering week can hide the start of a real slide.
• Using purchases as cost of sales. What you bought is not what you used. A week of stocking up for a bank holiday will look terrible on purchases alone.
• Comparing against someone else's target. Judge GP against your own costed menu first. External figures are context, not a pass mark.

Where to start

If you want the four lines kept in one place each week, the Weekly GP Tracker works out food cost %, GP % and GP £ for every week and shows the trend over 26 weeks, which is exactly the three-week drift you are looking for. For context on where UK kitchens tend to land, and why a number quoted online may not apply to you, read the UK food cost and GP benchmarks guide.