Photo: Jakub Zerdzicki / Pexels (photo page on Pexels, opens in a new tab)
If you run two to ten sites, the group P&L can look fine while one venue quietly bleeds labour or GP. You don't need to read every line every week. You need to spot the outlier fast, then dig. That "ten minutes a week" only works if you already have a ready-built weekly dashboard — same columns, same definitions, every site. Without that, you're rebuilding the spreadsheet, not managing the business.
For the shape of a single-site P&L, start with How to read a P&L without panicking. This piece is about rolling several of them up without lying to yourself.
Define like-for-like once
Definition (standard practice): like-for-like (LFL) sales compare sites trading in both periods, with new, closed or refurbished-and-shut sites excluded. Label it that way on every pack. Mixing a brand-new site into "growth" flatters this year's growth, and hides whether the mature sites are actually moving.
Write the rule on the first tab of the workbook. When someone asks "why isn't Site E in LFL?", point at the definition instead of having the same argument every Monday.
Define labour % the same way on every site
Labour % here means wages plus on-costs, not wages alone. For 2026/27 figures, treat on-costs as including:
- employer NI at 15% of pay above £5,000 per employee (0% for under-21s and for apprentices under 25)
- holiday pay (illustrative accrual figure — set your own rate with finance)
- employer pension at 3% of qualifying earnings for auto-enrolled staff
State the year on the pack. You'll find sources for those employer rates on GOV.UK (employer thresholds 2026–27; workplace pensions). If one GM reports "wages only" and another includes on-costs, your group % is fiction. Use one definition everywhere, then compare.
Sales in the denominator should be net of VAT. VAT isn't yours to spend as profit, and mixing gross till take with net labour will flatter every site.
Why averages hide a bad site
Group labour % must be sales-weighted: total labour £ ÷ total sales £. It's not the average of the site percentages. Equal weighting pretends a quiet site and a busy site matter the same. They don't.
Worked example (illustrative — not a real estate):
| Site | Weekly sales | Labour % | Labour £ |
|---|---|---|---|
| A | £30,000 | 28% | £30,000 × 0.28 = £8,400 |
| B | £25,000 | 28% | £25,000 × 0.28 = £7,000 |
| C | £20,000 | 28% | £20,000 × 0.28 = £5,600 |
| D | £15,000 | 34% | £15,000 × 0.34 = £5,100 |
Step by step:
- Total sales = £30,000 + £25,000 + £20,000 + £15,000 = £90,000
- Total labour = £8,400 + £7,000 + £5,600 + £5,100 = £26,100
- Group labour % = £26,100 ÷ £90,000 = 29.0%
- Simple mean of the four %s = (28 + 28 + 28 + 34) ÷ 4 = 118 ÷ 4 = 29.5%
Site D is 6 percentage points worse than the other three (34% − 28% = 6 pts). The sales-weighted group still looks almost fine at 29.0%. The simple mean (29.5%) is a different wrong number — it still doesn't show you which site is the problem. Only the site columns do.
What you do next: visit Site D with one question — is the rota wrong, is volume too thin for the fixed hours, or is the GM giving out overtime to be kind? The pack tells you where. The visit tells you why.
Averages answer "how are we doing?" Site columns answer "who needs a visit?"
GP % spread and controllable profit
Next to labour, put GP % per site. Use sales net of VAT and a clear cost-of-sales line — the walkthrough is in GP and margin, explained with a burger. Spread matters: one site at 62% GP and three at 70% will still average "okay" while the 62% site needs a menu, waste or supplier conversation.
Controllable profit per site (GP minus site labour and other site-controllable costs — agree the line list once with finance) is the monthly in-depth check. Keep the weekly review brief so the pack actually gets read. If you try to review twenty lines every Monday, you'll stop reviewing any of them by week three.
The weekly three-number check
Every site, every week, same three numbers:
- LFL sales % vs last year (using the definition above)
- Labour % (wages + on-costs, as defined)
- GP %
Flag anything that moves the wrong way for two weeks, or any site that sits more than a few points off the pack. Then schedule the visit. Check controllable profit monthly, not in the ten-minute pass.
A practical rhythm: Monday morning, sort the dashboard by labour % descending. Look at the bottom two sites on LFL. Glance at GP outliers. That's your visit list for the week — not a tour of whoever emailed first.
What this is not
This isn't a full board pack, a tax calculation or a substitute for your accountant. It's a way to review operations: same definitions, sales-weighted totals, three numbers, one clear next action. No employer-confidential data and no company names belong in a public Learn piece; use your own estate's numbers in private. You can use the method across sites.
If you also lead through GMs rather than living in every site, pair this dashboard habit with Leading through your GMs — numbers without a cadence become a graveyard of unread files.
