Running one site, you can feel how the week is going. With two or three, you cannot be everywhere, and it becomes easy to spend your time at whichever site shouts loudest instead of whichever one needs you most. A handful of numbers, checked at the same time each week, can fix that. This article is our own practical guidance, with some sources for context.
The short version
- Start with six numbers: sales, orders, average order, best sellers, busiest times and stock at risk.
- Compare each site with its own past before you compare it with other sites.
- Numbers can move for reasons nothing to do with you, such as the weather.
- A number only matters if you know what you will do when it changes.
The six numbers
| Number | The question it answers | What you might do |
|---|---|---|
| Sales by site | Where is the money coming from? | Spot a site that has changed direction |
| Orders | How many customers came? | Tell apart “fewer people” from “spent less” |
| Average order | How much did each spend? | Test bundles, or staff prompts |
| Best sellers by site | What do people here want? | Adjust ordering and range for each site |
| Busiest times | When do you need more staff? | Fit rotas to demand |
| Stock at risk | What is about to run out? | Reorder before you lose the sale |
Why sales alone can mislead
Sales are the product of orders and average order. A site with flat sales might have fewer customers spending more, which is a very different problem from the same number of customers spending the same. Splitting the two tells you whether to work on bringing people in or on what they buy once inside.
Best sellers also differ by site more than most owners expect. A station café and a riverside café can sell completely different things, even with the same menu. Ordering the same for both wastes stock at one and runs out at the other.
Compare fairly
The first comparison should be each site against itself, for the same period last week and last year. A site near an office will always beat one in a residential street on a Monday, which tells you very little.
Then allow for things outside your control. The Office for National Statistics publishes monthly retail figures for Great Britain, and they are a useful reminder of how much the wider market moves. Its September 2026 bulletin estimated that retail volumes fell by 0.4% in the month, with warm weather helping food and drink sales in the quarter, but hurting online clothing as autumn ranges sold less.1 If the whole market moved, your site did not necessarily do anything wrong.
So when a number moves, ask three questions before you act:
- Did it move everywhere, or just at one site?
- Is there an outside reason, such as weather, a local event or roadworks?
- Did something change here, such as a new manager, a price change or a broken machine?
Getting the numbers
Most tills can export sales data, and many can provide a live feed. Where neither is possible, a scheduled export is often enough to start.
Accurate sales records also matter for tax. If you are self-employed, you must keep records of your business income and expenses,2 and if you are VAT registered, some retailers use one of HMRC’s retail schemes to work out VAT with each return rather than for each sale.3 Your reporting should agree with those records, so it is worth checking that your numbers reconcile with what your accountant sees. For how long to keep them, see business records for HMRC.
A 15-minute weekly routine
Pick a time, for example Monday morning, and go through the same steps:
- Look at sales by site against last week and last year. Which site moved most?
- Split it into orders and average order. What kind of change is it?
- Scan the best sellers. Has anything risen or fallen unexpectedly?
- Check stock at risk. What needs ordering before the weekend?
- Look at the busiest times. Do next week’s rotas fit?
- Choose one action per site. One thing done is better than five noted.
Write the action down, and look at it again next week. Over time, this builds a record of what you tried and what worked.
Common mistakes
- Reacting to a single day. One bad Tuesday is usually noise, so look for patterns over weeks.
- Too many numbers. If a report takes an hour to read, it soon stops being read.
- Comparing unlike sites. Rank sites against their own past first.
- Numbers no one acts on. If nothing would change whatever it says, stop tracking it.
- Waiting for the accounts. Annual figures tell you what happened, far too late to change it.
See it on one screen
Our admin and sales dashboards put these numbers in one place you can check from your phone: sales by store, best sellers that re-rank for each store, busiest times and low stock alerts. The live demo on that page uses a sample café group, so you can tap a store and see how the picture changes. Once you know your busy times, our guide to rotas covers the rules to check before you publish one.
Sources
-
Office for National Statistics, Retail sales, Great Britain: September 2026, accessed 7 November 2026. ↩
-
GOV.UK, Business records if you’re self-employed, accessed 7 November 2026. ↩
-
GOV.UK, VAT retail schemes, accessed 7 November 2026. ↩
Admin and sales dashboards
Your sales, stock and to-do list in one place, with the numbers you actually use.
See how it works