, Retail
Your till has recorded every transaction for four years. Time, operator, items, prices, discounts, payment method. You have looked at the weekly total and, on a bad month, the weekly total for last year.
Nothing about that is unusual. Electronic point of sale systems are sold on features and bought for taking money, and the reporting sits behind a menu nobody opens because the first three reports anyone tries are useless. The useful ones take a few minutes to find and then change what you do on Tuesdays.
Three reports, and no more to start
Begin with these, monthly, printed if that is what gets them read.
Sales and margin by category, this year against last. Not sales alone. A category can grow while its margin shrinks, usually because the growth came from promotion, and the sales line will congratulate you the whole way down.
Sales by hour of day, by day of week. Almost every shop has a shape it has never examined: a dead hour it staffs fully, a Saturday peak that runs an hour later than the rota assumes, a Thursday evening that stopped being worth opening for two years ago.
Top and bottom 50 lines by margin contribution, not by units. Units tell you what is popular. Margin contribution tells you what is paying for the shop.
That is enough for a year. The temptation with any new system is to build a dashboard with forty measures on it, which is a reliable way of looking at none of them.
Hour by hour against the rota
Put takings by hour next to labour hours by hour and you have sales per labour hour, which is the single most actionable number a small retailer has.
It will be uneven, and the pattern is usually consistent: overstaffed in the morning because that is when the delivery arrives and everyone is in, understaffed in the late afternoon, and a lunchtime peak covered by whoever is not on their break. Moving two hours of labour from one part of the day to another costs nothing and shows up immediately in queue length and in average transaction value, because queues shorten baskets.
Use it to decide opening hours too. The question is not whether the last hour of the day is busy. It is whether the margin earned in that hour covers the wages, the lighting and the heat. Shops stay open to nine for reasons of habit and pride far more often than for reasons that appear in the figures.
The line that sells other lines
Basket analysis sounds like something a supermarket does with a data science team. In its useful form it is a question you can answer with an export and a spreadsheet: which products appear in the same transaction, and what does a basket containing X look like compared to one without it.
What that turns up is normally three kinds of line.
- Traffic drivers. Low margin, high frequency, and the basket around them is large. These are the lines to keep available and to price sharply, because they are buying you visits.
- Attachments. Things bought alongside something else, which respond enormously to position and to a member of staff mentioning them.
- Standalones. Sold on their own, to people who came for that and nothing else, and who therefore cost you a visit each.
The mistake basket analysis prevents is delisting a low-margin line because its own row looks bad, when it is the reason for a quarter of your transactions. Look at what leaves the shop with it before you cut it.
Average basket value and items per basket are worth tracking over time as well, because they respond to things you control: layout, queue length, staff prompts, the position of the impulse fixture.
The reports that mislead you
Some standard till reports are actively unhelpful.
Sales by operator, used as a ranking, mostly measures who works the busy shifts. It is genuinely useful for exceptions rather than for league tables: refunds without an original sale, voids, no-sales, manual price overrides. Those belong in a weekly report that somebody reads.
Year-on-year percentage on a small base swings wildly and means very little. Look at absolute pounds alongside it.
Stock on hand from the till is a book figure, not a count. It is only as good as your goods-in discipline, and treating it as truth is how a line shows as available for three weeks while the shelf is empty.
What HMRC expects the till to do
Two points worth knowing, because both have teeth.
If you sell a mix of standard, reduced and zero-rated goods, the VAT retail schemes set out in HMRC's Notice 727 govern how you work out output tax, whether through point of sale, apportionment or direct calculation. Scanning at line level and getting the VAT flags right on your product file means the point of sale scheme does the work for you. Getting those flags wrong quietly misstates your VAT return every quarter.
Second, electronic sales suppression, meaning software or configuration that hides or reduces recorded sales, has been a specific target since Finance Act 2022 gave HMRC powers to require information, issue penalties of up to £50,000 for making, supplying or promoting such tools, and to penalise possession. If a supplier ever offers you a till feature that helps with your takings, that is what they are describing.
Making Tax Digital for VAT also requires digital records with digital links between them, so a process that involves retyping the day's totals into your bookkeeping is a problem waiting for an inspection. Most till systems export to the common accounting packages, and turning that on is an afternoon.
Where this shows up in the game
Shops are part of Business Baron's full game, and e-commerce is free to run. The game models the retail trades and the till, and it gives you real financial statements rather than summary bars. Reading them is optional. It is also most of the game.