StoriesNovember 16, 20266 min read

Why doesn't your sales report match your register? It's almost never a calculation error

Two screens in your business can show a different sale for the same day with no visible error anywhere. The real (anonymized) case of a restaurant group, and how to check if it's happening to you.

Pol

Fundador de AutoBoost

Sales and cash registerBusiness data
Why doesn't your sales report match your register? It's almost never a calculation error

Two screens in the same business, looking at the same day, can show a different sales figure with no error anywhere. If this has happened to you, the first thing worth knowing is that it's almost never a calculation mistake: it's that the two screens are reading a different source for the same fact.

That's what happened to a restaurant group with several locations. And it's not a restaurant problem: as soon as a business uses more than one program to record the same thing (a point-of-sale system, a staff spreadsheet, a management screen, a file someone corrects by hand), it shows up sooner or later, whatever the sector.

Where does a difference nobody caused come from?

In this group, the weekly sales report was built only from each location's register data: whatever the point-of-sale system marks when something is charged. That sounds like the most reliable source, and it almost always is.

The problem is in the "almost." When the team manually corrects a charge that didn't go through the register (a transfer that came in another way, a mismatch someone adjusts afterward), that correction doesn't get recorded in the register: it gets recorded in the program where the group manages staff, which is where that kind of adjustment had always been made. The sales report was never going to look there, so it never saw it.

There was a second, subtler crack, this one inside the group's own management system: the screen that compares product cost against sales read the figure from one source, and the staff screen read the sale from another. The same week, for the same location, could come out with two different numbers depending on which screen you looked at. Nobody designed it that way on purpose; each screen was built at a different time, and each one got wired to whichever source was closest at hand then.

How do you know if there's a real problem, or just rounding?

Before fixing anything, it had to be measured. Every date and location combination was checked one by one, 211 of them, comparing the two sources against each other. The result: 205 matched down to the cent. Only 6 didn't, and in those 6 the reason was always the same: one of the two sources had no data for that day, or had it at zero.

That 97% match rate is the important part of the story. This wasn't a broken system or badly migrated data: it was a small, specific gap (manual corrections for charges that don't go through the register) that, added up location by location and week by week, was enough for two reports from the same company to never quite agree. And as long as nobody compared the two reports side by side, the gap went unnoticed: each screen, looked at on its own, seemed perfectly reasonable.

Which of the two screens needed fixing?

Neither one, on its own. The fix wasn't choosing which of the two sources was "the right one": both were correct, each in its own way, and both were needed to get the full picture. The calculation was unified so that, every day, the correct figure is taken by combining the two sources (the register plus the corrections that lived in the other program), and every screen in the group was changed to read that single, already combined figure, instead of each one fetching the sale on its own.

The practical result: the weekly report, the screen comparing cost and sales, and the one comparing staff against what's sold now say exactly the same thing for the same location and the same day. If anyone in the group needs to decide something based on that figure (whether the week went well, whether a shift needs adjusting, whether a specific location needs a closer look), it no longer matters which screen they open first.

It's not just a restaurant thing

The same pattern shows up in any business that records something important in more than one place. A pharmacy group that logs what's charged at the counter in its sales program, but keeps orders that get billed at month's end somewhere else. A training academy that records enrollments in a CRM, while the actual payment of the fee sits in the payment platform. A repair shop that logs the car's arrival in one diary and the part sold in a different program. In every case, the risk is the same: if nobody built the path for corrections and adjustments to also reach wherever the total is viewed, that total comes up short, and nobody finds out until someone compares.

Could this be happening to you without you knowing?

Three questions you can ask yourself today, without touching anything:

  • Is there any charge, order or adjustment in your business that someone corrects by hand, in a different place from where it was first recorded? If the answer is yes, that's the exact point where two figures can start drifting apart.
  • If you open two different screens in your own system and both say "this week's sales" (or "orders," or "customers served"), do they show the same number for the same period? Open them both at once, on any random Tuesday, and compare.
  • If you ask your provider "do all my screens read the sale from the same place, or does each one have its own path to the data?", can they answer you with confidence, or do they need to go check?

The decision rule

If two screens in your business can show a different number for the same fact, the failure doesn't announce itself with an error on screen: it announces itself the day someone compares the two reports and the numbers don't match, and by then it may have been happening for weeks without anyone noticing. Before trusting a report to decide something (placing a bigger order, changing a shift, closing out a month), it's worth checking that it comes from a single place, not from two paths that are supposed to agree and don't always.

This is, at bottom, the same problem we solved for a pharmacy group that brought together data from several locations and its own management program into a single place: you can see that case here. And if you want to check how the data in your own business is set up before you're the one who finds out the hard way, this is what we do.

If you'd like to see it with your own company's numbers, get in touch.

Share article