At first, everything seems fine. The ERP opens, the screens are always the same. You fill in what’s required, leave blank what isn’t, click save, and move on. Nobody thinks about it again.
It always starts this way: data shoved in haphazardly, just enough to keep the business moving without friction.
A date field that accepts anything: “12/03/24”, “12-3-2024”, “March 12” — take your pick.
The colour description is a free-text field, so you get: “Blue”, “BLUE”, “Sky Blue”, “Romantic Midnight Blue”, “as in photo” — all supposedly identifying a specific colour shade. Sure, but which one?
Marketing tags are a free-for-all: “promo”, “Promo”, “PROMOTION”, “Promo 2024”, “Offer!!!” (three exclamation marks included). Nobody ever defined a standard, and nobody ever felt the itch to check and ask questions.
Then there’s the price that should be calculated by the system, but sometimes gets overridden on the fly so as not to annoy the sales director. And the customer type typed into a free-text field instead of selected from a dropdown because “it’s faster”.
And sure enough, it is faster!
Day after day everything runs smoothly: shipments go out, invoices get printed, customers don’t complain. And when something goes wrong, the usual scapegoat gets hauled out — whoever entered the data into the system.
Someone has to be the idiot in this setup, and it’s never the person who designed the brown, vaguely foul-smelling swamp the company is wading through and calls a database.
Then comes the day someone urgently needs to pull some numbers — maybe to make a decision, maybe to feed into an LLM pipeline — and that’s when they discover exactly what kind of mess they’re standing in:
- “Active” customers aren’t defined anywhere: everyone tagged them their own way.
- Marketing campaigns can’t be compared because the tags were written however felt right at the time.
- Dates won’t sort and filters don’t work.
- Analysing a product category across multiple seasons becomes an odyssey worthy of Shackleton.
No technical errors occurred. The system simply recorded everything faithfully — including, above all, the chaos. Until the data that seemed harmless turned out to be a complete shambles. And yet it was all there, and the business owner was convinced that “everything was in the ERP”.
Except it isn’t.
Because the only good data is data you can easily extract, process, and present. Everyone was perfectly calm, because whoever should have been keeping watch had simply stopped doing so a long time ago.
Has this ever happened to you?
Do you know any Chief Information Officers who abdicated their role as guardians of data quality in exchange for a shiny job title and a jaw-dropping salary?