Home / Insights / The hardest part of any system isn’t the software — it’s the data

The hardest part of any system isn’t the software — it’s the data

One thing I’ve learned moving from nursery and school IT into business systems: the hardest part of any system is rarely the software. It’s the data people put into it.

At Genco, I’m working on our bespoke ERP platform. The build matters, but what really decides whether it’s useful is whether a customer, a product, or an order is entered the same way by everyone, every time. One duplicate account or a free-text field used three different ways, and the reports nobody trusts start to appear.

In nurseries it was the same lesson with higher stakes. Children’s records, allergies, emergency contacts, and safeguarding notes had to be accurate across every site, because the person reading them might be a practitioner who’d never met that child before. Wrong data was not an inconvenience. It was a risk.

Why software gets the blame

When reports look wrong, the instinct is to blame the system. Sometimes the system is clumsy. More often the inputs are inconsistent: two spellings of the same customer, products described freehand, statuses that mean different things to different teams. No dashboard can fix that after the fact without a painful clean-up.

People then stop trusting the numbers and keep shadow spreadsheets. The ERP becomes a filing cabinet you still have to check by hand. That is expensive in time and in decision quality — and it is usually a habit problem wearing a software costume.

IT can deliver screens and integrations. The organisation has to deliver habits. Otherwise you polish a platform that still serves muddled truth.

What worked in both worlds

Whether the record is a child profile or a customer order, the same patterns help:

• Fewer free-text fields, more drop-downs and validation — constrain the ways people can be “creatively inconsistent.”
• One clear owner for each type of record — someone accountable for quality, not “everyone and no one.”
• Make the right way the easiest way — busy staff will take the path of least resistance; design that path to be correct.
• Small, regular tidy-ups — rather than a heroic clean-up once a year when trust in the numbers has already gone.

In a nursery, that might mean a shared standard for how allergies are recorded so every site reads the same signal. In ERP, it might mean enforced customer codes, controlled product attributes, and a simple rule for who may create a new account. The domain differs; the discipline does not.

IT’s real job on data quality

IT should remove friction from doing the right thing: sensible defaults, required fields that matter, duplicate checks, clear naming, and training that matches real workflows. What IT should not do is own “data quality” alone while every other team types around the rules.

If you are implementing or living with ERP — or any line-of-business system — treat data standards as part of go-live and of business-as-usual. Agree the definitions before you celebrate the go-live cake. Schedule the small tidy-ups the way you schedule backups: dull, regular, non-negotiable.

The software is the shell. The habit is the product. Build both, or you will keep wondering why the reports feel wrong.

Start smaller than a “data project”

You do not need a formal data-governance programme to improve trust in the numbers. Pick one painful report — aged debt, stock, or order status — and trace three wrong rows back to how they were entered. Fix the field, the dropdown, or the ownership rule that allowed it. Then schedule the next small tidy-up. Momentum beats a whiteboard full of principles.