Data quality is a habit, not a project

Data quality is usually treated as a project: a clean-up effort with a start, an end, and a celebration. Six months later the data is messy again, and everyone is surprised. It should not be surprising. Quality is not a state you reach; it is a habit you keep.

Why clean-ups don't last

A one-off clean-up fixes yesterday's errors without changing the process that produced them. The moment the project ends, the same broken forms, unclear ownership and manual workarounds start generating the same problems again. You have treated the symptom and left the cause untouched.

Build routines, not rescues

• Catch errors where they enter, not months later in a report.
• Give each critical dataset an owner accountable for its quality.
• Make quality visible — a small set of checks people actually watch.
• Fix the process that created the error, not just the record.

These routines are modest. They are also the difference between data people trust and data they quietly work around with their own spreadsheets — the surest sign that trust has already been lost.

Quality as a standing commitment

The organisations with genuinely trustworthy data are rarely the ones that ran the biggest clean-up. They are the ones that made quality somebody's ongoing job, built the checks into everyday work, and kept them running when attention moved elsewhere. It is less dramatic than a rescue project, and far more effective.

So budget for the habit, not just the rescue. The clean-up gets you a clean baseline; the routine is what keeps it clean.