Data quality
Why do our numbers disagree, and how do we secure them?
Data quality is about cleaning, validating and securing, so two departments stop seeing two numbers for the same question. The weight is not on the clean-up but on securing it: ownership per definition and checks that keep running. Without that it goes wrong again within six months.
When do you need this?
Two departments, two numbers, the same question. Data quality is about cleaning, validating and securing, so the conversation returns to the outcome instead of the figures.
- Cleaning, validating, securing
- Assigning ownership per definition
- Checks that keep running
What happens if this is left alone
- One customer appears three times with three addresses, and the mailing goes out three times.
- Nobody trusts the numbers any more, so a private list is kept alongside the system, and then there are two truths.
- An error at the source runs through to the dashboard, the report and the annual figures, and is only found when an auditor trips over it.
Through which route do we deliver this?
The same field runs through different routes. Which one fits depends on the scope and duration of what is on the table.
Frequently asked questions
Why do our numbers disagree?
Almost always because the same term is defined differently in two places, not because something is broken. As long as nobody owns that definition, the difference keeps returning.
Is data quality a project or ongoing work?
The clean-up is a project, securing it is ongoing. Checks that keep running automatically are the difference between a tidy dataset and a dataset that stays tidy.
What do you record?
Per definition: who owns it, what the source is, and which check sits on it. That document is the real result; the cleaned table is the by-product.
Who carries this out?
Often a data student through New Data Student for the cleaning and checking, with a consultant through Squad for the definitions and ownership.
Are you a professional working in this field? Then have a look at the roles we fill and our vacancies.
The image on this page was generated with AI. What you see is an illustration and not a photograph of our reality.
Looking for someone for data quality?
Tell us what is on the table. Within two working days you know whether we have someone and what it costs.