What does data quality mean in practice?

Data quality is often discussed as if it were an absolute state. It is not. The question is always whether the data is good enough for its intended use.

Four measurable dimensions

Completeness: what proportion have the field filled in. Validity: whether the information could actually be correct, such as an organization number with the correct number of digits. Uniqueness: whether the customer appears only once. Freshness: when the information was last verified.

The purpose determines the threshold

If you are sending invoices, the address must be correct for every single customer. If you want to see which industries are growing, it is enough for the industry to be recorded for most customers. Requiring invoice-level quality for every field is a surefire way to never finish.

Measure before you clean up

Count how many records are missing each type of information before you get started. Without a baseline, you won’t know whether the cleanup is working, and efforts tend to focus on what is easiest rather than what matters most.