Home › Cases

What happens in the operational layer of data quality?

The concrete work: analysing, finding errors, cleaning data and improving the process that produced them. The operational layer is the backbone of the Metro model. It starts with an analysis phase in which problems are identified from complaints, patterns and deviations in the data, before anything is changed.

Aris Prins has more than twenty years of experience in data quality, from analysis and quality control to cleaning data and dealing with complex customer records. This is where data quality stops being a policy and becomes work.

Why the operational layer matters

It is the backbone of the Metro model. Here analyses are run, errors traced, data cleaned and processes improved. The examples are recognisable: invoicing that goes wrong because two systems are connected incorrectly, or customers who appear several times because nobody agreed what makes a customer unique.

From analysis to action

The starting point is the analysis phase: identifying data quality problems from complaints, from patterns and from deviations in the data. Sometimes the issues are already known; sometimes they only appear after a deeper look. AI tools can help predict patterns and flag anomalies, but they do not decide what counts as a problem.

What this means in practice

Cleaning data without fixing the process that produced the error means doing it again next quarter. The operational layer only pays off when the analysis reaches the process, not just the records.

Based on an episode of the Databewuster podcast, recorded in Dutch. The episode, the summary and the full transcript are on the podcast page. This English article describes what was discussed; it does not quote the guest directly, because the conversation was in Dutch.