You can watch this second episode about applying the operational layer from the Metro model on YouTube, or listen to it on Spotify and Apple Podcasts.
Understanding Data Quality from Within
Aris Prins, with over 20 years of experience in data quality, shares practical insights: from data analysis and quality control to data cleansing and managing complex customer data. This article is a must-read for anyone who wants to understand and improve data quality in practice.
Why the operational layer?
The operational layer forms the backbone of the Metro Model. This is where data quality becomes tangible: analyses are performed, errors are detected, data is cleaned, and processes are improved. Aris illustrates this using recognizable situations, such as faulty invoicing due to incorrect system integrations or customers who are incorrectly registered multiple times.
From analysis to action
An important starting point in this layer is the analysis phase. Aris explains how to identify data quality issues based on complaints, patterns, and deviations in
Datasets. Sometimes these issues are known, sometimes you only discover them after a deeper analysis. AI tools can help predict patterns and identify anomalies, for example, when a customer's birthday appears to be on January 1, 1900.
Rules and stakeholders
A recurring theme is the need for clear rules and collaboration between stakeholders such as data stewards, data owners, and IT. Who determines what is correct? How do we ensure this remains the case? Clear definitions and responsibilities allow for effective monitoring and improvement of data quality.
Cleanup, migration and monitoring
Data cleanup often proves more complex than anticipated. Sometimes, it's technically impossible to adapt systems, for example, due to vendor dependency. Workarounds are then sought or temporary manual solutions implemented. Migrations also carry risks, such as incorrect birth dates or duplicate customer registrations. Monitoring helps with this: how is the quality developing, and how many errors are being resolved or are new ones being introduced?
AI, automation and value
High-quality data is essential for successful AI applications and automation. Aris emphasizes that poor data can lead to incorrect predictions and inefficient processes. By prioritizing potential damages or fines, organizations can better target their investments in data quality improvement.
A practical start for enthusiasts
For those who want to get started with data quality, Aris Prins advises: start small, choose a specific pain point, analyze the data, and monitor progress. Use smart tools to recognize patterns, but above all, don't forget the importance of human insight and collaboration.
Stay informed!
Curious about more practical insights and tools related to data quality? Subscribe to our newsletter via The Sonny Side of Life. We'll share the materials discussed there and keep you updated on future episodes!
Want to implement the Metro Model within your organization? contact on.
