Quick Summary

Good AI depends on good data. Unfortunately, data quality across the globe is in a woeful state, causing errors and loss of trust within AI/ML projects, analytics, business intelligence, and executive decision-making.

Businesses are voluntarily or involuntarily shifting their investments and priorities to improving their data. The only lasting solution is to shift left, correcting data models and data quality as close to the source systems as possible (the applications that gather or generate the data).

The best way to do that is to get help, delegating modeling and data quality to the developers who know the requirements and data well when they write the source systems. This takes far less time and effort than most assume.

Turning developers into great data modelers requires:

1. Establishing reference enterprise data design standards

2. Mentoring developers in the best practices of data modeling

3. Implementing manual or automated design reviews

4. Programmatic design-time data contracts and data quality checks