50% of firms are using AI or ML to spot data quality issues

About half of firms are using AI or machine learning-based techniques to spot data quality issues.

Of 10 major banks and asset managers surveyed for our ongoing 2025 Automation in Data Management Benchmark, 50% have used AI or machine learning to identify data quality issues in the past year.

An additional 30% of firms are working to do the same. And the firms who successfully spotted data quality issues using AI or ML remediated less than 25% of issues using the same techniques, meaning at least three-quarters of issues were remediated manually.

There’s a long way to go before banks and asset managers fully automate their data functions.

If you’re interested in how your firm compares, our benchmark is still open. Reach out to me at emmahilary.gould@infopro-digital.com to receive a link.

Only firms who complete the survey will receive the full results.

The statistic is small but significant. Our research shows that data management offices vary widely in their means of automating, but almost all agree they are under high pressure from executives and their boards to automate more.

Still, “tremendous” progress has been made automating core data governance, says Junaid Farooq, the founder of Pegasus 19 Consulting, which is currently working with a large bank’s data and AI office.

Data quality, one of the “pillars” of core data governance, is used by industry groups like the EDM Association to measure and standardize data management. Quality assesses the accuracy, completeness, consistency, timeliness, validity, and uniqueness of datasets.

The bank Farooq is working with has programmed data quality agents to perform profiling (an exercise that evaluates the accuracy of data), write data quality rules, and detect anomalies in datasets, such as missing fields, fat-finger errors, or abbreviations where there should be full words or phrases.

“AI is very good at searching a large set of data, analyzing it, and summarizing it. When you can do that, identifying anomalies in your data from a data governance perspective is reliable, low-hanging fruit,” Farooq says.

Benchmarking is a new initiative on WatersTechnology to bring our readership trusted, independent information.

We will run more benchmarks next year covering how technology and data are used in financial markets.

If you are not a data professional but have ideas for future benchmarks, reach out at emmahilary.gould@infopro-digital.com.