Recently, the DNB and ECB have sharpened their knives regarding BCBS239 compliance. To quote directly, “non-compliance is no longer acceptable.” The guide on Effective Risk Data Aggregation and Risk Reporting (RDARR) has also raised the bar for compliance higher than ever.
Data management and governance specialists have been using banking regulation like the goose that laid the golden egg since the 2008 financial crisis. Unfortunately, the promise of reusing tools and programs built for compliance to add value to analytics has yet to deliver.
The difficulty and expense of becoming compliant drained budgets and attention spans. Data quality programs were mostly implemented to be barely compliant (if even that). Data lineage at the column or attribute level seemed like a pipe dream (pun intended). Identifying data stewards enthusiastic about their data was a lost cause. Data owners hardly existed without layers of delegation. Governance councils went from enthusiastic to apathetic.
Competition for attention to data was fierce—the appeal of big data, SaaS, the dawn of data science, the slow move from data centers to the cloud—and of course, AI innovation demanded by everyone from shareholders to graduates.
But all is not lost…now more than ever, modern data analytics and AI demand the same tools and processes for governance to ensure value. Users want democratization of data via data products in marketplaces. Deciding who is producing and transforming data requires lineage to know which data needs attention and which we’re wasting effort on.
Additionally, data products meant to be consumed “off-the-shelf” from a marketplace require that data owners and producers don’t need to be involved with every (approved) consumer.
To derive the most value from data in the age of AI requires a full understanding of metadata—business terms, definitions, and more. Does this mean we need to revert to the early 2000s and have stewards pen definitions from scratch? Absolutely not. How many derivations of a definition or data quality rule are necessary for “Customer first name”? Shouldn’t AI agents suggest the most reasonable definitions?
The tools and processes needed for BCBS239 compliance are finally at the point of being necessary to create value from data. The tsunami of data is only getting higher. To stay competitive and on budget, organizations must harness AI for data management to free up precious human resources—while keeping control of their AI.
There are many lessons to build on from the successes and failures of governance programs since BCBS239 demanded it in 2013.
Are you ready to take your data program to the next level?
Perhaps it’s time for a coffee chat with Cameron Consulting.
hashtag#BCBS239 hashtag#DataGovernance hashtag#DataStrategy hashtag#AIandData hashtag#BankingTransformation hashtag#DNB hashtag#ECB hashtag#DataManagement hashtag#AIGovernance