Be the Air Traffic Controller of Your Data
In a recent webinar, Customer-Centric DataOps for Trusted Golden Records at Bremer Bank, Leilani Moll, Bremer’s VP of Analytics and Data Services talked about their customer-centric approach to DataOps. She explained, “When you think about banking, you imagine a lot of money flowing. It’s actually data that’s flowing, not money. But the tight governance and control you imagine a bank putting around the money, is the same you need for the flow of data.” Moll described this as having an air traffic controller mindset.
Customer-Centric Air Traffic Control
Air traffic controllers need to know about all the planes that are arriving and departing and flying nearby their airport. Think of this as your data workflows or overall data operations. They also need to know about airports, which you can think of as your data sources. However, air traffic controllers do not know much about the passengers and crew on the planes. Organizations with a customer-centric focus need to know which passengers (customer records) and coming from which airports (data sources). And more importantly, in-depth knowledge of the variations and nuances of the airports is critical. Personally, I avoid flying through ORD in the winter, DFW in the summer, and JFK always, due to their inefficiencies and unreliable on-time flights. Every airport, like every 3rd party data source is not alike and has unique challenges for optimizing your data flow. Much like air traffic controllers, IT teams need the agility to handle and communicate with each airport. To do this, Moll recommends following the classic steps. “Start with the business objective and at the end, operationalize the model.”
3 Don’ts for Agile Air Traffic Control
Moll shared her tips on staying agile throughout the data operations optimization process.
- Don’t start by boiling the ocean. Start small by addressing one data source at a time. All of the systems (just like airports) start to blend together. It can get confusing for the team members working together. As you work through more and more data sources, the data ingestion and catalog process will become more efficient.
- Don’t expect magic with data quality. Moll said “Don’t expect a magical mastering process to solve your data quality issues up front. You need to catch your data quality issues before they enter the data mastering step.” For each new source you bring into your data lake, work with your business users to clean up and improve the data quality. Reject the bad records, put them through the cleansing process with your business counterparts, then ingest them back in. As you run ingestion over and over, you’ll get fewer and fewer bad records. (Bonus tip: Bring in the business side of your organization at the very beginning of your data project.)
- Don’t choose a data management platform with a steep learning curve. If the technology is too hard to use, this takes away from the real customer data governance work that needs to be done.