Looking at historical records of insurance quotes and key metrics from their web database, plus call tracking, our auto insurance client determined their crucial callback window to be 5 minutes or less for a 90% close rate. 66degrees was tasked with solving a data engineering problem: Get the data from their on-premise legacy database, enabling them to respond to calls quicker, get more out of their data, and improve revenue.

Our data practice lead and two of our data engineers handled this four-month engagement hand in hand with the customer’s data analytics and data science leads, creating a reference architecture that the client can reuse in other places for the same favorable business outcome.

IBM’s DB2 still runs the lion’s share of databases in the world, and while the market wants to move away from its expensive grip on business, there’s a lot of risk and regulatory structures that keep them from doing so. As we told our client, and as we tell other businesses in this predicament – you don’t have to move an entire platform to move your data and get more value out of it.

The challenge, however, is that it’s notoriously hard (but not impossible!) to get the data out of DB2.

To avoid any issues, we configured QLIK (the market leader in ATL pipelines for legacy systems like DB2 and SAP) to pull the data out along with its schemas such that if the schema changed, we’d know about it. We then brought data through the pipeline using Google Cloud Platform’s Dataflow and loaded it into BigQuery. To understand the results, we ran a Machine Learning model and sent it to an API and then on to the call center.

With this approach, insurance customers began receiving callbacks in 5 minutes or less, and in turn our client experienced increased referrals, sales, and conversion rates. In addition, the reference architecture that 66degrees created can now be used for other teams and departments within the business.