(US & Canada) Sudip Ghose, VP, Enterprise Data, Analytics, and AI at Equifax, speaks with Dennis Allen, Director of Security Programs, Strategy, and Risk at Stratascale, in a video interview about lessons learned in data, cost management, and the role of the cloud-enabling team in Equifax’s cost management journey and improved performance.

Equifax Inc. is an American multinational consumer credit reporting agency headquartered in Atlanta, Georgia.

Sharing the best lessons learned, Ghose says it is fundamental to understand the data landscape in the company first. Then, it comes down to understanding the stakeholders and their analytic demands.

Stressing analytics, Ghose states that data is done for analytics and not for storage. Delving further, he says, along with good governance, data, and technology processes, it is critical to have a product and customer mindset.

Adding, Ghose states that a product is for a customer and has a value, which must be kept transparent with the customer in case the cost increases due to AI usage. As AI costs are more than analytic costs, transparency helps the company.

At the end of the day, product benefit needs to be understood, and Ghose recommends CDOs to inculcate that thinking in the ecosystem. Everything has a cost aspect that must be justified, he notes.

Then, Ghose refers to the aspects of data quality and suggests having good observability, balance, and control to ensure good data quality. He maintains that while organizational maturity is crucial, there must be some engineering principles to have the right quality of data at the right time.

When asked how to keep an eye on costs, Ghose states that in the cloud, effective tagging is necessary for accurate cost allocation. He adds that failing to tag products with the right customers or business units means bearing the entire cost internally.

While proper tagging facilitates chargebacks, it is not sufficient, says Ghose. For instance, if a product incurs high costs due to inefficient queries, such as 25 failed attempts caused by bad query logic, it is essential to address those inefficiencies.

Poorly optimized queries can quickly consume the entire cloud budget for the month, says Ghose. To mitigate this, governance practices are in place to identify and manage such issues, which helps keep costs predictable and controlled.

Unraveling another key aspect, Ghose mentions understanding the data and compute footprint. Leveraging a usage matrix helps monitor who uses what, thereby enabling the decommissioning of obsolete resources.

Furthermore, Ghose also affirms conducting a “cost bullet train” analysis to identify cost-saving opportunities. He says that hyperscalers like Google constantly refine their cost models.

Taking the example of Google’s auto-class feature, he shares how it allows older datasets to transition into lower-cost storage tiers, but it must be explicitly invoked. Similarly, understanding delete management is crucial, as deleted datasets may still incur costs unless fully managed.

Moving forward, Ghose acknowledges the role the cloud-enabling team plays in operations. For instance, he says, the team helped them understand cost strategies with Google BigQuery, from deciding what to reserve, when to use open slots, and how to effectively utilize pooled slots.

Furthermore, the cloud team guides Ghose and his team in managing their portfolio, ensuring that they make optimal utilization of strategies. They also support adopting new technologies released by Google, including Analytics Hub features.

Concluding, Ghose states that the organization is pivoting to serverless solutions to better manage costs and improve performance.

CDO Magazine appreciates Sudip Ghose for sharing his insights with our global community.