Fragmented data is a sales and customer retention problem. Different pieces of data, like demographics, product ownership, spending behavior, or financial goals, are isolated from one another, preventing a holistic view. As a piece in the ABA Banking Journal notes, data may fail to line up with what another department sees, siloes don’t have immediate access to certain data sets that may help them do their jobs, and data provided by different vendors may not be compatible in the first place, forcing manual work and suspect results.

In the age of artificial intelligence (AI), efficiency is the tip of the iceberg for data management. AI models need consistent, correct, and up-to-date data to function well; fragmented data must first be sourced, formatted, and merged in ways that make it useful. Tens or hundreds of applications within different business units for different use cases will make it hard to do that. To effectively use AI models in the long run across the bank’s functions, data must be centralized and synced in real time.

To reach this end, FIs need a data strategy — establishing what data they will collect, how they will manage it, and how they will use it — driven by business strategy and supported by the right technology. That takes technical chops within the FI and a vendor-assessment process that acknowledges the role that data will play in the FI’s long-term success. It isn’t just about internal data sources controlled by the FI. External data sets are also a crucial source of business intelligence and analysis.

As we’ve written, there are a few crucial things to know about the data an FI holds before it implements a data strategy. Knowing what data you have, where it comes from, and its quality are early steps:

• Data storage: Where does customer data go once it’s produced? How do you centralize that data when it’s siloed between systems?
• Data ingestion and aggregation: What zero- and first-party data do you have, what third-party data do you have access to, and how do you extract and import it?
• Preparation: How is your data formatted when it’s extracted? What needs to be done to clean and standardize it in ways that make it usable for analytics?

Modern infrastructure is also fundamental, and better data governance is another step — sharpening the people, procedures, systems, and metrics that help an organization control and have visibility into its data.

An FI’s data is a vital competitive advantage — particularly what it knows about its customers. But data is of value only when it’s usable, and the volume of data required to train and feed AI-driven systems make that imperative even more important.