The fastest way to get ROI from a data platform isn't to load every data source before going live, it's to start with the one source that answers the most business questions, deliver value in parallel while you add more, and let trust in the data build gradually across the organization.
That's the core message from "Implementation Blueprint: How to Plan Your Data Project for Maximum ROI," featuring a conversation between Pat Lapomarda, Director of Data Science at Arkatechture, and Jamie Clisham, who leads data strategy at Machias Savings Bank, a $2.5 billion, 150-year-old community bank in Maine. Jamie has been building out the bank's data program for three years, and the discussion covers exactly how she prioritized sources, teams, and use cases and what it delivered.
Start with the source that answers the most business questions for most banks and credit unions, that's the core system. It holds the most concentrated data on deposits, loans, growth, and pricing, and if historical data exists, it can be loaded to give trending insight from day one. The bigger question to keep asking throughout the process is simple: where does this data live?
Prioritize by three factors: repeatability and scalability (how many people will use it and how often), alignment to strategic priorities, and actionability (a report nobody can act on isn't valuable, no matter how interesting the question). A visible, categorized backlog (tagging items as governance, strategic, or urgent) helps manage competing requests without losing sight of what actually drives value.
Because it delays the return on an investment the organization already approved. Trying to integrate everything at once means stakeholders see no value for months while the team builds in the background. When Sequencing sources go live with one, build in parallel on the next delivers value continuously and lets both the data team and the business mature together. As the webinar put it: the backlog never actually shrinks, and that's a sign of a healthy, growing program, not a stalled one.
Start with the people who approved the investment they need to see value first. Then find your champions: the people most excited to use the data, not the skeptics. Skeptics come around once they see the platform deliver real, trustworthy answers; trying to force early buy-in from resistors usually backfires. On team structure, the right setup depends on organizational readiness some institutions need pre-built dashboards, others are ready to build custom analytics with a partner acting as an extension of the internal team.
By building a clean, governed semantic layer first. AI, whether generative AI or machine learning, is only as reliable as the data underneath it. Structured, mapped, and consistently identified data (for example, a single household ID tied to a customer across systems) is what allows AI tools to produce trustworthy, testable results instead of guesses. The webinar's guidance: use AI intentionally, where clean data already supports it, not everywhere it's offered.
Hear the full story from Jamie Clisham in the webinar recording
Jamie Clisham is focused on developing and leveraging analytics and data-driven insights to drive the execution of the bank’s business strategy. She is passionate about balancing revenue growth with customer experience. She has played a versatile role at the bank to execute complex cross-functional initiatives and drive results.