How a global financial institution cleared one of finance's most rigorous AI security reviews and built its customer intelligence layer with Dovetail
One of finance's most demanding AI security reviews took three years. The outcome: a global bank's entire design and customer experience organization now builds its customer intelligence on Dovetail, with AI-powered analysis replacing days of manual research work.
3 years
to clear one of the strictest AI security reviews in global financial services
300+
survey responses analyzed in one session, replacing days of manual coding
1,000s+
of survey comments queued for AI analysis as Dovetail expands org-wide
Ch.01
Customer intelligence at one of the world's most recognized financial institutions
Few organizations have a more active relationship with customer feedback than one of the world's largest financial institutions, a company serving tens of millions of customers across consumer banking, investment banking, commercial banking, and asset management. The bank's design and customer experience organization runs one of the most active qualitative research practices in global financial services. Researchers conduct hundreds of moderated studies each year, generating hours of video interviews, thousands of survey responses, and research reports that inform product decisions across every line of business.
For years, that intelligence sat scattered. Reports lived on personal drives. Survey results were coded line by line in spreadsheets. Videos were downloaded from research platforms, moved from one tool to another, and rarely stored anywhere searchable. When a product manager needed customer evidence to back a design decision, the answer was buried in a researcher's notes or lost entirely. The bank's design organization had an enormous amount of customer intelligence. It just couldn't find it.
Getting to a better answer took three years.
Ch.02
The challenge: years of research with nowhere to find it
The scale of the problem was sharpest during survey cycles. The bank's design organization ran internal surveys that generated hundreds of free-text responses covering everything from team culture to tool effectiveness. With predefined categories and no automated way to sort comments into them, researchers spent days manually reading every response and placing it into a bucket. The survey tool the team had previously relied on couldn't handle the nuance. A word like "data" appeared throughout hundreds of responses, but the meaning of each instance required reading the surrounding context. No automation could do that reliably, so humans did all of it.
The deeper problem spread further. Research findings were generated constantly across dozens of teams, then scattered across shared drives, presentation decks, and personal files. There was no single place where a product manager could search for past findings about a specific user group or workflow. Teams duplicated work without knowing it. Hard-won customer intelligence vanished between projects.
For a bank operating at this scale, with a research practice spanning consumer products, commercial banking, and digital services, the cost of inaccessible customer intelligence wasn't just an inconvenience. It was product decisions made without evidence that already existed somewhere in the organization.
Ch.03
The solution: enterprise security cleared, customer intelligence deployed
Getting Dovetail into the bank required three years of enterprise procurement: legal reviews, security audits, contract negotiations spanning ten or more months, and an Architecture Review Board process to evaluate every AI feature before it could be enabled.
That timeline reflects what it takes to deploy a Customer Intelligence Platform inside a global financial institution. The bank's controls team needed to understand how every AI model processed data, where results were stored, and what happened to information after a query ran. Dovetail's trust infrastructure, running AI within a controlled cloud environment without sharing data with third parties or using customer data for model training, was the answer that cleared the path.
Today, the research operations team has deployed Dovetail as the system of record for customer intelligence across the design and customer experience organization. Videos from user studies are uploaded, transcribed, and stored in a single searchable workspace. Survey responses are imported from spreadsheets and processed using AI Analysis, which automatically surfaces sentiment and themes across hundreds of open-ended responses. What once took a researcher days of line-by-line coding now runs in a single session.
"Everyone is seasoned enough researchers where they know that they're still the next level. After the AI does its thing, then the researcher will go and make sure that, you know, they're reviewing everything and they're going to go through all the analyses and make sure everything makes sense."
— Delivery Manager, Research Operations
Ch.04
From research ops to org-wide intelligence
The Delivery Manager who championed Dovetail within the bank didn't bring it in to help researchers work faster. She did it to change how the entire design and customer experience organization relates to customer data. Her goal was a platform where product teams, designers, and senior leaders could self-serve answers from the research library without booking time with a researcher, where customer signals reach the people making decisions, not just the people conducting studies.
That shift is already underway. Internal champions run show-and-tell sessions to expand adoption across research teams. Team leads are establishing taxonomy standards to make the repository fully navigable. Senior leaders are among those asking to access the workspace. And the pipeline of open-ended survey responses waiting for AI-powered analysis, across the bank's broader design organization, runs into the 1,000s.
"If it goes into, hey, it's effective and it's efficient and it gives us the clarity we need, then I can definitely be a proponent of saying, hey, from the broader organization, like you should also consider using it, because that broader organization is dealing with probably… in the thousands of comments."
— UX Researcher, Design Organization
This is what customer intelligence looks like at the scale of a global financial institution: not a single use case, but an organizational shift. Research that once existed only in individual files becomes infrastructure. Analysts who once spent days hand-coding survey responses spend those days acting on what the data reveals. An organization that serves tens of millions of customers around the world makes decisions with the confidence that comes from knowing what those customers are actually saying.
