Shopping for personal loans in ChatGPT
Bringing Experian’s personal loan marketplace into the conversation
Role
Focus
Team
Platforms
Lead Product Designer
Product design, logic flows, and prototyping
Product Management, Development, Content, and Design
Web, iOS, and Android
Overview
In January 2026, we kicked off a project that would bring Experian’s personal loan marketplace into ChatGPT. They had just introduced apps into ChatGPT and while the opportunities were limited, we made assumptions about where they would take apps and wanted to make sure we had adopted early on. We anticipated that ChatGPT would eventually surface relevant apps within conversations without requiring people to enable them first, which was the limitation back in January, and we wanted Experian to be ready when that opportunity arrived.
We approached the work in two stages. The first was a way to see generic loan offer ranges by lender and then continue to Experian for personalized offers. That would become the foundation for our next deliverable in May, where we would connect to the user's Experian account, collect the necessary user details and provide personalized offers in a carousel, all right there in ChatGPT.
Approach
Approach
Establish Experian as the source
Designing inside another product meant deciding where to follow its patterns and where to use our own. ChatGPT had components and a color palette we could use, and doing so would have shortened development time, but given the sensitivity of finances, making sure users knew they were signed in and interacting with Experian was something I advocated for. Keeping Experian elements, like our buttons and lists, gave the loan information legitimacy and instilled trust.
Approach
Starting with generic offers
Delivering personalized offers was not yet available within ChatGPT, so, with just their state or ZIP code, we created a flow where people could see estimated APR and loan-amount ranges from Experian's marketplace. From there, people could continue to Experian and provide the information needed for more relevant, personalized offers. This established both our presence in ChatGPT and a foundation we could build on. The next phase would bring more of that shopping journey into the conversation itself.
Approach
Bringing personalized offers into ChatGPT
For the next phase, we expanded the journey to include account connection, information collection, and personalized offers. People could essentially experience our full flow all within ChatGPT, from signing in, to providing their Top 3 details (loan amount, loan purpose, and annual income), and seeing their offers. The fuller funnel, where we collected employment status, housing status, and monthly housing payment, was also available.
I wanted the experience to feel conversational. My preference was to keep people in the chat rather than move them into a traditional form, drawing on what we knew about resistance to filling out forms. For the initial implementation, as a faster, easier MVP we couldl build upon, Product and Development chose embedded forms. I designed within that direction, keeping the information collection inside the ChatGPT experience, with conversation around it.
Stay where you are
Improving the journey
Some participants questioned why they should leave ChatGPT and sign in to Experian. They wanted more help before being asked to move elsewhere. One put it clearly:
“I went to ChatGPT because ideally I would want to compare several of them before I went to the specific website.”
This was great feedback and told us user's were comfortable shopping for loans within ChatGPT. The hesitation was about jumping from app to app, it reinforced the importance of offering useful guidance within ChatGPT and explaining what additional information or personalization the next step would provide.
Requests for state or ZIP code felt reasonable and people were also open to sharing basic financial information, but there was more hesitation around sensitive personal data. Some were uncertain about what ChatGPT or Experian already knew about them. Those expectations highlighted the need to apply the same updates and support from the Top 3 modal to this flow. These two projects were happening around the same time so we weren't able to apply the learnings from the Top 3 at the time.
Using our components within ChatGPT paid off, one participant recognized the visual connection immediately:
“This button or this kind of branding looks like the app.”
Another described Experian as trustworthy and reliable when discussing their existing perception of the brand. Together, these responses supported my rationale for retaining a recognizable Experian identity.
Branding that works
We conducted five moderated, 30-minute usability sessions on the live ChatGPT experience, using participants’ smartphones and computers. We wanted to see how people searched for loans, discovered Experian, understood the generic estimates, and reacted to the hand-off to the Experian app.
Testing
Outcome
Building off a generic set of offers, we developed one of the first authenticated, personalized loan shopping experiences in ChatGPT. From testing, we learned that Experian branding was recogizable and translated into trust and confidence, and that features like offer comparison and staying in ChatGPT were desireable for the users. These learning were captured and planned for future explorations and updates.
I don't have verified conversion or revenue results for this project. The personalized carousel of offers was far from the last iteration though. As soon as we delivered phase 2 we started on the next iteration, which Im unable to share at this time.