Finley AI: Case Study
After years of focus on user optimization for the personal finance industry, we thought to ourselves "Is this industry ready for AI?" After months of experimenting, researching, designing and extensive user testing...
I bring you Finley.
My role:
Lead Creative Strategy and Planning
Active (IC) Contributor
What is Finley
Finley is a streamlined solution designed to connect users to products more efficiently than ever. It adds a layer of personalization, education, and automation to transform what would otherwise be a static and impersonal experience. Leveraging AI language models and proactive guidance, Finley delivers an engaging, tailored journey for users.
Welcome
Finley begins by asking a few situational questions to better understand the user’s intent. Its voice and tone have been carefully refined through extensive testing to ensure users feel comfortable and confident throughout the experience. The UI is streamlined, using tiles and simple tap interactions wherever possible to create an intuitive, user-friendly interface.
Educational moments
During initial user testing, some users expressed feeling “unsure” or hesitant to share certain information while still in the research phase, largely due to concerns around security and trust. Our primary focus with Finley was to strike a balance between avoiding the “AI creep” factor and fostering education and trust. We prioritized transparent communication by clearly explaining the purpose and process behind each step. As a result, users responded very positively when testing the prototypes.
Dynamic Decision-Making
With each question answered, Finley becomes smarter and more effective at delivering relevant results. Users are notified as each milestone is reached, speeding up the process while providing transparency into what’s happening behind the scenes. This approach helps build user confidence in the final recommendations.
Compare Options
After answering a series of questions using natural language, you’re presented with a comprehensive comparison experience. You can review and compare all the options each provider offers, along with Finley’s AI-driven recommendation on which option best suits your needs and why. You can also ask Finley questions about any visible (or missing) data points. The goal is to ensure users feel confident in their decision, increasing the likelihood of conversion within the current session.
The Ask
CEO wants to go all-in with AI. Asks around the teams for volunteers to take on this challenge. So yeah, of course I threw my name in the hat!
"I want you to build an entirely new AI powered APP - Blue Sky!"
​
"And you have 90 days!"
Ok, so what do we do now?
Immediately all team leads come together to discuss where the best opportunity lies for this effort.
-
Mortgage - High volume, Poor market, High rates (No monetization opportunity)
-
Credit Cards - We just built a new UX matching tool that we still need to test
-
Personal Loans - Traffic is at all time high "Bingo!"
Usability Testing
Before diving into concepts, wire framing, etc, I always like to start with usability testing on the current experience. It gives us a chance to see how users are engaging with the experience. What they are saying out loud, as they interact. Where are they struggling? Where are they dropping off? Where are they succeeding?
​
Basically how can we not only make this a powerful AI integrated product, but also HOW CAN WE MAKE THIS A BETTER PRODUCT AND UX IN GENERAL.
User Feedback
After user testing is completed, we spend time going through the videos and documenting user sentiment.
"Why does it have so many questions?"
"What are the differences between the lenders?"
"Is my information secure?"
"I don't feel comfortable giving you my personal information"
"This process is confusing"
Team Focus & Planning
Based on the usability testing, I then like to get the team together to agree on "Focus and Goals"
This was our focus and goal for this project (based on testing and business goals)
Focus
Simplify the experience
Educate and Inform the user throughout
Boost user trust
Increase user return rate for low qualifying users
Flare (Yeah, we went there)
Goal Statement
Improve the personal loan experience by simplifying the experience itself, building user trust, enhancing personalization and customization with AI, while educating and empowering our users to make informed financial decisions.
Journey Map
Before starting to work on anything, we need to make sure we are accommodating FOR ALL SCENARIOS. We again, sync up as a team (Design, Product, Engineering, Analytics) and map out together all aspects of the journey.
Personas
Once we know what the user journey entails, we can then start getting into each user mindset, and start designing for their specific journey wins and pitfalls. In this case, there were 2 personas that stood out to us immediately. We have the PRIME USER, "Bella" who has no trouble qualifying for a loan. She has a high credit score, and will be shown multiple pre-qualified offers. And then we have our SUB-PRIME USER, "Percy" who has a lower credit score, and likely will not pre-qualify for any offers (or if he does, they will be of a lower quality/chance to obtain).
Wireframes
I always like to start off with wireframes, especially for a new product. It allows both design AND product to focus on the UX and less on UI. Feedback will also tend to be more on UX/Journey, and less "Why is this button blue? Can we reduce the spacing, etc".
​
I like use Platform for these sessions, as it already has pre-existing patterns and components that the team can pull from to maintain consistency.
​
Engineering is also brought into these sessions towards the end when we are ready to present concepts. It's important to get their feedback early and often.
Visuals
Once aligned on wireframes, we can then start approaching visuals. It's important to make sure we are starting out on the right foot, so we also quickly recognized that we will need a toolkit built out (to make sure we maintain consistency) even in a tight deadline.
Toolkit
We realized that the # of mocks were rapidly adding up, and we were going to be making design changes as we go, so the UI toolkit was a necessity. This was one of the first "On the fly" toolkits that I've worked with/created, and I'm glad we did. We wound up making multiple changes to the branding throughout the process, and this saved us more times than not.
Prototypes
Whether it's low or high fidelity prototypes, I believe that if we have time, it always helps to have prototypes ready to share with stakeholders and engineering. As a designer, we are usually pretty good at explaining our vision through words, but product/engineering/etc may not always grasp what we are explaining, so this just makes things easier.
​
Another benefit of creating these prototypes is that we also have something to get out in front of users super early, for some moderated or unmoderated user testing. Leadership loves that we can be testing prior to build out. It also allows us to catch any "a ha" moments from users and rapidly iterate before handing things over to engineering.
S*&#@ Happens, so we pivot!
With all of this new AI tech running in the background, things were taking a little longer than we would have liked...
So we pivoted and worked on visual aids that acted as educational and informative cues for users. Something that felt more native, and less of a "cover up".
​
In this case, we worked on a brief site tour carrousel, as there was significant loading time coming from the lead form into the Finley dashboard / hub.
User Testing (again)
Prototypes: Complete!
​
Testing: After 2 rounds of user testing and iterating, Finley was ready to be handed off to development!
​
Important:
I always like to make sure that presenting usr testing findings that I include video snippets from the sessions. That allows the stakeholders to directly relate to the findings and less rely on my "take".
The user journey doesn't stop here though!
Off-site is sometimes even more important than on-site, so while we were handing things over to engineering to start building out Finley, we synced with the CRM team to start discussing the campaigns that we wanted to launch prior to Finley in the wild. Success campaigns, Re-engagement, Cross-promoting, etc.
Results & Impact
We made our deadline and presented Finley at our annual employee conference.
Major competitors in the space already started building out their own version.
Finley was a very successful product, running for 1 year.
Enhancements that the AI team brainstormed and implemented, eventually made their way into the “core” on-site experience
Since the implementation of the updated “smart compare” “micro-moments” “sub-prime educational content” and “voice and tone”, the on-site personal loans abandonment rate has decreased by 16% the return user rate increased +8.5% & CTL / Lead conversions increased +23%.
Seeing similar trends across all other core on-site experiences