Developing a user-centered lead flow
We designed a new, multi-step flow that boosted revenue and form rates.
Background
NerdWallet's auto insurance experience revolved around an integration with QuinStreet.
Users who came to NerdWallet looking for insurance options were redirected to QuinStreet’s lead flow (branded as Insure.com). There, users would answer a series of questions that resulted in a list of insurers.
Problem
A 3rd-party partnership made sense when the insurance vertical first launched — NerdWallet didn’t have the resources to develop and manage features.
But as the vertical evolved, the partnership’s constraints began to outweigh its benefits:
- We couldn’t track the end-to-end performance data (and neither could QuinStreet)
- The task of getting what little data they had was painstaking, and a poor use of resources
- QuinStreet’s flow was inconsistent with our design guidelines, patterns, and principles
Solution
We decided to build an end-to-end insurance flow on NerdWallet’s domain.
By replacing QuinStreet’s flow, we could:
- Track data internally, allowing us to test and improve the experience
- Scale the flow to other verticals
- Ensure a user experience that's consistent with NerdWallet’s standards
Challenges
- Support our users' needs while meeting:
- 3rd-party API requirements
- Legal language requirements
- Reuse existing components and content patterns in a new type of flow
- Include a new “product” type on the results page (an insurance broker)
Goals
- “Do no harm” to conversion metrics
- Improve the flow’s usability (compared to QuinStreet’s experience)
- Help users set accurate expectations for what will happen and what they’ll get
Approach
Discover
- Helped coordinate a moderated usability study with my UX research partner
- Tracked our 3rd-party partners' field mapping requirements to determine what data types were necessary for returning results to users
Frame
- Conducted competitor analyses and audited QuinStreet’s experience with my product design partner
- Mined existing data and UX research for similar flows to understand:
- Where and why users dropped off
- How users respond to requests for personally identifiable information (PII)
Ideate
- Explored flow concepts around user intent and PII with my product design partner
- Co-led a dot-voting workshop to solicit stakeholder feedback and align on a final concept
- Developed a content strategy based on the frustrations identified in our UXR study
- Created concepts for the new insurance broker product with my design partner
Execute
- Designed content for the end-to-end experience, including new patterns and dynamic interactions
- Translated field mapping requirements into user-friendly content
- Partnered with legal to understand their requirements and translated those into user-friendly content
Before
After
One of the project constraints was the requirement for reusable components and content patterns. Unfortunately, the wording on form components alternated between first and second-person pronouns. UXR studies found that this confused users in other flows, so I advocated that we only use second-person pronouns in ours. Alas, the change was deemed out of scope.
Impact
+20%
Increase in average revenue per transaction (ARPT)
+44.4%
Increase in form open rates
+40%
Lift in form complete rates
Originally released as an A/B test against QuinStreet’s flow, NerdWallet's variant significantly outperformed the control:
- Form open rates increased by 44.4% and form completion rates increased by 44%
- Our goal for ARPT was “do no harm,” but it increased by 20%
© 2026 April De Costa | Content designer and strategist