Insurance Pre-Fill Flow

NerdWallet's auto-insurance shoppers were sent to a third-party Quinstreet experience the company didn't control, where a confusing question order and repeated data entry cost both conversions and trust. As the sole designer, I led a NerdWallet-native pre-fill flow that asks the right questions in the right order, earns trust before requesting sensitive information, and hands a complete profile to top carriers, so people can compare 2 to 3 of them in far less time, without the repetitive data entry and frustration that pushed them away before.

NerdWallet's auto-insurance vertical earned roughly 95% of its revenue through a third-party experience powered by Quinstreet, branded as Insure.com. Because NerdWallet didn't control it, the company couldn't fix its usability problems, and Quinstreet ordered the results by who bid highest rather than by the best deal for the user. Two things were quietly costing us:

The brief I was first handed was narrow: recreate the Quinstreet flow and pre-fill known fields to cut duplicate data entry. After walking the flow myself and tearing down competitors, I argued that pre-fill alone wouldn't fix the deeper problems, and got the scope widened to a full, NerdWallet-native redesign.

I framed the work around a single question the team aligned on:

The primary user was "Thriver Eddie," a NerdWallet member archetype. He is a 37-year-old enterprise sales manager, married with two kids, financially comfortable and optimizing rather than scrambling. His most relevant frustration: "having to log into multiple platforms to check all his account balances." Eddie doesn't want to do the legwork. He wants NerdWallet to do it for him.

That pointed straight at pre-fill as the core value. Every field we could fill for Eddie was a moment we respected his time and reinforced that NerdWallet was working on his behalf.

Before designing anything, the team grounded the work in evidence rather than assumptions. Our UX researcher ran interviews and usability tests on the existing Quinstreet flow, sessions I helped synthesize. They confirmed the problems went past duplicate data entry: people dropped off when asked for PII without context, and the question order felt illogical to them. My own deep dive was the competitive teardown.

I audited roughly a dozen flows (Zebra, Progressive, Insurify, Gabi, ValuePenguin/LendingTree, Jerry, Hippo, Farmers, and more), capturing each one's questions in order, screen by screen, across desktop and mobile. I sorted them into two models, insurance marketplaces (multiple carriers, like what we were building) and single-source carriers, because the two set very different user expectations about what happens at the end.

I ran the same teardown on the incumbent Quinstreet flow, annotating it on the dimensions that drive drop-off: number of multi-select questions, raw data-entry fields, pages until a user sees a rate, and how much of the downstream carrier quote actually came pre-filled. That gave the team a shared, specific vocabulary for why the existing experience underperformed, instead of a vague sense that it "felt clunky."

Carriers dictate most of the questions, so the real design problem was deciding what to ask, in what order, and for whom.

I made the optimization criteria explicit so the flow wasn't being tuned by gut feel. We were optimizing to improve usability over Quinstreet, set the right expectations so users know what they're opting into, guide people coherently so the flow never feels arbitrary, and build trust as users share increasingly sensitive PII by keeping personal questions toward the end.

Then I segmented users by intent, since a single ordering can't serve everyone:

A flow with more than 20 questions depends on consistency. Rather than design screens one at a time, I built a small system of question templates on top of NerdWallet's design system, each with documented rules for when to use it:

This initiative touched more than ten stakeholders across Business Development, Category, Content, CRM, Performance Marketing, Biz Ops, User Ops, Legal, and Data Platform, each with a stake in what shipped.

Two moves kept it from stalling:

The redesigned flow leads with the question a car-insurance shopper expects first, their vehicle, under NerdWallet's own brand, with clear "Section 1 of 4" progress so people always know how much is left.

Underneath that simplicity, information NerdWallet already knew was pre-filled, every step's answers persisted so users could leave and come back, and sensitive questions carried short context explaining why the information was needed and how it would be used. The whole experience handed a complete profile to top carriers so Eddie could shop across several of them without re-entering anything.

The work continued after launch. A/B testing and early post-launch data showed users abandoning the flow before finishing, so I reordered the question sections to remove the friction that was costing completions.

The reorder targeted three points where the flow was losing people. Each change maps to one of them, with the reasoning behind it:

It shipped, and after iterating on live data, it beat the bar. We set out to match the incumbent Quinstreet experience on ARPV. The redesigned, NerdWallet-native flow did better: