Research → Product
American Family: the auto-quote funnel
An award-winning online auto-quote tool that wasn’t converting, and analytics that had been misreporting why.
Evidence
Measured
26.6% → 33.8%
American Family · quote completion
Measured
16.5% → 22.2%
American Family · bind conversion
Stakes
The online auto-quote tool had won an award and wasn’t converting. Leadership wanted binds up and had no reliable funnel data.
The bet
Measure the funnel honestly, then fix the pages where people actually drop.
Two moves
Rebuilt the funnel. SQL funnel modeling found that background reports could silently reject customers, so the analytics had been misleading. Why: you can’t fix a funnel you can’t see.
Tested before building. Paper prototypes, call-center ride-alongs, moderated interviews and a three-variant counterbalanced test. Why: cheap evidence first.
For the methodologist
Unmoderated A/B/C prototype test (n = 42) analyzed with Welch’s F; chi-square goodness of fit on screen-level preference. Outcomes are H1 2017 vs H1 2016.
What I’d do next
Revisit with today’s instrumentation: event-level analytics would have replaced weeks of funnel reconstruction.
Request the full case
The full case, with the working artifacts, is shared with people reviewing my work for a role or engagement. Confidential, participant-identifying or collaborator-owned material is never shared.