Optum's product pages and acquisition funnels needed constant tuning to keep users engaged and moving toward their health goals, not bouncing off generic layouts.
Role Led test strategy and prioritization — turned product-analytics hypotheses into a monthly experiment pipeline
Optum's digital products needed to optimize how users moved through topic pages, search, and app-download flows so the experience matched each customer's underlying objective.
Provider search, topic pages, and landing pages all needed structured experimentation to find layouts that moved users toward their goals.
Users with different category affinities (e.g. mental health, depression topics) were shown the same generic content regardless of demonstrated interest.
Generic CRO tooling can run tests, but prioritizing and interpreting them for sensitive health topics requires in-house strategy and analysis.
At 50-70 tests a year, every stage — generating hypotheses, choosing which to build, and handling sensitive topics — needed its own process.
Sustaining 50-70 tests per year meant every step from hypothesis to insight needed a defined, monthly-reviewed process.
Product analytics teams surfaced hypotheses faster than they could be built and shipped, requiring a prioritization layer before development.
Tests on pages like "depression" topic content needed careful layout variants to find engagement gains without compromising user experience.
A six-stage loop — recommend, hypothesis, prioritize, develop, go live, analyze — ran on a monthly review cycle, with strategy and execution owned end-to-end.
Product analytics teams identified hypotheses; strategized and prioritized which to build each month based on expected impact.Hypothesis · prioritization
Collaborated across teams to build test variants — from CTA layout changes to multivariate landing page tests (MVT).A/B · MVT
Designed variants around observed category affinity, offering next-best-action content based on a user's demonstrated interests.Personalization · next-best-action
Shipped tests, analyzed results monthly, and fed insights back into the next round of hypotheses — a continuous loop.Analysis · monthly review
Optum moved from one-off page tweaks to a monthly pipeline running 50-70 experiments a year across search, topic pages, and app-download flows — with most tests landing significant, positive lifts.
Next-best-action content based on demonstrated category affinity gave users more relevant paths through sensitive topic areas.
More engaging content helped customers progress toward their health goals, contributing marginally to overall NPS improvement.
Let's talk
Tell me your biggest data or AI bottleneck and I'll send you how I'd approach it — the same way I worked this problem. No résumé dump, no pitch. Just a useful conversation.
Start the conversation →