Measuring Wellbeing - Thrive Score

An unvalidated sales pitch, turned into the behavioral score that became Thrive's leading KPI.

Role: Design lead + acting PM (Q2–Q3)
Team: 5 engineers, PM partner, data science
Timeline: Fall 2024 – Summer 2025

The result: one score you can always read - every number shows what's driving it and what to do next.

TL;DR

Thrive recommended off engagement, not whether anyone's health improved. I led the work to fix that.
Built one behavioral score from five behaviors, with the “because” always visible.
Won a protective Head of Design over to a 23-question survey by testing it to a 94% completion rate.
It became the company's leading KPI.

The problem

Thrive was flying blind - recommending off engagement (did you open the notification?), not whether your health improved. Someone sleeping four hours a night got the same generic content as someone thriving.

One system, two audiences: the same behavioral data that personalizes the experience is what proves outcomes to the buyer.

Finding the right metric

The pitch was “already validated.” It wasn't. The first concept - a radar plot - drew the death sentence from the VP: “This looks nice, but I don't understand it.” Users said the same. The pattern was clear: Whoop and Oura earn trust because sub-scores show their work; a credit score doesn't, because it's a number with no because. So I reframed it - not “how do we measure users,” but “how do we make every recommendation earned.”

The system

The real unlock: the score isn't the end, it's the basis for personalization. One baseline becomes a breakdown becomes a recommendation that's actually earned - tuned to what's holding this person back, not what they last clicked.

Baseline assessment question with a Likert-scale response — 1 · Assess. A 23-question baseline - the product doing the work to understand you first.

Thrive Score with its five sub-scores, showing what's driving the number — 2 · Break it down. Sub-scores make the 'because' visible - exactly where you're strong and where you're not.

Insight banner: an opportunity in Movement, with a prompt to update your intention — 3 · Earn the recommendation. The weakest area becomes a targeted nudge - personalization the score makes possible.

Recognition

Score high in an area and you earn a superlative badge - a small extrinsic nudge you can keep in your profile or share. One per behavior, shipped in 15+ languages.

Active Achiever badge - awarded for Movement

Winning the org over

Adding 23 questions during an engagement crisis was a non-starter. I made the flow cleanly skippable, then user-tested it - and came back with the number that ended the debate.

94%

survey completion

among users who started it - the number that ended the debate and shipped the survey.

Org rollup

The same score rolls up to an HR view - one system, two audiences, made concrete.

HR Pulse dashboard: overall organizational Thrive Score (78 of 100, up 4.0% vs. last quarter), key driver flagged as Sleep, and trends over time segmented by Employees, Managers, and Executives — Same metric, organizational lens - with the lever to pull (Sleep) surfaced for action.

Impact

Co-developed with Microsoft as the beta partner - we worked closely with them to develop and validate the score before broad release.
Launched to the entire user base, localized in 15+ languages.
Became the company's leading KPI - shifting the org from engagement metrics to real user outcomes, and making progress visible for the first time.

Reflection

The lesson the false start taught: “already validated” is a claim to test, not an assumption to inherit. The veto that felt like a setback is exactly what made the product trustworthy.

There's more to this one

The scoring-model tradeoffs, the roadmap influence, and exactly what shipped. Happy to walk through it.

Email me View other work