A perimenopause funnel experiment at Flo Health

Context Flo Health is a women's health app used by 75 million people worldwide. Perimenopause is its fastest-growing user segment and the company priority.

My team is responsible for acquiring new perimenopause users and introducing them to Flo's perimenopause mode. The Peri Score was our most successful funnel, the metric gives women a way to measure the impact of their symptoms on their daily lives.

The aim of this project was to introduce the Peri Timeline, a feature the company knew users wanted, stages users in their menopause journey. Based on how they answer certain questions, an algorithm determines whether they are likely in late reproductive, early perimenopause, late perimenopause, or menopausal phases.

First iteration We chose the Peri Score funnel, our most successful funnel, as the testing ground for the timeline. It's the funnel where users who want to know if they're in perimenopause end up. I wrote the copy and co-created the design for the first version. We added a staging layer before the Peri Score questions, using the Peri Timeline algorithm to identify where each visitor was in their menopause journey.

It failed. The result was a net negative 8% outcome.

I helped analyze the data to understand specifically what had gone wrong. A key staging question about cycle changes was medically ambiguous, users couldn't answer it confidently, and it had significant drop-off. Around 76% of visitors couldn't be assigned a perimenopause phase at all, creating a dead end in the funnel. The screen confirming the perimenopause phase had a 2.51% drop — users who reached it and saw an early-stage result appeared to feel less urgency to continue. We were losing more visitors overall in the test than in control, around 21% versus 16%.

Second iteration I used the data to redesign the experience.

I rewrote the ambiguous staging question with the medical team — clearer language, same medical accuracy.

For visitors whose phase couldn't be determined, I removed the phase confirmation screen entirely and moved them straight to symptom questions.

For those whose phase could be identified, I consolidated two screens into one, using the title to bridge their timeline result directly into the Peri Score questions.

Throughout, the copy had to do more than feel right. The questions were inputs feeding a background algorithm. Working across medical, science, and product simultaneously, the language had to serve the user, satisfy the algorithm, and meet medical compliance requirements all at once.

Result The winning iteration drove a +24% uplift in conversion.

The biggest driver was the cohort we'd previously lost: visitors whose perimenopause phase couldn't be pinpointed. By removing the phase confirmation for that group rather than showing them an uncertain result, we maintained more of the audience we needed to convert. Only around 11% of visitors were identified as being in early or late perimenopause — meaning the majority of the funnel's performance depended on handling the unclassified group well.

The experiment has since been rolled out. A follow-up experiment is planned to understand the impact on activation of saving the timeline assessment results within the app.

Role Senior Content Designer | Growth | Flo Health Collaborators Senior Product Designer, Senior Product Manager, Medical Advisor, Lead Research Scientist

Perimenopause funnel experiment

A perimenopause funnel experiment at Flo Health

Context Flo Health is a women's health app used by 75 million people worldwide. Perimenopause is its fastest-growing user segment and the company priority.

My team is responsible for acquiring new perimenopause users and introducing them to Flo's perimenopause mode. The Peri Score was our most successful funnel, the metric gives women a way to measure the impact of their symptoms on their daily lives.

It failed. The result was a net negative 8% outcome.

Second iteration I used the data to redesign the experience.

I rewrote the ambiguous staging question with the medical team — clearer language, same medical accuracy.

For visitors whose phase couldn't be determined, I removed the phase confirmation screen entirely and moved them straight to symptom questions.

For those whose phase could be identified, I consolidated two screens into one, using the title to bridge their timeline result directly into the Peri Score questions.

Throughout, the copy had to do more than feel right. The questions were inputs feeding a background algorithm. Working across medical, science, and product simultaneously, the language had to serve the user, satisfy the algorithm, and meet medical compliance requirements all at once.

Result The winning iteration drove a +24% uplift in conversion.

The experiment has since been rolled out. A follow-up experiment is planned to understand the impact on activation of saving the timeline assessment results within the app.