Retention Lifecycle.
Lead designer · A designed journey from purchase to day 60, with the 30-Day Milestone shipped as the first phase.
Lead designer · A designed journey from purchase to day 60, with the 30-Day Milestone shipped as the first phase.
The first weeks run on novelty — a photo here, a song there, a question for the AI. But nothing turns into routine, so when the novelty fades there's no habit to hold on to. Weekly use slides from day 30, and the gap is widest by day 60.
Each fix was scoped to a single screen, then started from zero on the next drop-off — the experience only got noisier, and nothing compounded.
Give people a reason to keep reaching for the glasses until it becomes a habit worth keeping — and retention follows. Each stage has one goal.
One system, every surface. Each stage's goal plays out across
SocialEmailApp — heavier early, lighter later.
Shipped at day 30 — a five-page story that looks back just enough to feel personal, then turns forward: what to try next, and how other people actually use the glasses.
The first direction was a Wrapped for your glasses — photos, top songs, personal stats, all built around your own month. We built it, put it in front of users, and it didn't land.
So the ratio flipped: one page looking back, three pointing forward. We renamed it too — not a Recap but a Milestone. A recap only looks back; a milestone marks where you are on a road.
The milestone marks that turn — and it's built to feel like it: story-style pages, a cinematic opening, full-bleed visuals. A moment to scroll.
Not everything survived the redesign — knowing what to leave out is the judgment.
A holdout group saw no Milestone. The rest were split by hero — capture-heavy users got Listening promoted, listening-heavy users got Capture — always the next most retentive category.
In both treatment groups, usage of the promoted feature rose — an early, positive read on the depth-to-breadth bet.
The entry was built to show once, then step aside — yet people dug the Milestone out of their notification history to reopen it, some again and again. A pushed promo had quietly become something personal: observed behavior, not proven motive — and enough of a signal to reshape the roadmap that followed.
I ran all four phases — concept, prototype, user testing, production — with AI in the loop. AI expanded directions and pre-tested the flow; I built the front-end in Claude Code and committed to the same codebase as engineering, then fixed UI directly in QA instead of filing tickets.
AI pushed execution toward zero. Judgment became the bottleneck — choosing the direction, holding the context, owning the quality.
Once the Milestone held, the framework moved upstream — to the days before onboarding even starts. Today the app shows a blank page until the glasses arrive; the redesign turns that wait into a pre-education moment that captures a signal, then spends it at day 0 to personalize onboarding — connecting touchpoints that already exist.
Get in touch for a closer look.
The right moment, the right signal, and the right level of interruption can turn retention from a push into a pull.