Lumen — A calmer habit tracker

Context

Lumen Labs makes a habit-tracking iOS app with about 180,000 monthly active users. The team came to me after six months of flat growth: installs were steady, but 30-day retention had dropped to 34% and App Store reviews were trending negative — not because the app was broken, but because users described it as “stressful.”

The problem

The original streak mechanic punished any missed day by resetting the counter to zero. Internal analytics showed 62% of churned users quit within 48 hours of breaking a streak they’d held for 10+ days. The loop that was supposed to drive habit formation was instead generating guilt and uninstalls.

Process

I started with a week of 1:1 interviews — eight active users, six churned users. The pattern was unmistakable: people weren’t quitting because the app was hard; they were quitting because it made them feel like failures.

From there I ran a concept sprint with the founding team. We tested four alternative mechanics against the original in a two-week prototype study with 40 participants. The winner, which we called “weekly streaks,” counted consistency across a rolling 7-day window instead of demanding an unbroken chain.

Design decisions

Soft resets, not hard ones. Missing a day reduces your streak by one instead of zeroing it. This single change reframed the emotional payload of the app.

Calendar over counter. The home screen became a 7-day grid where each day is a dot that fills in. You can see your consistency at a glance without a single number shouting at you.

Gentler copy. Every notification got rewritten. “You broke your streak!” became “Tomorrow’s a new day — your week so far is still 5/7.”

Dark mode as default. Late-evening check-ins were the most common usage pattern. We flipped the default theme to dark and tuned the color palette for low-light environments.

Outcome

Eight weeks after rollout:

• 30-day retention climbed from 34% to 51%
• Daily check-in rate among retained users rose 18%
• App Store rating moved from 4.1 to 4.6 across 2,800 new reviews
• Organic installs grew 22% over the following quarter, driven largely by word-of-mouth

The team has since applied the same “reduce, don’t reset” principle to their premium goal-setting features.

What I’d do differently

I shipped the new mechanic as a full replacement rather than an A/B test. The metrics validated the call, but I left data on the table — I can’t quantify the counterfactual, and a small cohort of power users who liked the strict streak felt blindsided. Next time I’d run a 2-week opt-in beta first and let the original mechanic remain as an advanced toggle.