ABA therapy app: 0.7★ → 4.4★ in 18 months

The situation

A behavioral-health startup had a React Native app that therapists were supposed to use for daily ABA workflows. The team was biased toward shipping new features, but the existing feature set was actually the problem.

Users hit constant pain points. The app was unreliable. The App Store rating reflected it: 0.7 stars at the start of engagement. Crashes numbered in the hundreds per top offender per cycle. Cold-start was over 3 seconds. Licenses stalled around 500.

What we did

First, we ran a mini-audit and demonstrated to the org that the existing features (not missing features) were the user pain point. They were reluctant but agreed to pause new-feature work.

Second, we did a major refactor across several files to ensure the app was writing to the local database (WatermelonDB) correctly, then reading back and updating the Redux store in a coherent order. A side-effect win surfaced in the Redux layer: we started using the store as a change-detection signal to suppress unnecessary re-renders. That unlocked the cold-start improvement.

Third, we stood up a biweekly crash-triage process with Firebase Crashlytics. Every two weeks the team pulls the top 2–3 crashes, prioritizes them for the next cycle, and proactively notifies affected customers so feedback loops on the fixes stay tight. The cadence is still running today.

The numbers

Over 18 months, with Chris leading: App Store rating climbed from 0.7 to 4.4. Customer count grew from ~100 organizations to 500+. License count (roughly 5 per customer) went from ~500 to 2,500+. Cold-start dropped from 3 seconds to under 0.5. Crash volume on the top offenders dropped from 500–1000 per cycle to 50–100.

Numbers confirmed by Chris. Not rounded. These are outcomes of one specific engagement, not generalized claims about company-wide portfolio performance.

What this says about Rescue

The biggest lesson from this engagement: you cannot rescue a product by shipping more features onto a broken foundation. The audit's job is to surface that, and then the rescue's job is to convince everyone that stability is the feature. This is why /rescue at CAM Software is audit-first and why we'll rescope larger work into a Build or a Partner rather than stretch a Rescue past 8 weeks.