User Differentiation in the Aviva App

The app rating improved from 3.3 to 3.7–3.9 stars on the Play Store. The onboarding redesign reduced access confusion and enabled two new prospect acquisition channels: the waitlist and kiosk appointment scheduling.

The Aviva app had a 3.3-star rating on the Play Store, driven mostly by non-customer reviews from users who couldn’t log in. Two root problems: there was no clear path for users without an Aviva account, and the app failed to communicate that a prior kiosk application was required to access the service.

The starting point was an analysis of negative Play Store reviews to identify friction patterns. Key finding: most complaints came from users without an active account who didn’t understand the business model. Validation included a 4-to-8-week A/B test distributed via Firebase Remote Config, with a minimum sample of 1,000 unique users per variant.

I designed a new onboarding experience that segments users into three journeys: active customer (existing loan), user with a prior kiosk application, and new user with no history. For new users, I created waitlist and kiosk appointment scheduling flows. The solution was validated through an A/B test comparing a manual form against automatic INE scanning.

70.36% conversion rate (full account activation) in the first 90 days post-launch. Average activation time: 1.5 minutes from download to active account. Most frequent funnel errors concentrated in OTP (code not received) and INE scanning (capture failures), quantified in GA4 with design events instrumented by the team.

1. The largest non-conversion group has no design solution: users who download the app without visiting an Aviva kiosk cannot progress through the activation flow. The problem is channel and expectation management, not UX. This requires a pre-download communication strategy.

2. Analytics event creation was an invisible blocker: without the right events in GA4, measuring the redesign’s success would have been impossible. Internalizing this process was a critical decision that accelerated the team’s ability to iterate with data.

3. OTP and INE scanning are the most measurable friction points: errors from these two actions are quantifiable and prioritizable, representing clear opportunities to improve the remaining 29.64% non-conversion rate.

1. Reduce INE scanning errors — investigate failure patterns (lighting, device, orientation) and improve real-time feedback during scanning.

2. Pre-download communication for kiosk-less users — design a ‘not yet available’ screen with a CTA toward the kiosk channel or waitlist.

3. Migrate analytics infrastructure to DBT — the Data team has started migrating from BigQuery/Looker to DBT; keep design events aligned with the new data model.