Play Store Rating — A/B Experiment

Phase A (NPS Pre-filter, May 26→Jun 25): 7,597 eligible clients, 14.1% response rate (1,068 users), NPS score 4.5/5, 92.3% promoters, ~192 new reviews, rating 4.5★, +1,800% vs baseline. Phase B (Direct In-App, Jun 26→Jul 25): ~161 new reviews, rating 4.68★, 92% with 4-5★, +1,500% vs baseline. Total impact: Play Store went from 3.8★ to 4.7★, from 250 to 724 total reviews (+190%). Aviva sits +0.33★ above the median for similar financial apps (4.3★).

Aviva Crédito had no structured mechanism to request store ratings, resulting in low conversion (users had to leave the app to rate), no prior satisfaction filter, attribution gap, and risk of saturation without frequency control.

Cohort: clients with an active loan from their 5th payment onward. Baseline (Phase 0, Apr→May 25, 2026): ~10 reviews/month, cumulative rating 3.8★, 250 total reviews. Interpolation using Play Store Console 28-day windows to isolate reviews per phase. Technical limitation: Google In-App Review API does not confirm if the user completed the review (conversion inferred from store delta). Firebase-BigQuery did not process data during Phase B — no interaction metrics available for that phase.

Sequential A/B experiment in three phases: Phase 0 (baseline without intervention, Apr 21–May 20), Phase A (NPS pre-filter + In-App Review API for scores 4-5, May 21–Jun 20), Phase B (Direct In-App Review API without prior NPS, Jun 21–Jul 20). Shared trigger: customers with an active loan from their 5th payment slip generation.

Phase 0 — Baseline (Apr 21–May 20, no intervention): minimal volume of 1 to 5 ratings per week, with no stable trend. Phase A — NPS pre-filter (May 21–Jun 25): 151 total ratings over 5 weeks, avg 4.51★, 90.1% positive (4-5★). The first launch week (May 21–27) concentrated the highest volume with 83 ratings at a 4.40★ average, consistent with accumulated opinion from existing customers. Phase B — Direct In-App (from Jun 25): first week with 85 ratings, avg 4.64★, 91.8% positive. Phase B matched 56% of Phase A's total volume in a single week with an equal or higher positive rate, without NPS pre-filtering. By carrier: Telcel 72 ratings (4.68★), AT&T MX 23 (4.52★), ALTAN 22 (4.73★), Movistar 7 (4.71★). Key observation: the population that responds organically to the direct prompt is positive at similar rates to the pre-filtered group, which questions the need for the NPS as a prior filter to improve review quality.

1. NPS pre-filtering converts volume into quality: Phase A generated more reviews than Phase B (~192 vs ~161) with better perceived quality. The pre-filter didn’t reduce volume — it directed it toward positively predisposed users, raising the average rating to 4.5★.

2. Controlled friction has returns: Adding an extra step (5-level NPS) might seem counterproductive for volume, but users who passed the filter left more detailed reviews with higher ratings. The friction wasn’t abandonment — it was selection.

3. The 5th payment was the right trigger: Generating payment slip #5 is a milestone and relief moment, not a frustration point. That positive disposition explains the 92.3% promoter rate.

4. Google In-App Review API has an attribution cost: There’s no confirmation of whether the user completed the review after the prompt. In Phase B, all metrics are inferred from Play Store Console delta — proxy data, not hard data.

5. +1,800% justifies the experiment on its own: The baseline was ~10 reviews/month. Phase A delivered ~192. The mechanism works; the strategic decision is which variant to activate permanently.

1. Permanently activate the winning mechanism — implement the A/B winning strategy and ensure the trigger remains active. (Engineering · Current sprint)

2. Expand NPS to all 4 lifecycle moments — activate post-KYC, post first loan, and renewals. (Product · Q3 2026)

3. NPS Dashboard in Looker Studio — monitor score health, context chips, free text, and response rate. (Data Analytics · Q3 2026)

4. Negative review management — response process for 1–3★ reviews within 48h with CX escalation. (UX/CX · Ongoing)