Nikita Boyarkin
← Volta Neobank

Market & Jobs · project 16 of 23

Age-sliced A/B HTE: 35–44 +11.0pp and 18–24 +5.3pp, but 45+ +1.4pp (ns). Referral (trust) converts 45+ best — friction is trust, not UX.

Situation
The KYC fix lifted conversion overall — I check whether it closes the 45+ segment specifically
Task
Isolate the senior segment and see whether the KYC fix closes its drop-off, not just the overall lift
Action
Age-sliced A/B HTE by age group (two-proportion z-test), KYC completion by channel, chi-square age × completion
Result
35–44 +11.0pp (p<0.001), 45+ +1.4pp (ns), 45+ in treatment 53.2% vs 61.2%
Stack
Pythonpandas / NumPySciPy / Statsmodelsscikit-learnMatplotlib / Seabornuv + ruff
Source View on GitHub →
On this page
  1. Situation
  2. Task
  3. Actions
  4. Result
  5. Recommendations
  6. Documentation

Volta — 45+ KYC Deep-Dive

Situation

The KYC fix lifted conversion overall — but does it close the 45+ segment specifically, which dropped off the funnel the most.

Task

With the overall lift confirmed, my job was to isolate the senior segment — the one that fell out of the funnel hardest — and see whether the KYC fix closes it too.

Actions

  • Age-sliced A/B HTE of KYC (two-proportion z-test).
  • 45+ vs 25–34 KYC completion by channel.
  • Chi-square age × completion within treatment.

Result

  • 35–44 +11.0pp (p<0.001) and 18–24 +5.3pp (p<0.05) — significant; 45+ +1.4pp (p=0.59, ns) — the fix does not close 45+.
  • The 45+ gap persists in treatment: 53.2% vs 61.2% for 25–34 (chi² p<0.001).
  • Referral (trust) converts 45+ best (64.1%) with the smallest gap (−2.4pp vs −10.6pp on the website).

Recommendations

  • Don’t ship a UX-only fix for 45+ — the barrier is trust.
  • Separate track: assisted onboarding (video call / in-branch KYC) + partner/referral channel.
  • Test the economics of the trust track (Project 18).

Documentation

Charts

Source: github.com/NikitaBoyarkin/volta-banking — 22 projects; figures recomputed from the repo's own datasets (data/*.csv) via its analysis scripts. Funnel counts from data/volta_funnel_data.csv (10,000 users); A/B, retention, segmentation, churn, RFM, CLV, attribution, anomalies, spend, support, NPS, JTBD, unit economics, premium, KYC deep-dive, referral, assisted CAC, FX sourcing, premium offers, anchor CAC and dormant win-back follow the published project narrative (README + part pages).

KYC-fix lift by age (HTE)

Heterogeneous effect of the KYC progress bar by age group (treatment − control, pp). The fix works for 18–44 but not for 45+.

0 5 10 15 18-24 — Lift: 5.3 5.3 25-34 — Lift: 3.1 3.1 35-44 — Lift: 11 11 45+ — Lift: 1.4 1.4 18-24 25-34 35-44 45+ Age Lift, pp
Key takeaways
  • 35–44 +11.0 pp (p<0.001) and 18–24 +5.3 pp (p<0.05) — the lift is significant; 45+ +1.4 pp (p=0.59, ns) — the fix does not close 45+.
  • The 45+ gap persists in treatment too: 45+ 53.2% vs 25–34 61.2% (chi² p<0.001) — friction is trust, not UX.

Volta project map

Volta overview →

Core

Extended

Market & Jobs

RAT v2

Causal