What Is Genie A/B Testing On Its Paywall CTA?
Lazyweb Research detected 12 distinct experiments at Genie (July 2026), of which at least 10 touch the paywall. [1] In the CTA dataset Genie runs 10 paywall CTA experiments, 4 of which changed the CTA text — a program weighted toward offer and layout over button copy. [2] These are observed before/after variations with inferred rationale, not company-confirmed A/B tests.
Lazyweb Research detected 12 Genie experiments (July 2026); 4 of its 10 paywall-CTA tests changed the button.
The finding
Lazyweb Research detected 12 distinct experiments at Genie, with at least 10 on the paywall. [1] In the CTA dataset it runs 10 paywall CTA experiments, 4 of which changed the CTA text. [2] Genie is a paywall-concentrated AI-assistant app that iterates the offer and layout more than the button copy.
How it compares
Genie's 4-of-10 CTA-text-change rate sits below the balanced middle, closer to copy-light programs. It contrasts with LockWidget (6 of 10) at similar total volume.
How to apply it
Genie is a reference for an AI-assistant paywall that keeps most CTA copy stable (4 of 10) and tests the offer/layout instead. If your assistant app is over-indexing on button wording, Genie is evidence the offer structure may carry more weight. Test both levers separately. All are detected variations, not proven winners. [2]
Caveats
All figures are observed variations with LLM-inferred rationale, not company-confirmed A/B tests — no lift is measured. [1] Surface splits are lower bounds (screen category unlabeled on 1,425 of 4,814 experiments). [cat_null] CTA claims use the 795-experiment CTA dataset. [2]
The numbers
| Stat | Computed from |
|---|---|
| 12 distinct experiments; at least 10 paywall | company_total:genie (value 12; paywall 10) |
| 10 paywall CTA experiments, 4 changed CTA text | paywall_cta_by_company genie 10/4 |
| LockWidget 10 CTA experiments, 6 changed CTA text | paywall_cta_by_company lockwidget 10/6 |
| 1,425 of 4,814 experiments have no screen category | screen_category_null_on_experiments (1425/4814) |
Sources & citations
- [1] Lazyweb Research analysis of 12 detected experiments (Genie, ~800-app mobile corpus), July 2026. COUNT(DISTINCT experiment_id) on before/after diffs; splits from is_paywall. ↩
- [2] Lazyweb Research analysis of 795 paywall CTA experiments (146 companies), July 2026. paywall_cta_experiments; genie 10/4. ↩
- [3] Lazyweb Research analysis of 795 paywall CTA experiments (146 companies), July 2026. paywall_cta_experiments; lockwidget 10/6. ↩
- [cat_null] Lazyweb Research analysis of 4,814 detected experiments (276 companies, ~800-app mobile corpus), July 2026. screen_category is NULL on 1,425 experiments, so all surface splits are lower bounds. ↩
Source: Lazyweb Research — proprietary analysis of real, in-market app screens. Cite as Lazyweb Research, 2026-07-07.