# What Is ChatOn A/B Testing On Its AI-Chat Paywall?

Source: Lazyweb Research
Published: 2026-07-07
Sample size: n=14
Tags: paywall, pricing, monetization, experiments, mobile, saas
HTML: https://www.lazyweb.com/research/chaton-paywall-experiments
Markdown: https://www.lazyweb.com/research/chaton-paywall-experiments.md

**Answer.** Lazyweb Research detected 14 distinct experiments at ChatOn (July 2026), of which at least 13 touch the paywall. [1] ChatOn is a nearly pure-paywall experimenter in the AI-assistant category, concentrating almost all its detected iteration on monetization. These are observed before/after variations with inferred rationale, not company-confirmed A/B tests.

> Lazyweb Research detected 14 ChatOn experiments (July 2026), at least 13 on the paywall.

## The finding

Lazyweb Research detected **14 distinct experiments** at ChatOn, with at least 13 on the paywall. [1] ChatOn is an AI-chat assistant whose detected iteration is almost entirely on the subscription surface — a monetization-first pattern shared with other assistant apps like Genie in this corpus.

## How to apply it

ChatOn joins Genie as an AI-assistant that runs nearly all detectable iteration on the paywall (at least 13 of 14). If you build a consumer AI-chat app, these two are evidence the paywall is the dominant test surface in the category — worth a dedicated, continuous testing program. No ChatOn experiments were detected in 2026. [1]

## 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 because screen category is unlabeled on 1,425 of 4,814 corpus experiments. [cat_null]

## The numbers

| Stat | Computed from |
| --- | --- |
| 14 distinct experiments; at least 13 paywall | company_total:chaton (value 14; paywall 13, signup 1, in-2026 0) |
| 1,425 of 4,814 experiments have no screen category | screen_category_null_on_experiments (1425/4814) |

## Methodology

Universe: 14 distinct ChatOn experiments (COUNT(DISTINCT experiment_id)) within 4,814 detected before/after UI diffs across 276 companies, July 2026. Extraction: LLM-inferred rationale on observed variations. Caveat: detected variations only, never company-confirmed A/B tests.

## Sources & citations

- [1] Lazyweb Research analysis of 14 detected experiments (ChatOn, ~800-app mobile corpus), July 2026. COUNT(DISTINCT experiment_id) on before/after diffs; surface splits from is_paywall + screen_category.
- [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.

## Related questions

- [What Is Genie A/B Testing On Its Paywall CTA?](https://www.lazyweb.com/research/genie-paywall-cta-experiments)
- [What Has ChatGPT Changed On Its Paywall And Signup This Year?](https://www.lazyweb.com/research/chatgpt-signup-paywall-experiments-2026)
- [Which Apps Run The Most Paywall CTA Experiments?](https://www.lazyweb.com/research/which-apps-run-the-most-paywall-cta-experiments)