New paid-AI adopters had the highest observed Grammarly cancellation rate
Explicit cancellation over the next 13.5 months among 4,464 direct-billed Grammarly subscribers active June 30, 2025, grouped by paid-AI adoption before that date
Share with an explicit Grammarly cancellation
Source: YipitData U.S. e-receipt panel. New paid-AI adopters n=658; established paid-AI holders n=796; no paid AI by baseline n=3,010. Paid AI includes ChatGPT, Claude, or Perplexity. Explicit cancellation emails undercount total churn, so rates are floors.
Chart data
| Series | Share with an explicit Grammarly cancellation |
|---|---|
| New paid-AI adopter | 19.1% |
| Established paid-AI holder | 17.2% |
| No paid AI by baseline | 14.2% |
Did the higher cancellation rate isolate an AI-specific effect?
No. Among subscribers who had never adopted paid AI, 125 people who added a new non-AI subscription in the prior year canceled Grammarly at 17.6%, compared with 14.0% for 2,885 people who added no new paid subscription. The 17.6% comparison-group rate was statistically indistinguishable from the new-AI group, although its small sample limits precision. The result makes new paid-AI adoption a sign that the subscriber was also changing other paid subscriptions, not proof that AI replaced Grammarly.
How common did paid-AI adoption become in the Grammarly cohort?
By June 2026, 50.1% of the fixed mid-2025 Grammarly cohort had adopted a paid ChatGPT, Claude, or Perplexity subscription at some point, up from 32.6% by June 2025. This is cumulative ever-adoption, not the share necessarily holding both subscriptions at the same time.
Assumptions & Methodologies
We analyzed 4,464 U.S. e-receipt panelists with a directly billed Grammarly paid subscription active on June 30, 2025. Active required a paid receipt in the prior 12 months, no later cancellation signal before baseline, and continuous panel observation. Paid-AI adoption used observed direct or app-store receipts for ChatGPT, Claude, or Perplexity before baseline; free signups and OpenAI API invoices were excluded. New adopters first added paid AI in the prior year, while established holders adopted earlier. Grammarly cancellations were explicit cancellation or expiry emails observed after baseline through Aug. 17, 2026. Apple-billed Grammarly subscriptions were excluded because their lifecycle could not be classified consistently. Cancellation rates are floors, the panel skews slightly male and upper-middle income, and the non-AI comparison indicates the gap may partly reflect broader changes to paid subscriptions. Results do not show that AI caused or substituted for Grammarly.