ChatGPT retention was lower among subscribers also paying for Claude
Six-month ChatGPT retention from December 31, 2025 to June 30, 2026 among U.S. e-receipt panelists active on ChatGPT at baseline
Share still active on ChatGPT after six months
Source: YipitData U.S. e-receipt panel. Cohorts were active on ChatGPT on December 31, 2025 and observable through June 30, 2026: ChatGPT only n=40,857; ChatGPT + Claude n=2,439. Some direct ChatGPT cancellations may be missed, which can make retention appear higher.
Chart data
| Series | Share still active on ChatGPT after six months |
|---|---|
| ChatGPT only at baseline | 80.4% |
| ChatGPT + Claude at baseline | 72.5% |
Which service did dual subscribers keep when they consolidated?
Claude was much more often the only service left. By June 30, 2026, 24.7% of the 2,439 December dual subscribers retained only Claude, while 5.0% retained only ChatGPT. Claude alone was about 4.9 times as common as ChatGPT alone. Roughly 92% still had an active paid Claude signal, either alone or alongside ChatGPT, so the lower ChatGPT retention was not simply broad subscription churn across both services.
Assumptions & Methodologies
We analyzed U.S. e-receipt panelists active on ChatGPT on December 31, 2025 and observable through June 30, 2026, separating those also active on Claude. For subscriptions with repeat receipts, active required a paid signal in the prior 45 days. For direct subscriptions, we treated a start as active for up to 400 days unless we observed a cancellation. We excluded developer API and third-party app signals. Some direct ChatGPT cancellations may be missed, making retention appear higher. The panel skews slightly male and upper-middle income. Results show association, not that Claude caused churn.