Khanmigo subscribers paid ChatGPT or Claude more often than other paid education-technology subscribers and active panelists overall
Share also paying OpenAI or Anthropic from August 2024–August 2026, among active U.S. panelists
Share also paying ChatGPT or Claude
Source: YipitData U.S. e-receipt panel. Khanmigo subscribers n=127; other paid ed-tech subscribers n=46,673; all active panelists n=1.31 million. Khanmigo results are directional because the cohort is small.
Chart data
| Series | Share also paying ChatGPT or Claude |
|---|---|
| Khanmigo subscribers | 53.5% |
| Other paid ed-tech subscribers | 26% |
| All active panelists | 6.3% |
Which paid AI subscription usually appeared first?
ChatGPT or Claude appeared before the first observed Khanmigo receipt for 69.6% of the 79 subscribers with both signals. An AI subscription could begin before Khanmigo and still remain observable afterward: 79.7% had a paid AI receipt on or after Khanmigo adoption. In 63.3%, the last observed paid AI receipt came after the last Khanmigo receipt. These receipt sequences show coexistence, not product usage or true subscription start dates.
Assumptions & Methodologies
We analyzed YipitData’s U.S. e-receipt panel from August 2024–August 2026. Khanmigo subscribers were 127 active panelists with Stripe-billed receipts from Khan Academy; an unrelated Khan & Khan Tutorial biller was excluded. Paid AI meant direct OpenAI or Anthropic receipts. The benchmark included 46,673 subscribers to Duolingo, Coursera, Chegg, Quizlet, or Grammarly under the same panel constraints. ChatGPT and Claude app-store billing is under-captured, so overlap and observed tenure are floors. The small Khanmigo cohort makes exact percentages directional, and first observed receipts may occur after the true subscription start.