Paid AI appeared before cancellation for a rising share of Chegg cancelers
Share whose first observed OpenAI, Anthropic, or Perplexity receipt preceded their first Chegg cancellation, by academic-year cancellation cohort
Share already paying for AI before canceling Chegg
Source: YipitData U.S. e-receipt panel. Academic-year canceler cohorts: AY21–22 n=2,771; AY22–23 n=2,180; AY23–24 n=1,693; AY24–25 n=1,088.
Chart data
| Series | Share already paying for AI before canceling Chegg |
|---|---|
| 2021–22 academic-year cancelers | 0.1% |
| 2022–23 academic-year cancelers | 10.8% |
| 2023–24 academic-year cancelers | 28.3% |
| 2024–25 academic-year cancelers | 35.2% |
Did paid AI distinguish Chegg cancelers from retained subscribers?
Only slightly. Paid AI appeared among 36.9% of 2024–25 Chegg cancelers and 33.2% of retained subscribers, a 3.7-percentage-point gap that the source treated as no meaningful difference. Among 862 AI-first cancelers from 2023–24 and 2024–25, the median observed coexistence before cancellation was ~11.8 months.
Did prior paid AI reduce Chegg reactivation?
No meaningful difference appeared within 12 months. Chegg activity resumed for 19.1% of 855 AI-first cancelers and 17.0% of 1,400 cancelers with no observed prior paid AI. The source treated the 2.1-percentage-point gap as no meaningful difference, so the evidence does not show that paid AI prevented a return to Chegg.
Assumptions & Methodologies
We analyzed YipitData’s U.S. e-receipt panel. Chegg cancelers were long-tenured panelists whose first explicit cancellation fell in an academic-year cohort from September 2021–August 2025. Paid AI meant a direct OpenAI, Anthropic, or Perplexity receipt strictly before cancellation. Reactivation required a later Chegg signal 31–395 days after cancellation. The panel over-represents technology adopters, men, and upper-income users, so absolute AI levels may exceed the student population. Receipt order establishes timing, not causality; some OpenAI billing may represent non-ChatGPT spending, and the newest cancellation cohorts have slightly shorter follow-up at the edge.