Who does AI name for test prep and tutoring?
SAT, ACT, GRE and LSAT prep, plus private tutoring for students.
The ranking
Last measured: October 22, 2026. 32 answers measured, 0 failed. AI answers change over time: this is a snapshot of that day, not a permanent verdict.
| Business | Times named | Share | Avg. position |
|---|---|---|---|
| The Princeton Review | 17 | 53% | 4.4 |
| Kaplan | 15 | 47% | 5.1 |
| Wyzant | 13 | 41% | 3.8 |
| Khan Academy | 10 | 31% | 5.4 |
| Mathnasium | 9 | 28% | 4.6 |
| Magoosh | 8 | 25% | 4.4 |
| Varsity Tutors | 7 | 22% | 2.7 |
| Sylvan Learning | 7 | 22% | 4.7 |
| Huntington Learning Center | 5 | 16% | 4.2 |
| 7Sage | 4 | 13% | 2.5 |
| GregMat | 4 | 13% | 2.8 |
| PrepScholar | 4 | 13% | 4.5 |
| College of DuPage | 4 | 13% | 4.8 |
| Blueprint | 3 | 9% | 2.3 |
| Tutor.com | 3 | 9% | 3.3 |
Showing the 15 most-named businesses, only those named at least 2 times. Ties go to the one the AI names earlier in its answers. Variants of the same name count as one business.
Left out: 4 names the AI mentioned that aren’t businesses competing in this industry (products, manufacturers, banks, universities, media or directories), such as College Board (public body or association), Bluebook (product), ETS (public body or association) and LSAC (public body or association).
The questions we ask
The same 8 buyer questions every month, none of them naming a business:
How we measure this
Engines asked: ChatGPT, Gemini, Perplexity, Claude. Each one answers as it would for someone in the US.
We send every question to every engine, then a separate model lists, in order of appearance, the specific businesses each answer named. Generic terms are dropped (search engines, government agencies, cities, review sites): only real businesses count.
Variants of the same name are merged: capitalization, accents, legal suffixes (LLC, P.C.), the website (“name.com”) and generic industry words around the name. Each answer counts once per business, even if it names it several times. Names that don't compete in this industry are left out even when the AI mentions them: products and manufacturers, banks and lenders, universities, media and directories (the place where you look for a provider isn't the provider).
A call that fails (the engine doesn’t respond, or the answer can’t be read) never counts as “this business was not named”. It is left out entirely, so the number of measured answers is always less than or equal to the number of questions sent.
Each business’s share is out of the answers we actually measured, never out of every question sent: if some calls fail, nobody’s percentage drops because of it.
Are you in this ranking?
This ranking covers the whole industry. The free check tells you how often YOUR business is named by the same four AIs, in what position, and who they name instead. No card needed.
Get my free check