Research
What AI recommends, and what fractional CMOs charge.
I publish original data on two questions my clients ask most: which brands ChatGPT recommends in their category, and what senior marketing leadership costs. Every study is dated, the method is below, and it’s free to cite with a link.
By Monica Hacker, fractional CMO. Updated October 28, 2026.
The studies
Eight studies, all dated.
Each ChatGPT index is 15 answers: five buyer questions, asked three times each. Results are observed samples, not platform rankings.
| Study | Tested | Sample | Headline finding |
|---|---|---|---|
| ChatGPT Baby Care Index | Nov 7, 2026 | 15 answers | Aquaphor, CeraVe and Pipette each named in 6 of 15; every skin care answer cited the National Eczema Association seal |
| ChatGPT Pet Index | Nov 7, 2026 | 15 answers | Purina Pro Plan first in 6 of 6 dog and cat food answers; treats were the open lane |
| ChatGPT Telehealth Index | Oct 28, 2026 | 15 answers | Nurx named in 6 of 15; 25 companies named overall |
| ChatGPT Supplement Index | Oct 28, 2026 | 15 answers | Nature Made named in 9 of 15; every answer cited a third party seal |
| ChatGPT Skincare Index | Oct 26, 2026 | 15 answers | La Roche-Posay named in 10 of 15; clean skincare was the open lane |
| ChatGPT Med Spa Index: New York | Oct 2026 | 15 answers | One med spa named in 12 of 15; dermatologists took 13 of 27 injectable picks |
| ChatGPT Med Spa Index: Miami | Oct 2026 | 15 answers | One med spa named in 7 of 15; dermatologists and surgeons took 13 of 24 injectable picks |
| Fractional CMO Pricing 2026 | Oct 12, 2026 | 20 providers | Median starting retainer $6K a month; median top tier $13K |
Free to cite with a link to the study page. For press questions, reach me through the contact form.
The method
How every ChatGPT index is run.
Buyer questions, not keywords
Five questions phrased the way a shopper or patient asks them, such as “What’s the best magnesium supplement brand?”
Three runs each, fresh chats
Every question is asked three times in new ChatGPT temporary chats with no memory, history or custom instructions.
Count what’s named
A brand counts once per answer if it is recommended or listed as an option. Brands named only as a warning are counted and flagged separately.
Note the context
The date, the location the test ran from (New York so far) and anything that shaped the answers, such as local results, are recorded with the data.
Rerun and keep history
Indexes are rerun monthly. Earlier results stay published with their dates, so changes over time are visible.
Limits
What these numbers can’t tell you.
Fifteen answers is a sample, not a census. AI answers change from run to run and after model updates, and independent studies have found recommendation lists rarely repeat in the same order. That’s why I report how often a brand is named, not its position.
The indexes test ChatGPT only, from one location, in English. Treat them as a dated snapshot of what a typical buyer would have seen, and a starting point for testing your own category.
Can I cite this research?
Yes. Every study is free to cite with a link to its page. Please include the test date.
How often is the research updated?
The ChatGPT indexes are rerun monthly. The pricing benchmark is refreshed each quarter. Earlier results stay published with their dates.
Why only 15 answers per index?
It’s enough to see who is consistently named and who appears once, and small enough to rerun every month. Repeated monthly samples say more than one large snapshot.
Can you run this for my category or brand?
Yes. That’s the AI Visibility Check: your real buyer questions, tested across AI platforms, with a prioritized fix list. $2,500, delivered in 5 days.
Want this for your category?
The AI Visibility Check runs the same method on your buyer questions and shows what to fix first. $2,500, delivered in 5 days.
Book a call →Thirty minutes. Bring your numbers, or just bring the problem. You'll leave knowing what I'd fix first.
