What ChatGPT reads before it recommends an expert

7 in 10 sources were the named experts' own sites.

The 72 answers listed 551 sources, and 387 of them (70%) were on the websites of the experts ChatGPT named. Directories, profiles, rankings and other outside pages made up the rest.

Management consultants

78%

HR consultants

72%

Business coaches

70%

Executive coaches

69%

Leadership consultants

67%

Fractional CFOs

66%

B2B sales coaches

61%

The first name mattered too. In 49 of the 72 answers, the site of the first expert named was among the sources for that answer.

What it means: outside mentions help, but ChatGPT mostly learned about these experts from what they say about themselves. If your own site is vague, it has little to work with.

Half of what it read was the homepage. Blog posts were about 1%.

We sorted those 387 own-site sources by the kind of page, going by each page's address.

Homepage

190 (49%)

Service or program page

98 (25%)

About or team page

46 (12%)

Blog post or article

4 (1%)

Other

49 (13%)

What it means: when the question is "who should I hire", ChatGPT reads the pages that say who you are and who you help. Many experts' homepages open on a slogan. If yours doesn't say your profession, who you work with, the problems you solve and where you are, in plain sentences, fix that first.

This doesn't make articles useless. They may help in ways this count can't show, such as earning links or answering other kinds of questions. But they weren't what ChatGPT cited on these hiring questions.

Ranking on Google didn't predict being named.

For each question, we also took Google's top 10 for the short search a client would type, such as "executive coach for ceos toronto". Of the 183 expert websites ChatGPT named, 11 were in that top 10. Counting any search within the same profession, it was still only 22 of 165.

The lists are different because the questions are different. A client's question to ChatGPT describes a situation, and Google's top 10 for a short search holds many directories, big firms and job boards.

What it means: a small practice can be named without ranking on page one, and page one doesn't guarantee being named. Measure both separately. You still want Google, because people check you there too.

Small practices get named, and the lists aren't settled.

• Big names were rare. Of the 373 times ChatGPT named someone, only 7 were large global firms: Deloitte, PwC and McKinsey.
• People get named. In the executive coaching answers, 33 of 50 mentions were of a person rather than a company.
• Lists are short. An answer named between about 4 and 7 experts on average, depending on the profession. Fractional CFO answers were the shortest, at 3.8.
• Lists move. Of the 28 executive coaches named across 9 answers, 16 came up only once. Ask again, and you may get someone else.

What it means: nobody owns these answers yet, so it's worth trying now, and a single check tells you little. Ask each question several times.

Five changes, in the order we'd make them.

1. Rewrite the top of your homepage. Say your profession, who you work with, the problems you solve and where you work. Use the words a client would use.
2. Give each service its own page. Name the situation it's for: "coaching for newly promoted VPs", for example, rather than "leadership coaching".
3. Make your About page factual. Your background, credentials, the kinds of clients you've worked with, and your full name, spelled the same everywhere.
4. Get listed where your field is listed. Look at the outside sites in your own answers. These differ by profession: Clutch came up for management consultants, CoachVouch for executive coaches, and Find Finance Pros for fractional CFOs.
5. Check the answers, more than once. Ask your best client's question three times, in new chats, and see who comes up. Here's how we do it for coaches and consultants.

See what ChatGPT says about you

We'll show you every answer, name and source for your line of work, on one private page. It's free.