The founder posts, the company page keeps the record

Should the founder post, or should the company page carry the content?

The founder's profile carries the content, and the company page carries the record. That is the answer we give in the room, and this post lays out the two pieces of evidence behind it: an audit of 703 citations from three AI engines, in which 35 of the 37 LinkedIn feed posts the engines cited were published from a personal profile (38 feed-post URLs, two of them pointing at the same post), and three client results we are cleared to publish by name, each from our LinkedIn program, which publishes from the executive's own profile.

The question comes up in most first meetings about LinkedIn, often from a founder who would rather the company carried the exposure. The two evidence sets answer it from different directions. The audit measures where AI engines went when they needed a source. The client results measure what a program run from a personal profile produced. Neither is a controlled experiment, and we say that once here so it does not need repeating: the audit records what was cited and cannot say what was available to cite, and the client results are outcomes of a program with no company-page version running beside it for comparison.

definitionFounder-led LinkedIn program

A founder-led LinkedIn program publishes under the executive's own name, from the executive's own profile, and builds each post from an interview with the executive, in the executive's voice, with the executive approving each post before it goes live. The company page holds the record the posts point to: the offer, the published results, and the reference pages a buyer checks after the meeting.

Which LinkedIn posts did the AI engines cite?

In August 2026 we ran 36 questions drawn from real sales conversations through ChatGPT, Claude and Perplexity with web search on, and the full audit is published separately. The run produced 108 usable answers carrying 703 citations across 489 unique domains, counting a domain once per answer; linkedin.com was the most-cited domain, appearing in 28 of the 108 answers, ahead of reddit.com at 18, and all 28 of those answers were Perplexity's.

Those 28 answers cited 55 distinct LinkedIn URLs, and the URL says who published each one. In our count, 38 were feed-post URLs, whose path carries the account handle before the underscore; 15 were LinkedIn articles, whose slug usually ends in the author's name; and 2 were topic pages LinkedIn assembles itself. Of the 37 posts behind those URLs, 35 were published from a personal profile and 2 from a company page. Of the 15 articles, 10 carry a person's name in the URL, 4 carry a company's, and 1 carries neither.

The classification is a reading of the URL and a handle could be misread at the margin, so the count to hold onto is the shape: in our audit, when an engine reached for LinkedIn, it reached for a person 35 times out of 37 feed posts. The 28 answers were spread across all four question types we asked, 10 comparison, 6 diagnostic, 7 how-to and 5 definition, so the reach for a person held whether the buyer was choosing between two options or asking what a term meant. The largest group of the cited posts, 15 of the 37 by our coding, opens by stating a position, and that shape gets its own post.

Two limits belong next to the number. All 28 answers that cited LinkedIn came from Perplexity; ChatGPT and Claude cited linkedin.com on none of their 72 answers in our run, so the finding describes the engine that cites most rather than all three equally. And because the audit records what was cited, it cannot say how many company-page posts existed on these questions and were passed over. It can say that with the whole web available, the engine's LinkedIn sources were people, 35 of 37.

Our explanation, offered as a hypothesis, is that a citation needs a specific claim attached to a name that can be held to it, and a person's post carries both in its first line while a company page's post more often carries an announcement. The run does not test that. The test is the same 36 questions re-run at 60 days, with a count of whether any company page has entered the cited set.

What did programs run from a personal profile produce?

The second evidence set is smaller and closer to home: three client results we are cleared to publish by name, each from our LinkedIn program, which publishes from the executive's own profile. Wen Zhang, founder of If Not Now Wen, had a single post reach 6,989,013 views. Kjael Skaalerud, Managing Partner at Skaling Ventures, drew 1,122 newsletter clicks in one month. Bardo, a sustainability platform, received 61 inbound leads in one month.

Views on a single post, newsletter clicks in a month and inbound leads in a month are three different things at three different depths of a funnel. The first measures how far one post travelled, the second measures traffic into an audience the executive owns, and the third measures conversations the executive did not have to start. A program is judged on whichever of the three the executive set out to get, and the table keeps them apart for that reason.

The three executives differ in audience size, in subject and in standing before the program began, and a single post at 6,989,013 views says at least as much about that post and its moment as about the account it was posted from. The results show what our program, publishing from the executive's own profile, produced for three clients who agreed to be named. They do not show that the same posts from a company page would have produced less, and our hypothesis for why they land is the same one the audit suggested: a reader, like an engine, responds to a person holding a position.

What is the company page for?

The company page is where we point the buyer after the meeting. In the first meeting the founder makes the argument; afterwards, the prospect looks up the site and the page, and what they need there is the record: the offer, the published results and the reference pages the founder quoted. Every result in the table above is published on our company site, where a prospect can read it after the meeting, and the three rows are there because a page is where a number can be filed with its source. Our data supports a division of labour rather than a winner.

How does a founder publish 12 posts a month without writing them?

The mechanics are on the service page and they are short. We interview the executive for 30 minutes a month and turn it into 12 posts in the executive's voice. Before the interview we research the topics in the executive's niche that are drawing the most response, so each post starts from a subject with demand behind it. After it, finished posts land in the executive's queue for approval, and the executive controls what goes live. We schedule and publish at the times the audience is on the feed, support distribution in the first hours after a post goes up, and review the month's analytics to feed what worked back into the next interview.

Two boundaries hold the program together. Each post is built from the interview, in the executive's voice, and the executive approves each post before it goes live. Those two rules are what let a post under the founder's name be the founder's position: the interview supplies the position and the approval step confirms it.

So, founder profile or company page? The founder's profile carries the content and the company page carries the record. In our audit the one engine that cited LinkedIn reached for a personal profile on 35 of the 37 posts it cited, and the three client results we can publish came from our LinkedIn program, which publishes from the executive's own profile. We will re-run the same 36 questions at 60 days, and if company pages have entered the cited set by then, this post gets a correction with the same numbers attached.