The 37 LinkedIn posts an AI engine cited, read from their first lines
Perplexity cited 37 individual LinkedIn posts in our 36-question audit. Coded from their opening words, 15 state a position and 7 open on the same hard hire. In this set, the first line was the answer.
[ key takeaways ]
- In our August 2026 audit of 36 buyer questions across three AI engines, linkedin.com was the most-cited domain, and all 28 answers that cited it came from Perplexity; ChatGPT and Claude cited it on none of their 72 answers.
- The 28 answers pointed at 55 distinct LinkedIn URLs: 37 individual posts (one cited under two URLs), 15 articles and 2 topic pages, and 35 of the 37 posts were published from a personal profile rather than a company page.
- Coded from their URL slugs, 15 of the 37 cited posts open by stating a position a reader could disagree with, and 7 open on hiring a GTM engineer, all 7 cited on one how-to question about not being able to make that hire.
- Comparison questions drew a LinkedIn citation on 10 of 12, and the how-to questions tied comparison at 12 individual posts each, 7 of them on the single GTM-engineer hiring question.
- Six of the 37 cited posts open with hashtags, so their first line is not readable from the citation.
Which LinkedIn posts did an AI engine cite when buyers asked about go-to-market?
When we put 36 buyer questions to ChatGPT, Claude and Perplexity in August 2026 and published what the engines told our buyers, one domain led the citation count. The run produced 108 answers carrying 703 citations, counting each cited domain once per answer, across 489 unique domains, and the most-cited of the 489 was linkedin.com, cited in 28 of the 108 answers. This post is about what sat behind those 28, because a founder who wants a post pulled into a buyer's answer should see what the pulled ones look like, and the honest way to show that is to show the posts.
Three facts about the 28 frame everything below. All 28 came from Perplexity, which cited linkedin.com in 28 of its 36 answers; ChatGPT and Claude cited it in none of their 72. The 28 answers pointed at 55 distinct LinkedIn URLs: 38 pointing at individual posts, one of which was cited under two URLs, so 37 posts; 15 long-form articles published through LinkedIn's article format; and 2 LinkedIn topic pages. And 35 of the 37 posts were published from a person's profile rather than a company page, judged from the handle in each URL, which is the one sentence we will spend on that point here, since a companion post covers it.
The 37 individual posts are the material for the rest of this post.
What do the cited posts look like?
A LinkedIn post URL carries the post's first words as its slug, so the opening line of a cited post is readable from the citation itself. We read four of the 37 in full; the other 33 are described from their opening lines and from the audit question that pulled each one, and every count below is built that way.
[ fig. 01 · thirteen of the 37 cited posts, by opening words ]
Two things about the table need a note. Six of the 37 posts have a URL slug that carries hashtags rather than words, so their opening lines are unknown to us; none of the six is in the table, and the next figure counts them as unknown rather than guessed. And one of the 37, cited on a definition question, was our founder's own post. It stays in the count because removing it would misstate the count. Perplexity cited it on the same terms as the other 36, and the earlier post's callout counts it alongside our maturity model page as two GrowthMasters citations in one Perplexity answer.
What does the first line of a cited post do?
We coded each of the 37 opening lines into one of seven shapes, one shape per post, from the slug alone.
[ fig. 02 · the 37 cited posts by the shape of their first line ]
Inside the hiring bar: five of the seven say in the first line that the hire is hard, and all seven were cited on the same how-to question.
The figure carries the counts. Two of them matter for what follows: 15 of the 37 open by stating a position a reader could disagree with, and seven open on hiring a GTM engineer, five of those saying in the first line that the hire is hard.
The counts sharpen a summary we published earlier, which described the cited posts as one specific claim with a number or a named mechanism. The claim holds as the first line's job in 15 of the 37. The number arrives later, where it arrives at all: two of the 37 slugs carry one in the first words.
The four we read in full show what follows the first line.
The post under the handle sreedharpeddineni, whose first line calls GTM engineering the new architecture function, makes a dated prediction in its second sentence, names the problem in one line, borrows an analogy from infrastructure engineering, and closes its argument on precedent: three earlier roles, two of them with the years each went from a title to a function and one with a count of companies attached. It ends in the first person, naming the role early so the people already doing the work have a word for it.
The post under the handle noemiejacquemin, which opens by asking why hiring a GTM engineer is so difficult, is 132 words by our count. It answers its own question four times, each answer a one-line reason with a one-line explanation under it, and closes with a three-part formula for getting the hire right.
The post under the handle james-hickey-787261201, which opens with the outsourcing question and answers yes and no in the same line, is 163 words. It splits the function into the parts that can be handed out and the parts that cannot, names two well-known companies that kept it in-house, adds a qualifier for companies going from zero to one, and ends by asking the reader what they have outsourced.
Juan's own post opens with its claim, contrasts two ways a team's workflows can run, and lists four conditions the second way rests on, one per line.
- definitionCitable post
A citable post is a LinkedIn post whose first line states the answer to a question a buyer is asking, in the words the buyer would use, with the reasons in the body short enough to read in a feed and specific enough to check. A citable post can be lifted into a machine-assembled answer one sentence at a time, because each sentence carries a claim and the thing that supports it.
What the four share is the order. The first line is the claim, the body is the reasons, and the reasons are checkable: a date, a count, a named company, a formula. Each fits the time a feed allows, and each is specific enough that a machine assembling an answer to whether RevOps can be outsourced can lift one sentence from it as the answer.
Which buyer questions pulled LinkedIn posts?
[ fig. 03 · linkedin citations by question type ]
One question carries the how-to row: what to do when you cannot hire a GTM engineer drew 7 of that row's 12 posts.
By question count, comparison questions drew the most LinkedIn citations, 10 of the 12 in our set. By post count, the how-to band tied the comparison band at 12 posts each, and one question did most of that work. Asked what a B2B company should do when it cannot hire a GTM engineer, Perplexity cited seven individual posts, and all seven open on hiring a GTM engineer. Seven of the posts it cited state that premise in their first line.
Eight of the 28 answers cited LinkedIn without citing an individual post; those cited the long-form articles or the topic pages only. Twenty of the 36 questions drew at least one individual post.
What we measured is which URLs one engine cited on 36 questions on one day in August 2026. The explanation we hold, as a hypothesis, is that a first line matching the words the engine searched for is what got a post pulled, since the engine composes its own searches and the cited posts' opening lines read like answers to them. The run does not control for how large each author's audience is, how each post performed in the feed, how recent it was, or what the engine's index held that day, and a post about opening lines cannot rule those out. The test is a re-run of the same 36 questions, and the earlier post commits to one at 60 days.
How do you write a post that gets pulled into a buyer's answer?
Write the first line as the answer, and let the counts above set the rest.
- State the position in the first line, as a sentence a reader could disagree with. Fifteen of the 37 cited posts open that way, the largest shape in the set.
- When the subject is a hard thing, say it is hard in the first line. Five of the seven posts pulled on the hiring question in our set do exactly that.
- When the buyer's question is a versus, put the versus in the first words. Three of the 37 do, and 15 of the 56 searches the engines composed for themselves in the earlier post were explicitly versus-shaped.
- Keep the body to reasons a reader can check. The two shortest of the four we read in full ran 132 and 163 words, and each carried a count, a named company or a formula.
- Put the first words where a scanner will read them. Six of the 37 cited posts open with hashtags, so their first line is not readable from the citation.
This is the shape our LinkedIn content work is built to produce. A 30-minute interview a month is where an executive's positions get said out loud with the reasons attached, and the writing moves the position to the first line and the reasons under it, 12 posts a month in the executive's own voice.
In our 36-question audit the most-cited domain was LinkedIn, all 28 of the answers that cited it were one engine's, and behind them sat 37 posts whose first lines mostly state a position, seven of them about the same hard hire. In this set, the posts that got pulled into a buyer's answer were the ones that answered in their first line.