Executive summary and key findings#
TL;DR
Perplexity cited companies more often than individual authors. That's the opposite of much current AEO advice, which tells brands to put a named expert behind everything, drawing on findings like Meltwater's study with LinkedIn: 75% of LinkedIn citations traced to individual member profiles. We tested whether that holds beyond LinkedIn, across 5,770 Google page-one web pages for 400 questions.
Perplexity cited 2,022 of 5,115 pages published by companies (39.5%) and 98 of 361 pages by individual authors writing under their own name (27.1%). A page by an individual author had 60.4% lower odds of being cited than a page published by a company in the same position.
To get cited by Perplexity, rank on Google page one first, then publish your strongest expertise on your company's own site, not only on personal channels.
Key findings on companies vs individual authors in Perplexity citations#
- Perplexity cited companies more often than individual authors. It cited 2,022 of 5,115 pages published by companies (39.5%), 98 of 361 pages by individual authors (27.1%) and 85 of 294 pages by anonymous authors (28.9%).
- The gap held for pages ranking in the same Google position on similarly strong websites. A page by an individual author had 60.4% lower odds of being cited than a page published by a company in the same position. If you're choosing where to publish expertise, your company's own site is associated with better odds of citation.
- Companies were cited more often in 7 of 8 industries.
- Google ranking position mattered far more than who published the page. On its own, Google ranking position explained more than 30 times as much about which pages Perplexity cited as publisher type or domain strength did. Check that a page ranks on Google page one before worrying about who signs it.
Cite this study
Discovered Labs (2026). Companies outperform individual authors in Perplexity citations: 39.5% vs 27.1%.
https://discoveredlabs.com/research/perplexity-citations-companies-vs-individual-authors
Charts and tables on this page may be reproduced with attribution and a link to the source.
Methodology and scope#
We ran 400 questions across eight industries through Perplexity, split evenly between commercial and advice-style questions. Each question ran three times, because answers vary between runs: 6.7% of citations differed between at least one run and another for the same question. A candidate page (one of the Google page-one results for its question) counted as cited if it appeared in at least one of the three runs.
Each question's candidate pool was its Google page-one organic results, captured in the same session as the citations. Every page in the pool was labeled as a company, an individual author, or an anonymous author before citations were examined, using a rubric fixed in advance. 95.4% of pages were classified with high confidence from the URL, domain, and search-result title and snippet alone. The remainder were flagged as ambiguous and are included in the totals throughout.
Of the 7,097 candidate pages, 5,770 were regular web pages, and every result in this study is based on those. The other 1,327 were posts on Reddit, YouTube, Quora, Facebook, and LinkedIn, excluded because publisher type on those platforms is largely fixed by the platform itself: every Reddit and Quora post is anonymous. Including them does not change the main result.
The analysis compares pages that ranked in the same Google position on similarly strong websites, and checks the result is not a fluke of repeated websites or of running several tests. Technically, the primary model is a logistic regression of citation on publisher type, ranking position, and domain authority, with standard errors clustered by question, a two-way question-and-domain clustering check, and multiple comparison correction across the primary tests. Full model output is in the technical appendix at the end of this page.
What LinkedIn AI search studies say about individual profiles vs Company Pages#
LinkedIn studies say personal profiles win AI citations#
Two widely cited 2026 industry reports shaped a lot of current AEO advice on author attribution. Meltwater, working with LinkedIn on a study of 9.5 million AI citations, reported that 75% of LinkedIn citations traced to individual member profiles versus 25% to Company Pages, and that follower count wasn't the gate for who got cited: about 51% of the cited individual profiles had fewer than 10,000 followers.
Semrush's analysis of 89,000 cited LinkedIn URLs across 325,000 prompts found the split depends on the AI engine. Individual members made up 59% of LinkedIn citations on ChatGPT Search and Google AI Mode, while Company Pages made up 59% on Perplexity, the engine this study uses.
General AEO guidance points the same way as the LinkedIn studies. Search Engine Land's analysis of what ChatGPT quotes credits genuine expertise, and Quora's guide to getting cited by ChatGPT tells brands to back claims with named expertise.
Research disagrees on whether author bylines matter#
The Meltwater and Semrush reports count what AI engines cited, not what they cited out of everything available. Neither can show whether a personal byline changes a page's chances of being cited, only what the final mix of citations looked like.
A third study, from Digital Authority Partners, scored 497 Google organic results across 30 queries, with two independent reviewers coding the signals, and found that author bylines and credentials had no measurable effect on citation. That study asked whether a byline was present, not whether a company or a person published the page, and used a smaller sample.
This study compares citation odds for companies and individual authors against a pool of pages that were all eligible to be cited, built before any citation was observed.
Perplexity cited companies more often than individual authors#
Swipe to explore →
Perplexity cited 2,022 of 5,115 pages published by companies (39.5%), against 98 of 361 pages by individual authors (27.1%) and 85 of 294 pages by anonymous authors (28.9%).
Source: 5,770 Google page-one web pages for 400 questions, each run three times on Perplexity; see Methodology and scope.
Perplexity cited individual authors 12.4 points less often than companies: 27.1% against 39.5%, a 31.3% lower rate. After accounting for Google ranking position and domain authority, a page by an individual author had 60.4% lower odds of being cited than a page published by a company in the same position. The first comparison uses raw citation rates. The second compares pages that ranked in the same position on similarly strong websites, which is the fairer comparison.
Individual authors still get cited. Perplexity cited pages by individual authors more than a quarter of the time. But between a page published by a company and a page by an individual author ranked in the same Google position, Perplexity cited the company's page more often.
In practice, if a piece of expertise could live on your company site or a personal channel, the company site is associated with better odds of being cited.
How we classified companies, individual authors and anonymous authors#
Every page in the candidate pool was assigned one of three labels before we looked at whether it was cited.
Company covers organizations, products, and publications, including a company blog post with a named staff byline. A page on a company's own blog that says "written by [a named staff member]" is still classified company, not individual author, because the page trades on the publication's authority. The test applied: would this page exist, and carry the same weight, with a different staff writer's name on it? If yes, it's company. That same person's own personal site would be classified individual author.
Individual author covers content published under a real, identifiable personal name, where the content trades on that person's own authority rather than an employer's, such as a personal blog or a Substack under someone's own name.
Anonymous author covers a real person publishing under a handle that discloses no real name, such as a forum post or a Stack Exchange answer. It is kept separate from individual authors because a reader has no way to verify who a pseudonymous account is, while an individual author's identity can in principle be checked. Anonymous authors were cited at a similar rate to individual authors (28.9% against 27.1%), and both trailed companies (39.5%).
Each page was classified once against a written rubric.
Companies stayed ahead in five stricter checks#
We tested the gap between companies and individual authors five more ways, and it held every time.
Domain strength doesn't explain why companies win#
Personal sites usually have weaker backlink profiles than company sites, so a weaker domain, not the byline, could explain why individual authors were cited less. Pages by individual authors in this sample did sit on somewhat weaker domains than pages published by companies. But once domain strength was accounted for, the gap between companies and individual authors barely changed.
Domain strength still matters for AI search in general. Our earlier study of 2 million AI citations found AI-perceived domain authority, a measure of how much trust AI engines have built up for a domain, roughly 6 times as influential as the strongest page-level signal other than prompt-content alignment (how closely a page's wording matches the way people phrase questions to AI). Domain strength just doesn't explain this particular gap.
Companies stay ahead under a stricter statistical test#
Some websites appear as candidates for more than one question, which can make a result look more certain than it is. Rerunning the analysis with a stricter method that accounts for repeated websites widened the margin of error, but the gap between companies and individual authors stayed statistically significant, meaning it's very unlikely to be chance. The link between domain strength and citation did not survive the stricter test, another sign that domain strength isn't driving the gap.
Companies led for both buying and how-to questions#
The gap between companies and individual authors did not differ meaningfully between buying questions ("which tool should I buy") and how-to questions ("how do I do this"). Individual authors were cited slightly more often on buying questions (30.0%) than on how-to questions (26.1%), but the sample is too small to call that a real difference.
Companies led in 7 of 8 industries#
Companies were cited more often than individual authors in 7 of 8 industries.
Swipe to explore →
Companies were cited more often than individual authors in 7 of 8 industries, with home and DIY the one exception at 40.3% against 41.7%.
Source: 5,770 web pages across eight industries; samples of individual authors range from 22 to 108 pages per industry.
Home and DIY was the one exception, with the two rates close enough to read as no difference. Crypto and personal finance had the fewest pages by individual authors, so those two rows carry the most uncertainty.
Companies led in every version of the data#
The gap between companies and individual authors held in every version of the data we tested.
Swipe to explore →
Companies were cited more often than individual authors in every version of the data, from 39.5% against 27.1% across all pages to 58.1% against 41.4% in the top 10 Google results.
Source: 5,770 Google page-one web pages for 400 questions; each version filters the same dataset.
Google ranking position matters far more than authorship#
Of everything measured, Google ranking position was by far the strongest predictor of which pages Perplexity cited. Who published the page and domain strength each had a small association of about the same size, and the publisher association held up better under the stricter test.
Swipe to explore →
Google ranking position scored 0.163 on its own, more than 30 times the 0.005 for who published the page and the 0.004 for domain strength.
Source: how much each factor explains about citation on its own, N = 5,770 web pages. Full model in the technical appendix.
The full model output is in the technical appendix at the end of this page.
Why we tested Perplexity and what comes next#
The test ran on Perplexity because a pilot run showed it searches and cites sources consistently for both buying and how-to questions. Another major AI assistant often answered how-to questions without searching the web, which leaves no citations to compare. Semrush's LinkedIn data shows the individual-versus-company split differs between engines, so ChatGPT and Google AI Mode need their own test.
The candidate pool was Google's page-one results, not everything Perplexity can retrieve, so the result describes pages that already rank on Google page one.
The study compares real pages as they exist on the web. It did not republish the same page under different names, and it did not measure content format or freshness, which makes those the natural next tests.
How to get cited by Perplexity: the priority stack#
Here is the priority stack for a marketer deciding where to spend effort to get cited by Perplexity. Each item depends on the items above it already being in place.
Four priorities, in order
- 1. Get the page ranking on Google page one. Google ranking position explained more than 30 times as much about which pages Perplexity cited as publisher type or domain strength did.
- 2. Build domain-level authority as a standing foundation. In this study, domain strength had only a small link to citation among pages already ranking in a similar band. Our earlier study of 2 million AI citations found AI-perceived domain authority roughly 6 times as influential as the strongest page-level signal other than prompt-content alignment. Treat domain authority as the layer underneath everything else.
- 3. Decide where a piece is published before deciding who signs it. A page by an individual author had 60.4% lower odds of being cited than a page published by a company in the same position. Publishing high-value expertise on company-owned pages, not only on personal channels, is a decision most content teams can make right away, without waiting for a domain-authority campaign to pay off.
- 4. Treat authorship as a secondary layer. A staff byline on a company's own site still counts as company content. Get the page ranking, build the domain, choose the publication deliberately, and then decide who signs it.
Priority summary#
Swipe to explore →
Google ranking position comes first, explaining more than 30 times as much about citation as the next factor. Who published the page ranks third.
Source: this study (N = 5,770 web pages) and our earlier study of 2 million AI citations; byline evidence from Digital Authority Partners.
Technical appendix: full model output#
Model: cited_any ~ C(attribution) + rank + domain_rank, logistic regression, N = 5,770 web pages, standard errors clustered by question unless noted.
| Term | Odds ratio | 95% CI | p |
|---|---|---|---|
| Individual author (vs. company) | 0.40 | [0.30, 0.53] | < 0.001 |
| Anonymous author (vs. company) | 0.58 | [0.44, 0.77] | < 0.001 |
| Ranking position (per position) | 0.81 | [0.80, 0.82] | < 0.001 |
| Domain authority (per point) | 1.0004 | [1.0001, 1.0008] | 0.022 |
Standalone contribution of each predictor (McFadden pseudo-R² against an intercept-only model): ranking position 0.163, attribution 0.005, domain authority 0.004, full model 0.171.
Two-way clustering. 17.5% of domains appear in more than one question's candidate pool. Re-clustering standard errors by both question and domain gives p = 0.001 for individual author and p = 0.004 for anonymous author. Domain authority's p-value moves from 0.022 to 0.19.
Question-type interaction. Individual author x advice: OR 0.71, 95% CI [0.37, 1.36], p = 0.30. Anonymous author x advice: OR 1.67, 95% CI [0.91, 3.07], p = 0.10. Raw individual-author rates: 30.0% commercial, 26.1% advice. The interaction test is underpowered relative to the main effect.
Domain strength by category. Median domain score, on a 0 to 1,000 scale where higher is stronger: individual author 299, company 378, anonymous author 483. The prior-research comparison, AI-perceived domain authority against page-level signals, comes from a SHAP analysis across 2 million citations.
Industry 95% confidence intervals, company / individual author. Project management 35.9% [32.3, 39.7] / 21.6% [12.5, 34.6]. Marketing / SaaS 41.7% [38.1, 45.4] / 28.6% [15.3, 47.1]. Fitness 41.3% [37.7, 45.0] / 25.8% [13.7, 43.2]. Career / job search 41.2% [37.4, 45.2] / 30.2% [20.2, 42.4]. Dev tools 36.4% [32.5, 40.5] / 25.9% [18.6, 34.9]. Crypto / investing 39.0% [35.4, 42.7] / 4.5% [0.8, 21.8]. Personal finance 39.9% [36.4, 43.6] / 36.4% [19.7, 57.0]. Home / DIY 40.3% [36.3, 44.4] / 41.7% [27.1, 57.8].
Frequently asked questions#
How do you get cited by Perplexity?
Start by ranking on Google page one, because Google ranking position explained more than 30 times as much about which pages Perplexity cited as publisher type did. Then publish on a strong company domain. Across 5,770 Google page-one pages, Perplexity cited 39.5% of pages published by companies and 27.1% of pages by individual authors.
Should you use LinkedIn or your company website for AI search visibility?
Use both, with your company website as the home for your strongest expertise. LinkedIn studies show individual profiles earn most LinkedIn citations, but Perplexity cited pages published by companies more often than pages by individual authors ranking in the same Google position. Publish the full piece on your company site, then share it through personal LinkedIn profiles.
Do AI search engines cite LinkedIn personal profiles or company pages more?
It depends on the AI engine. Meltwater found 75% of LinkedIn citations came from individual member profiles. Semrush found individual members made up 59% of LinkedIn citations on ChatGPT Search and Google AI Mode, while Company Pages made up 59% on Perplexity. In this study, Perplexity also cited web pages published by companies more often than pages by individual authors.
Do author bylines help content get cited by AI?
Bylines on their own have not been linked to more AI citations. Digital Authority Partners scored 497 Google results and found author bylines and credentials had no measurable effect on citation. In this study, a staff byline on a company's own site still counted as company content, and companies were cited more often than individual authors.
Do Google rankings affect which pages AI search engines cite?
Yes, strongly. Among 5,770 Google page-one pages, Google ranking position was by far the strongest predictor of whether Perplexity cited a page, explaining more than 30 times as much as who published it. Top 10 results were cited far more often: 58.1% of pages published by companies and 41.4% of pages by individual authors.
Does domain authority matter for AI citations?
Domain authority matters, but it did not explain the gap between companies and individual authors. Our earlier study of 2 million AI citations found AI-perceived domain authority roughly 6 times as influential as the strongest page-level signal other than prompt-content alignment. In this study, companies kept their lead after accounting for domain strength.
Is a company blog or a personal blog better for getting cited by AI?
A company blog is the better bet on current evidence. Perplexity cited 39.5% of pages published by companies and 27.1% of pages by individual authors, and a page by an individual author had 60.4% lower odds of being cited than a page published by a company in the same position. Companies were cited more often in 7 of 8 industries tested.
More research#
Continue with the other Discovered Labs studies on AI citations:
- Reddit and LLM citations: 144,284 AI citations analyzed. Reddit's outsized presence in AI answers, measured across 144,284 citations on ChatGPT, Google AI Overviews and Gemini.
- What actually drives AI citations: a statistical analysis of 2M AI citations across 10K pages. The source for the domain-authority comparison in this study's stricter checks.