How ChatGPT and Perplexity Decide Which Brands to Recommend (And Why They Disagree)

Start with the finding that should reorganize your entire AI search plan.

Only around 11% of the domains ChatGPT cites are also cited by Perplexity, according to analysis of 680 million citations, with a separate study of 118,000 responses arriving at the same figure.

Eleven percent.

These two systems are answering the same questions, for the same buyers, from the same internet, and they are almost entirely disagreeing about which sources are worth using. BrightEdge benchmarking across ChatGPT, Google AI Mode and AI Overviews found the engines disagreed on which brands to recommend 62% of the time.

So if someone offers you "AI search optimization" as one service with one method, they are running the same campaign on two channels that share almost no audience. It is not a strategy. It is a hope.

Here is how each one actually behaves, and what to do about it.

They do not even mention brands at the same rate

Before tactics, calibrate your expectations.

A 2026 study of 34,234 AI responses found ChatGPT named brands in just 0.59% of responses, against 13.05% for Perplexity, a 46-fold difference. Grok was higher still at around 27%.

ChatGPT is extremely reluctant to recommend specific companies. Perplexity is comparatively happy too. That single fact should change where you spend effort first, and it explains why so many brands conclude "we tested ChatGPT and we're invisible." Most brands are invisible there. That is the default state, not a failure of your SEO.

What ChatGPT trusts

ChatGPT leans on what the broader consumer internet has already established about you. Directories, Wikipedia, news coverage, LinkedIn, and a long tail of editorial sources.

Notably, user-generated content from Reddit and Quora dropped out of ChatGPT's top sources entirely after platform updates in late 2025, falling to roughly half a percent.

Translated into work: ChatGPT visibility is earned slowly, through the things PR teams have always done. Tier-one media mentions. An accurate Wikipedia presence if you qualify for one. Consistent directory listings. Being described the same way in fifteen places by people who are not you.

This is the slow lane. It compounds and it is hard for competitors to undo.

What Perplexity trusts

Perplexity behaves almost oppositely. It favors high-authority specialist sources, refreshed recently, and it is friendlier to community content.

The practical implications are concrete: publish frequently, put visible dates on everything, structure content as short extractable answers, and have a presence in the communities relevant to your category.

This is the fast lane. You can move Perplexity visibility in weeks. You can also lose it in weeks, because freshness cuts both ways.

The LinkedIn detail that nobody is acting on

This is the most specific and immediately usable thing in this article.

LinkedIn has climbed sharply in AI citation share, ranking behind only Reddit overall across major platforms. One 325,000-prompt study found LinkedIn cited in 14.3% of ChatGPT Search responses and 5.3% of Perplexity responses, and for professional and B2B queries LinkedIn is the single most-cited domain across all six major AI platforms. 

Now the part that matters: ChatGPT and Google AI Mode take 59% of their LinkedIn citations from individual member posts, while Perplexity inverts this and takes 59% from Company Pages. 

So the instruction splits.

Want ChatGPT visibility? Your founders and senior people need to be publishing substantive analysis under their own names. Not reposts. Not "excited to share." Actual arguments with something at stake.

Want Perplexity visibility? Your company page needs to be complete, current, and posting regularly with dates.

Most B2B companies do exactly one of these and wonder why they show up on one platform only.

The reason most brands are invisible has nothing to do with strategy

Before optimizing anything, check whether you have made yourself readable. The data here is brutal.

One analysis of over a million AI citations found 73% of sites carry technical barriers blocking AI crawler access. Three quarters. If you are blocking the crawlers, nothing else in this article applies to you.

It gets worse at the page level. Across 2,225 pages audited, 36% were thin or non-extractable, 77% carried no visible date, and only 21.2% showed author signals.

Read that again in terms of your own site. Does every important page have a visible publication date? A named author with a real bio? Enough substance on the page to extract a claim from?

Roughly 53% of brands do not appear in AI answers at all, and it is worse in trust-sensitive categories: finance brands were cited in just 2.1% of checks, and 86% of legal and professional services brands were never cited once.

If you are in one of those categories, the opportunity is enormous, because your competitors are almost certainly invisible too.

The unglamorous checklist, in order: unblock the AI crawlers, put visible dates on every page, add named authors with credentials, make sure each key page contains an extractable claim in its first hundred words, and add schema. That is most of the gap for most companies, and none of it requires a strategy meeting.

Where third-party sources fit

Your own site matters more than the discourse suggests. The most-cited format across audits is the brand's own website. But you cannot rely on it alone.

Analysis of 680 million citations found the top 15 domains capture around 68% of consolidated AI citation share, with Reddit alone accounting for roughly 40%.

Worth being honest here: the studies disagree. Reddit dominates consolidated share across all engines while having fallen sharply within ChatGPT specifically. Both things are true, which is exactly the point of this article. Aggregate AI search numbers hide platform-level realities. Read the per-platform breakdown or you will make the wrong call.

One analysis of 200 million prompts found that even the most-cited domain on any single platform rarely exceeds 5% of total citations, with the remaining 95% spread across thousands of domains. That is a long-tail market, not a winner-take-all one, which is genuinely good news if you are not a household name. 

Measure per platform or do not bother

The single biggest mistake we see in AI visibility reporting is the aggregate score. One number that averages ChatGPT, Perplexity, Gemini and AI Overviews tells you nothing, because those engines disagree with each other most of the time.

Track separately. Same prompt set, run monthly, logged per platform. Record whether you appeared, whether you were cited, which URL or third-party source got the citation, which competitors appeared, and how you were described.

That last one earns its place. If ChatGPT is describing your business inaccurately, there is no other way to find out, and it is fixable.

Why this is worth the effort

Because the traffic is unusually good.

A synthesis of six independent studies covering 680 million citations put AI search conversion at 14.2% against 2.8% for Google organic, roughly a five-fold advantage, with Claude users converting at 16.8%, ChatGPT at 14.2% and Perplexity at 12.4%.

Fewer visitors, dramatically better ones. Which makes sense: someone arriving from an AI recommendation has already had their question answered and their options narrowed. They are not browsing. They are checking you out.

That is the actual argument for doing this work, and it is a stronger one than "everyone is talking about AI search."

The short version

There is no formula, and anyone selling you one is guessing.

What there is: two engines with almost no overlap in what they trust, a majority of brands who have accidentally blocked themselves from both, and a small number of boring fixes that most competitors have not made.

Fix the readability first. Then pick your lane deliberately rather than running one campaign at two different audiences.

Want to know where you stand on each platform separately, not as one meaningless average? Book a free growth audit. Thirty minutes, and you leave with your per-platform baseline and a written first move, whether or not you hire us.