Perplexity vs ChatGPT: who gets named.
Same question, two different shortlists. The difference is not random, and it changes what you should fix first.
One shows you its sources. The other just tells you who to call.
That single difference explains most of what we see when we ask the same question on both.
- Perplexity shows receiptsEach claim carries a numbered source you can open. If you are not on a page it cites, you are unlikely to be in its answer.
- ChatGPT gives a shortlistNames appear in prose, usually with a reason attached. The sources are implied, not listed, so you cannot see exactly which page earned the mention.
- Both depend on written factsNeither invents a company profile. If your own pages, listings and registration disagree, both engines have to pick one version.
- Neither is stableAsk twice, get two answers. Treat any single reading as a sample, not a score.
Four things we do differently per engine.
For Perplexity: be on a page worth citing
Because the citation is visible, the question becomes concrete: which published pages would a careful researcher cite for this question? A directory profile, a trade association listing, a comparison article, a supplier page. Writing another page about yourself does not create a citation; being included somewhere that already gets cited does. When we work on this, the target is a specific URL on someone else's domain, not a word count on yours.
For ChatGPT: be consistent across several places
With no visible source list, the practical lever is repetition: the same company description appearing in several independent places, worded similarly. That is why a single directory listing rarely moves anything, while five consistent listings plus one partner page can. Absence here usually means "not enough places", not "bad website".
Both: give them one version of the facts
Legal name, registration number, address, what you sell, who you serve. If these disagree between your site, your registration and your listings, each engine resolves the conflict its own way, and you end up with two different descriptions of the same company. Fixing that is unglamorous and it is the highest-yield hour in the whole engagement.
Both: speak the buyer's language, literally
A question asked in Chinese and a question asked in English can produce different shortlists, because the underlying sources are different. If your buyers search in both, you need presence in both, not a translated page on one domain.
Two engines, two separate counts.
- Same question setIdentical wording, so the only variable is the engine. Changing the question between engines makes the comparison meaningless.
- Separate recordsOne column per engine, per month. A combined "AI visibility score" hides the only thing worth knowing: where you are missing.
- Counts, not percentagesWith a handful of questions, "60% visibility" is arithmetic theatre. Named out of asked is honest and checkable.
- Keep the raw answerEvery reply, word for word, with date and language. Without it, a month later nobody can tell what changed or why.
Four straight answers.
Is one engine more important than the other?
Depends on your buyers, and you can find out: ask them which assistant they use, or watch which referrers arrive on your site. For B2B research in Singapore we see both, with Perplexity more common in early research and ChatGPT more common in the final shortlist.
Why is my competitor named and I am not, when we do the same work?
Usually because they are written about in more places, not because they are better. Ask the reverse question: what would a researcher have to read to conclude they are the obvious choice? Then get your name onto those pages.
Can I influence the sources Perplexity cites?
Indirectly, yes: by being included in the pages it already cites, and by publishing facts clear enough to be quoted. You cannot submit a citation, and anyone claiming to place one directly is describing advertising, not citation.
Do Google AI Overviews follow the same rules?
Closer to Perplexity: sources are shown, and they are drawn from what Google already indexed. Which means for that surface, ordinary SEO health is not optional, it is the entry condition.
Find out what AI says about you.
Free, with no obligation. You keep the report either way.