How does Perplexity choose its sources?
Know how Perplexity finds information before deciding what to change on your website.
Steen Stones · Reviewed
Perplexity looks for sources that match the words and meaning of your question. It removes clearly irrelevant or stale results and ranks what remains. Pro Search then brings information from several searches together in an answer.
Search its index using word matching and meaning
From candidates to ranked results
- Combine the two sets of results
- Filter clearly irrelevant or stale content
- Rank the remaining candidates through several stages
A simplified view of Perplexity’s own API description. It is not a trace of every answer in the consumer app.
Sources: Perplexity: Search API architecture
How does Perplexity narrow down the sources?
Perplexity searches by words and by meaning, then combines the results. Its technical explanation describes several stages that remove poor matches and rank the remaining sources.
It can score a whole page or a smaller part. When you review your content, read the section that answers the buyer’s question on its own. Check that it still makes sense.
Why can the sources change during a conversation?
Pro Search uses the selected focus to decide where to look. It can search articles, academic papers, forums and videos. The source choice therefore depends in part on the kind of search the user chooses.
Follow-up questions keep the context of the earlier conversation. When you compare answers, record the search focus and previous questions rather than treating the final question as the whole test.
Can Perplexity read your website?
Ask your website manager to check access for PerplexityBot. Perplexity recommends allowing it in the site’s crawler rules and allowing requests from its published network addresses. This bot collects information for search results, not AI training.
Perplexity-User has a different job. It may visit a page because someone asked a question. Perplexity says these user-requested visits generally ignore the crawler rules in robots.txt, so do not confuse them with routine crawling.
What should you change on your pages?
Make the useful information easy to follow. Perplexity says it aims to preserve a page’s structure when reading it and can select smaller parts of a document. Clear sections help a reader check the same information.
Adding more text is not a useful goal by itself. Perplexity’s technical explanation says irrelevant information can hurt model performance. Keep the detail that answers the question and remove the material that gets in the way.
Most traditional SEO best practices still apply, according to Jesse Dwyer of Perplexity in Search Engine Journal. His comment is guidance from the company, not a promise about your page.
But yes, it is currently accurate to say (most) traditional SEO best practices still apply.
How should you check a Perplexity answer?
Open the linked sources and check the information against them. Perplexity itself recommends this. Pro Search brings findings from multiple sources together, so read the source behind the claim that matters to your decision.