1DOT
AI Visibility8 min read

How Perplexity chooses which sources to cite

Author
Mo Walji
Published
12 APR 2026
Category
AI Visibility

Perplexity built its own 200-billion-URL index and runs a crawler most businesses have never heard of. Your Google rankings do not transfer here. Neither do your assumptions about what gets cited.

Why does Perplexity ignore some businesses?

A company can rank on page one of Google, hold a strong backlink profile, and have a decent Bing presence, and Perplexity will still have no idea it exists. This happens constantly.

80% of URLs cited by AI assistants do not appear in Google's top 100 results (Ahrefs, 2025). Only 11% of cited domains show up across multiple AI platforms (Profound, 680M citations). Perplexity built its own index, runs its own crawler, and pulls supplementary data from Google and Bing APIs. The balance shifts per query. Nobody outside Perplexity knows the formula.

The reverse happens too. Smaller sites with fresh, structured content get cited regularly while established competitors stay invisible. The patterns are there. The published playbook is not.

That is the honest starting point. Perplexity does not explain how it selects sources. What is known comes from reverse engineering, large-scale citation studies, and continuous testing. Not a spec sheet.

How does Perplexity decide which sources to cite?

Nobody has the complete answer. Perplexity does not publish its ranking criteria. What is known comes from reverse engineering and large-scale citation studies, not a spec sheet from the company.

The signals AI platforms use to recommend businesses overlap, but Perplexity weights them differently from every other platform. Five signals stand out.

Content freshness

Half of all Perplexity citations come from content published in the current year, and 80% from the last two years (Seer Interactive, 5,000+ URLs analysed, 2025). On ChatGPT, older evergreen content still performs. On Perplexity, recency is the price of entry.

Around half of all Perplexity citations come from content published in the current year, and 80% from the last two years.

Source: Seer Interactive, 2025

Earned media

82% of AI citations across platforms come from earned media (Muck Rack, 1M+ links, 2025). Being written about in third-party publications makes a business structurally more citable. Perplexity's source selection appears to reward this even more than other platforms.

Across platforms, 82% of AI citations trace back to earned media rather than a brand's own website.

Source: Muck Rack, 2025

Reddit presence

46.7% of Perplexity's top 10 most-cited domains are Reddit (Profound, 680M citations, 2025). Community-generated, experiential content wins. Unlike ChatGPT's licensed Reddit partnership, Perplexity accesses Reddit's public web directly.

Reddit accounts for 46.7% of Perplexity's top 10 most-cited domains.

Source: Profound, 680M citations, 2025

Extractable structure

Short, declarative passages that function as standalone answers. Numbered steps. Comparison tables. The patterns across audits are consistent: structure for extractability, not just readability. A Q&A format is preferred.

Domain authority

80% of URLs cited by AI assistants do not appear in Google's top 100 results (Ahrefs, 2025). Strong backlink profiles, publishing consistency, and recognised author credentials still matter. The SEO fundamentals do not disappear. They carry different weight.

Citation volume, publisher programmes, and technical requirements

Perplexity cites more sources per response than any other platform. 21.87 on average, versus ChatGPT's 7.92 (Qwairy, 118,101 answers, Q3 2025). More citations per answer means more chances to be the source. Source cards appear at the top of every response. Numbered references sit inline. This platform shows its working.

Perplexity averages 21.87 citations per response, versus ChatGPT's 7.92.

Source: Qwairy, 118,101 answers, Q3 2025

The Comet Plus publisher programme shares revenue with cited publishers. Partners include the LA Times, The Independent, CNN, and the Washington Post. Nobody has confirmed preferential citation treatment. But the financial incentive exists.

One technical requirement is non-negotiable: PerplexityBot does not render JavaScript. If content loads through JS frameworks, expandable menus, or hidden tabs, Perplexity cannot see it. Server-side rendering or invisibility. Those are the options.

What matters most on Perplexity?

Perplexity built its own index. It does not borrow Google's view of the web and repackage it. That single fact changes everything about how visibility works on this platform. Three areas matter more on Perplexity than anywhere else.

  • 01
    Freshness cadencePerplexity weights recency at roughly 40%, the highest of any AI platform. Around half of all citations come from content published in the current year. Content older than two years is structurally disadvantaged. This is not a ranking signal. It is a publishing strategy problem.
  • 02
    Earned media and Reddit presence82% of AI citations come from earned media (Muck Rack, 2025). On Perplexity specifically, Reddit accounts for 46.7% of top cited domains (Profound, 680M citations). Your own website is necessary but insufficient. Third-party mentions on sources Perplexity actually reads are what drive citations.
  • 03
    Technical accessPerplexityBot does not render JavaScript. If your content requires client-side rendering, Perplexity cannot see it. Server-side rendering and clean HTML structure are non-negotiable. Allow the crawler in robots.txt. Keep canonical tags consistent. Fast page speed. No exceptions. Surfacing blockers like these is the job of a technical AI visibility review.

The difference from ChatGPT optimisation is structural. ChatGPT depends on Bing's index and rewards referring domain count above all else. Perplexity runs its own index and rewards freshness. Only 11% of cited domains appear across both platforms (Profound, 2025). Visibility on one does not guarantee visibility on the other.

Anyone who claims to have cracked Perplexity's algorithm is selling confidence they do not have. The platforms are opaque. The architecture shifts without notice. This work requires continuous research and adaptation. Not a checklist.

For how this fits across all six AI platforms, see the AI search visibility services. For the methodology behind structuring content for AI citation, see generative engine optimisation (GEO).

Why Perplexity visibility is worth the work

Google organic converts at 1.76%. Perplexity converts at 10.5% (Semrush, 2025). That is not a rounding error. It is a 4.4x multiplier.

Perplexity traffic converts at 10.5% against Google organic's 1.76%, a 4.4x multiplier.

Source: Semrush, 2025

Perplexity users arrive with intent already formed. They have read the synthesised answer. They are clicking through to confirm, compare, or buy. The question is whether they land on your site or a competitor's.

A practical first step is to run your real buyer queries through Perplexity, record every response, and score your presence against your top competitors. That answers three questions. Can PerplexityBot reach your pages? Is your content structured for citation extraction? Where do you stand against competitors for the queries your buyers actually ask? Where the answer exposes pages and structure that AI cannot parse, a Visibility Rebuild fixes the page roles, schema, and entity structure at the root. If you want a full assessment across all six AI platforms, start a free AI visibility audit.

About the author
Mo Walji

Mo Walji founded 1DOT in 2015. The company now helps B2B companies become visible to AI search platforms across ChatGPT, Perplexity, Gemini, and Google AI Overviews. Mo works directly with construction, manufacturing, professional services, and education companies in the UK and US.

Read more about Mo →

Frequently asked questions

Perplexity does not publish its ranking criteria. What we know comes from independent research and reverse engineering. The strongest signals identified so far are content freshness, earned media placement, domain authority, answer-first content structure, and Reddit presence. Perplexity has its own index alongside Google and Bing API access, so the source mix for any given query is not disclosed. Traditional search rankings are not a reliable proxy for Perplexity visibility.
No. Perplexity has no paid advertising product. The Comet Plus publisher programme shares revenue with publishers whose content gets cited, but participation does not guarantee citation. You cannot buy placement. You earn it through content quality, freshness, and authority.
Technical fixes like crawler access and server-side rendering can take effect within days. Content restructuring typically shows citation changes within four to eight weeks. Authority signals like earned media placement take months. Freshness improvements may show results faster than on other platforms because current research suggests Perplexity weights recency more heavily.
Not reliably. Perplexity has its own index alongside Google and Bing API access. Only 11% of cited domains appear across multiple AI platforms (Profound, 2025). Strong SEO foundations help because domain authority, content quality, and technical health influence citation across all platforms. But high Google rankings alone do not guarantee Perplexity citations.
PerplexityBot is Perplexity's independent web crawler. It indexes content separately from Google and Bing. It does not render JavaScript, so you need server-side rendering. Allowing PerplexityBot in your robots.txt is the first step to becoming citable.
Related resources
More insights

Related reading

See where your company stands in AI search.

Start with a free audit. We test your visibility across the major AI platforms and walk you through the results on a live call.

Start with a free audit