SEO Ranking Factors: Confirmed, Leaked, Myths

SEO ranking factors are the signals a search engine uses to decide which pages to show for a query and in what order. Google says it looks at hundreds of them, but it confirms only some, the 2024 documentation leak shows many more that it stores without saying how they are weighted, and a long list of popular “factors” are myths Google has explicitly denied. This page sorts them into those three levels of evidence.

It is for marketers and owners who have seen “200 ranking factors” lists and want to know which items deserve budget. Every claim below links to a Google document, a named analysis of the leak, or a named study. Where the evidence is only that an attribute exists, we say so.

On this page

• What are SEO ranking factors?
• Confirmed ranking factors from Google
• What the 2024 leak shows Google stores
• Ranking factor myths
• What are the top 3 SEO ranking factors?
• Every factor, graded by evidence
• Do ranking factors apply to AI answers?

What are SEO ranking factors?

A ranking factor is any measurable property of a page, site, query or user that changes where a page appears. Google’s guide to how Search works says relevance is determined by hundreds of factors, which could include the user’s location, language and device. So some factors are about you and many are about the searcher.

The mistake most lists make is treating every claimed factor as equally real. The diagram below shows the three levels of evidence we use for the rest of this page.

Confirmed ranking factors from Google

These are factors Google’s own documentation says its systems use. The ranking systems guide is the primary source, and it describes systems rather than individual signals:

• Query understanding and relevance. BERT, RankBrain, neural matching and MUM match queries to pages by meaning, not only exact words. Passage ranking scores individual sections of a page.
• Links. Google calls PageRank one of its core ranking systems, used since launch. Backlinks remain a confirmed input.
• Freshness. Freshness systems show newer content where recency is expected, such as news or product releases. For evergreen queries, age matters less.
• Originality. Original content systems aim to rank original reporting ahead of pages that merely cite it.
• Reviews. The reviews system rewards reviews with insightful analysis and original research.
• Reliability. Reliable information systems surface authoritative pages and demote low-quality content, especially on sensitive topics.
• Page experience. Google’s page experience documentation says Core Web Vitals are used by its ranking systems, and that good scores don’t guarantee a top ranking. Details in Core Web Vitals explained.
• Demotions. Spam detection, deduplication, the exact match domain system, removal-based demotion, and site diversity, which generally limits a site to two listings in top results.

For local results, Google names three factors in its Business Profile help: relevance, distance and prominence.

What the 2024 leak shows Google stores

In May 2024, internal documentation for Google’s Content Warehouse API became public. Rand Fishkin’s SparkToro write-up describes more than 2,500 pages covering 14,014 attributes. Mike King’s iPullRank analysis counts 2,596 modules with 14,014 attributes as of March 2024.

The caveat is the most important part. Fishkin notes the documents don’t show the weight of any element in ranking, nor prove which elements are used. King writes that we don’t know how features are weighted or whether everything available is used. Google stores these attributes; the leak does not show how much any of them is weighted. The table below lists ten that matter for everyday SEO, with Google’s own field description where the documentation gives one.

What the leak changes in practice is modest but real:

• Titles are measured against queries. QualityNsrNsrData.titlematchScore is described as a site-level signal of how well titles match user queries. Clear, query-shaped titles are worth the effort. See SEO title length.
• Topical focus is recorded. QualityAuthorityTopicEmbeddingsVersionedItem.siteFocusScore and siteRadius measure how focused a site is and how far pages stray from it. Publishing far outside your subject may dilute more than it adds.
• Clicks are stored. NavBoost attributes such as goodClicks and lastLongestClicks exist. What they feed, and how strongly, is not in the documentation. Our NavBoost explainer covers what is known.
• New hosts are tracked. PerDocData.hostAge is described as used to sandbox fresh spam at serving time.

For the full background, see the Google leak explained.

Ranking factor myths

Some popular factors are not “unproven”. Google has said they do nothing. Its SEO Starter Guide and related documentation address them directly:

• Meta keywords. Google’s list of supported meta tags says the keywords meta tag is not used and has no effect on indexing and ranking.
• Word count. There’s no magical word count target, minimum or maximum. Length only matters as far as it covers the question.
• Heading order. Google says it doesn’t matter if headings are used out of order. Headings still help readers and passage matching.
• Keywords in the domain. They have hardly any effect beyond appearing in breadcrumbs. The exact match domain system exists to stop them earning excess credit.
• E-E-A-T as a score. Google’s guidance on helpful content says E-E-A-T itself isn’t a specific ranking factor, and that rater data is not used directly in ranking algorithms. Our E-E-A-T guide covers what it is instead.
• Paying Google. Google doesn’t accept payment to crawl more often or rank higher.

What are the top 3 SEO ranking factors?

Nobody outside Google can rank factors by weight, and anyone who gives you an exact top three is guessing. What the evidence supports is a working order of priority:

1. Relevance to the query and its intent. Most of Google’s named systems exist to judge it.
2. Content quality and originality. Several systems (original content, reviews, reliable information) reward it; spam and deduplication systems punish its absence.
3. Reputation from other sites. PageRank is a confirmed core system, and the leak shows site-level authority attributes are stored.

Technical accessibility is not on the list because it is a gate, not a factor. A page that fails Google’s technical requirements is not ranked badly; it is not in the running.

How to check a ranking factor claim yourself

New “factors” appear every month, usually in a vendor’s blog post or a conference slide. Four questions sort most of them in a couple of minutes:

1. Is there a Google document that names it? Search Central documentation, the ranking systems guide, or the spam policies. If yes, it is confirmed, though the document rarely says how much it counts.
2. Is it an attribute in the leaked documentation? If the claim cites the leak, check that the exact attribute name exists, and remember that existing is all the leak proves. A post that says “the leak confirms Google ranks by X” has overstated it.
3. Is the evidence a correlation study? Correlation studies are useful for spotting patterns, but pages that rank well tend to share many traits at once. Long pages correlate with rankings partly because well-resourced sites write long pages.
4. Has Google denied it? Several popular factors, from meta keywords to word count, have been addressed directly. A denial is not proof the thing is irrelevant in every case, but it is strong evidence against paying for it.

One more habit helps: test changes on a handful of comparable pages and compare them with untouched pages over the same weeks. Google’s own advice is to wait a few weeks before judging a change, and seasonal swings or an algorithm update can easily be mistaken for the effect of your edit. Search Console’s page-level query data is the most honest record you have of whether a change moved anything.

Every factor, graded by evidence

Do ranking factors apply to AI answers?

The gates do; the weighting shifts. Google’s AI features documentation says a page must be indexed and eligible for a snippet to be a supporting link in AI Overviews or AI Mode. After that gate, brand mentions appear to matter more than raw link counts. In Ahrefs’ study of 75,000 brands, branded web mentions correlated with AI Overview visibility far more strongly than backlinks, as the chart shows.

Correlation is not causation: well-known brands earn mentions and AI visibility for the same underlying reason. Still, the pattern fits how answers are assembled, from sources that describe a brand consistently. We cover the mechanics in how AI engines choose citations and the tactics in digital PR for AI search.

Using SEO ranking factors to prioritize your site

Treat the three bands as a budget rule. Spend first on confirmed factors, where Google has told you what it rewards. Use stored attributes to break ties and to shape habits, such as query-shaped titles and a focused topic, without pretending you know their weight. Spend nothing on myths.

If you want your own site graded this way, page by page, with AI answers checked alongside rankings, our AI visibility audit uses the same evidence levels and shows which fixes fall in which band.