What You'll Learn
- The difference between confirmed, likely, and speculative ranking factors
- All verified content quality signals from official Google documentation
- How link signals including PageRank and anchor text influence rankings
- Which technical signals Google has officially confirmed as ranking factors
- How E-E-A-T is evaluated and why it is not a single direct metric
- How Google's AI ranking systems (RankBrain, BERT, MUM) work
- The most widely repeated ranking factor myths — debunked with sources
What Makes a Ranking Factor Confirmed
Google has confirmed it uses over 200 signals to rank pages, but it has not published a complete list. This creates a large space for speculation in the SEO industry. For this guide, a ranking factor is classified as "confirmed" only if it meets one of the following criteria: Google's official Search Central documentation states it directly, a senior Google employee (such as Gary Illyes, John Mueller, or Danny Sullivan) has explicitly confirmed it in a public forum, the factor appears in an official Google research paper or patent, or it was disclosed in the 2024 Google Search API document leak subsequently acknowledged as authentic.
Everything beyond these criteria — correlation studies from SEO tools, anecdotal observations, and interpretations of vague statements — is speculative, however widely repeated. This guide covers confirmed factors only.
The 2024 API documentation leak
In May 2024, a large volume of internal Google Search API documentation was leaked and subsequently verified as authentic by Google. While the documents describe internal systems and do not constitute public-facing guidelines, they provided significant corroboration for several previously suspected ranking signals. This guide notes where the leak provides additional confirmation.
| Category | Number of Confirmed Factors | Primary Source |
|---|---|---|
| Content quality | 8+ | Google Quality Rater Guidelines, Search Central |
| Link signals | 5+ | PageRank patent, Search Central, API leak |
| Technical / UX | 6 | Google announcements (HTTPS, CWV, Mobile-first) |
| E-E-A-T related | Multiple | Google Quality Rater Guidelines (public) |
| Query understanding | 5 systems | Google AI announcements |
Content Quality Signals
Content signals are the most significant category of ranking factors. Google's entire purpose is to return the most relevant, highest-quality content for a given query — content evaluation is therefore central to the ranking algorithm.
<title> element is a direct on-page relevance signal. Google may rewrite title tags in SERPs if it determines they do not accurately represent the page content, but the original title still influences how the page is understood.Link Signals
Links were the original differentiator of Google's algorithm. PageRank, the link-based authority scoring system described in the 1998 Brin and Page research paper, remains a core ranking signal. The fundamental principle — that links from authoritative, relevant pages pass more authority than links from low-quality pages — is confirmed and unchanged, though the implementation has become significantly more sophisticated.
John Mueller (Google Search Advocate) has confirmed that Google does not always pass negative signals from low-quality links to the linked page. In many cases, Google simply ignores links it considers unnatural or manipulative rather than applying a penalty. The disavow tool remains available for sites that believe they have received a manual penalty related to links.
Technical Ranking Signals
Google has officially announced a small number of technical signals as direct ranking factors. These are noteworthy because they represent rare cases where Google explicitly stated that a specific technical property influences ranking — something Google seldom does publicly.
E-E-A-T: Experience, Expertise, Authoritativeness, Trustworthiness
E-E-A-T is a framework from Google's Search Quality Rater Guidelines — a document used to train the human quality raters who evaluate search result quality. Google added the first "E" (Experience) in December 2022 to the existing E-A-T framework, recognising that first-hand personal experience with a topic is a quality signal distinct from formal expertise.
The Four Components
- Experience. Does the content creator have direct, first-hand experience with the topic? A product review written by someone who has used the product is higher quality than one written by someone who has not. A travel guide written by someone who has visited the destination carries more weight than one assembled from secondary sources.
- Expertise. Does the content creator have the knowledge and skill required to produce accurate, reliable content on this topic? For YMYL (Your Money or Your Life) topics — medical, legal, financial, safety — formal expertise is weighted heavily. For other topics, demonstrated knowledge and skill qualify as expertise.
- Authoritativeness. Is the content creator or the website a recognised authority on this topic? Authority is built through external recognition — citations, mentions from other authorities, reputation within a field — not self-declaration.
- Trustworthiness. Google's Quality Rater Guidelines state that trust is "the most important member of the E-E-A-T family." An untrustworthy website has low E-E-A-T regardless of apparent expertise. Trust signals include transparency (who is behind the site), accuracy and factual correctness, security (HTTPS), and legitimate business signals.
Google's quality raters use E-E-A-T as a qualitative evaluation framework, not a quantitative metric. There is no "E-E-A-T score" in Google's algorithm. What exists are machine learning models trained on the outputs of quality rater evaluations — so E-E-A-T influences rankings indirectly through signals that correlate with high quality assessments.
YMYL Pages Have Higher E-E-A-T Requirements
Google's guidelines identify "Your Money or Your Life" pages — those that can significantly impact a person's financial stability, health, safety, or wellbeing — as requiring the highest levels of E-E-A-T. These include medical information, legal advice, financial guidance, news on important topics, and safety information. For these categories, demonstrated formal expertise and strong trust signals are particularly important.
Google's AI Ranking Systems
Google has publicly confirmed five major AI systems that influence how queries are understood and how results are ranked. These are not separate ranking factors but rather systems that mediate how all ranking factors are applied.
| System | Year | Function | Scale |
|---|---|---|---|
| RankBrain | 2015 | Maps queries and pages to a concept space; handles never-before-seen queries | Confirmed as top 3 ranking signal at launch |
| Neural Matching | 2018 | Matches query concepts to page concepts even when exact words don't match | Applied to ~30% of queries at launch |
| BERT | 2019 | Bidirectional language model; understands word context within full sentences | Now applied to nearly all queries in English and many other languages |
| MUM | 2021 | Multimodal; can process text, images, video; understands across 75+ languages simultaneously | Applied selectively to complex searches |
| Gemini integration | 2024 | Powers AI Overviews in SERPs; influences understanding of complex, multi-part queries | AI Overviews rolled out globally 2024 |
These AI systems collectively mean that keyword-matching strategies that may have worked historically are far less effective. Google now understands the intent and meaning behind queries at a sophisticated level — optimising for search engines by stuffing keywords is both unnecessary and potentially counterproductive. Writing clear, comprehensive, accurate content for human readers is the only durable approach.
Common Ranking Factor Myths — Debunked
The SEO industry generates a significant volume of unverified and incorrect claims about ranking factors. Here are the most widely repeated myths, with official sources for correction.
Google has explicitly stated it does not use the meta keywords tag for web search rankings. Bing also confirmed it ignores it. This tag has been irrelevant since approximately 2009. Source: Google Webmaster Blog, 2009.
Domain Authority is a metric invented by Moz. It is a proprietary third-party metric that attempts to predict ranking potential — it is not used by Google, acknowledged by Google, or correlated with Google's actual internal authority calculations. Google's equivalent concept is PageRank, which operates at the page level.
Google has repeatedly confirmed that social signals — Facebook likes, Twitter/X shares, LinkedIn engagement — are not used as direct ranking factors. Google cannot reliably crawl and process the full social media graph. Social media can indirectly help SEO by increasing content visibility and earning links. Source: Multiple John Mueller statements, 2014–2024.
Whether Google uses user engagement signals (click-through rate, bounce rate, dwell time) as ranking factors is one of the most debated topics in SEO. The 2024 API leak included references to click-related metrics, but Google has officially denied using GSC CTR data as a ranking input. The relationship remains disputed and should not be treated as confirmed.
Google has never confirmed keyword density as a metric, and Gary Illyes explicitly stated in 2014 that keyword density is not a ranking signal. Modern Google understands synonyms, related concepts, and entity relationships — writing naturally for readers is more effective than attempting to hit keyword density targets.
Authentic Sources Used in This Guide
Official documentation, research papers, and verified disclosures only.
Public document defining E-E-A-T, YMYL, and quality assessment framework.
Brin & Page (1998). Describes PageRank, anchor text signals, and original ranking architecture.
Google's announcement of Core Web Vitals as ranking signals (2020).
Official overview of Google's ranking systems and approach.
Google's Helpful Content system announcement and integration into core algorithm.
Google's BERT announcement (2019), confirming scale and impact on query understanding.