Ranking Factors: Which Are Confirmed, Which Are Inferred

Most SEO advice collapses three different things into one list: statements Google has made, patterns found in third-party datasets, and claims repeated because they sound plausible. That confusion makes it difficult to decide whether to improve a page, build links, change a template, or ignore an expensive recommendation.

This guide separates SEO ranking factors by evidence level. “Confirmed” means Google has publicly described a system, signal, or best practice as relevant to Search—not that the factor guarantees a position. “Inferred” means a correlation study found an association, not that the factor caused higher rankings. “Folklore” means the claim lacks reliable public evidence or is commonly overstated.

How to read evidence about SEO ranking factors

Search rankings are produced by multiple systems, queries, locations, devices, languages, and user contexts. A page can rank well because it satisfies search intent, is discoverable, is supported by credible links, and is technically accessible. There is rarely one switch that explains the result.

Evidence also has different strengths:

• Confirmed by Google: Google documentation, statements from named Google representatives, or observable requirements such as crawling and indexing guidance. These sources establish that a factor or system exists, but usually do not reveal its weight.
• Inferred from correlation studies: Large-scale analyses comparing ranking positions with page characteristics. These are useful for generating hypotheses, but they cannot establish causation.
• Folklore: Claims repeated in forums, sales pitches, or simplified checklists without a dependable source, or claims that turn a limited observation into a universal rule.

A correlation is not causal evidence. If high-ranking pages tend to have more referring domains, that does not prove that buying more links will produce the same result. Those pages may also have stronger brands, better content, older businesses, more publicity, and better distribution. Treat studies as diagnostic clues, not as Google’s published recipe.

Confirmed: what Google has actually supported

Factor or claim	Evidence level	Source of the claim
Relevance between the query and the page’s content	Confirmed	Google Search Central documentation on how Search works and ranking systems
Links help Google discover pages and can be used as a ranking signal	Confirmed	Google Search Central documentation and Google’s public explanations of PageRank and link analysis
HTTPS is used as a lightweight ranking signal	Confirmed	Google Search Central announcement about HTTPS as a ranking signal
Core Web Vitals are used in page experience evaluation	Confirmed	Google Search Central documentation on page experience and Core Web Vitals
Freshness matters for queries where newer information is expected	Confirmed, query-dependent	Google Search Central documentation on freshness systems
Longer content ranks better	Inferred at best, not a Google rule	Correlation studies such as Backlinko’s analysis of 11.8 million Google search results
A specific word-count threshold improves rankings	Folklore	Common SEO advice; no general Google threshold has been published

Relevance and search intent

Google has consistently described relevance as central to Search. Its systems analyze the words in a query, the meaning behind them, and the content of candidate pages. This does not mean repeating the exact keyword a fixed number of times. It means answering the task implied by the query.

For “enterprise SEO migration checklist,” a 600-word opinion article is less aligned than a practical checklist covering redirects, canonicals, testing, logs, templates, and monitoring. For “best accounting software for freelancers,” a page may need comparisons, pricing context, limitations, and suitability—not merely a definition of accounting software.

Links and discoverability

Google has publicly described links as useful for discovering pages and understanding relationships between documents. Google’s original PageRank research also established the importance of link analysis. That confirms links matter; it does not confirm that every link passes meaningful ranking value.

A relevant editorial link from a trusted publication is materially different from hundreds of automatically generated directory links. Google’s spam policies prohibit link schemes intended to manipulate rankings. Do not treat “more referring domains” as a license to buy packages priced at $99 to $500 per month. Investigate relevance, editorial control, placement, anchor text, and whether the link would exist without an SEO transaction.

Technical access, HTTPS, and page experience

A page must generally be crawlable, indexable, and eligible to appear. Robots.txt, noindex directives, canonical signals, server errors, and JavaScript rendering can prevent a page from performing regardless of how good its copy is. These are not glamorous ranking tactics; they are prerequisites for receiving a fair evaluation.

Google announced HTTPS as a lightweight ranking signal, while also emphasizing that it affects a small fraction of queries compared with stronger relevance signals. HTTPS should therefore be standard security hygiene, not a promise of a ranking lift.

Google’s Core Web Vitals thresholds are specific: a good Largest Contentful Paint is 2.5 seconds or less, Interaction to Next Paint is 200 milliseconds or less, and Cumulative Layout Shift is 0.1 or less, measured at the 75th percentile. These metrics are part of Google’s page experience guidance. Passing them is worthwhile, but it does not compensate for irrelevant content or weak authority.

Freshness, originality, and people-first content

Freshness is confirmed but conditional. A query about election results, software pricing, or current regulations benefits from current information. A query about a stable mathematical definition may not. Changing the publication date without materially reviewing the content is not the same as making it fresher.

Google also publishes guidance about creating helpful, reliable, people-first content. That guidance supports quality decisions, but it should not be reduced to a mechanical “helpful content score” that an editor can increase by adding 300 words. Original reporting, first-hand experience, clear sourcing, and useful interpretation are stronger signals of quality than cosmetic expansion.

Inferred: useful patterns from correlation studies

Correlation studies can help prioritize investigation. They are particularly useful when a pattern appears across many queries and agrees with sound technical or editorial reasoning. They still cannot tell you what caused a page to rank.

Referring domains and link authority

Backlinko’s analysis of 11.8 million Google search results reported that pages in higher positions tended to have more referring domains. That is a meaningful association and supports examining link prominence when competing against established pages.

It does not prove that acquiring an identical number of domains will move a page from position 20 to position 3. Links may be a consequence of ranking rather than the sole cause: visible pages attract citations, mentions, journalists, and partnerships. Use the finding to assess whether your page is credible enough to compete, not to set an arbitrary monthly link quota.

Content length and topical coverage

Large studies often find that top-ranking pages are longer than lower-ranking pages. The Backlinko study mentioned above reported an average first-page result of 1,447 words. That number is descriptive, not a target issued by Google. A page can rank with 300 precise words, while a 3,000-word page can fail because it is repetitive or misses the user’s decision.

Longer pages may rank because complex topics require more complete answers. They may also attract more links or cover more related concepts. Those explanations are competing hypotheses, not proof that word count itself caused the ranking.

Brand mentions, authority, and traffic

Studies frequently show that strong brands, high-traffic sites, or pages with many mentions perform well. The association is commercially useful: recognizable companies often have better distribution, stronger links, clearer expertise, and more returning users.

However, “brand searches are a ranking factor” is an overstatement. A brand’s visibility can be both a cause and an effect of good rankings. Traffic can also reflect rankings rather than produce them. Treat brand demand as an asset that supports SEO, not as a metric to manufacture through artificial searches.

Structured data and rich results

Structured data helps search engines interpret eligible content and can make a page eligible for certain rich results. Correlation data showing that pages with rich results receive more clicks does not prove that the markup itself improves the blue-link ranking. The enhanced appearance may attract attention, or the pages using markup may already be better maintained.

Implement structured data only when it accurately describes visible page content and meets Google’s guidelines. Do not add fake reviews, hidden details, or unsupported schema in pursuit of a display feature.

Folklore: claims to remove from your strategy

“There is an ideal keyword density”

There is no universal percentage that makes a page rank. Repeating “SEO ranking factors” 30 times in a 1,000-word article can reduce clarity and create a spam signal for readers even if it satisfies a spreadsheet. Use the primary term where it is natural, then cover the vocabulary and questions genuinely required by the subject.

“More words always win”

This popular advice is wrong. A longer page often loses when a shorter competitor answers the query faster and more precisely. Do not turn a pricing page into a 2,500-word essay because a tool recommends a target. Add detail only where it resolves uncertainty, supports a decision, or demonstrates expertise.

“Domain age is a ranking factor”

An old domain may have accumulated links, recognition, useful content, and technical stability. That does not establish that age itself earns rankings. Buying an expired domain because it is 12 years old is not a substitute for relevance and a clean history.

“Domain Authority is Google’s score”

Domain Authority is a proprietary third-party metric, not a Google ranking factor. It can be useful for comparing link profiles within one provider’s dataset, but a score of 60 does not represent a Google threshold. Do not reject a relevant publication solely because its third-party score is lower than a prospecting target.

“Click-through rate is a direct ranking lever”

Search results naturally produce different click patterns by query, position, device, brand, and intent. Google has discussed user interactions in the context of evaluating and improving search systems, but that does not validate the simplistic claim that increasing organic CTR by five percentage points will automatically raise a page.

Improve titles and snippets for accurate persuasion, not manipulation. Never use misleading promises, fake urgency, or clickbait that causes users to return immediately because the page did not deliver.

A practical evidence-led workflow

1. Check eligibility first. Confirm that important URLs return a 200 status, are not blocked, are indexable, have a sensible canonical, and can be rendered. Review coverage in Google Search Console and server logs where available.
2. Match the page to the query. Identify whether the searcher wants a definition, comparison, transaction, local provider, template, or current update. Compare the task your page completes with the pages already earning visibility.
3. Use confirmed factors as guardrails. Secure the site, improve Core Web Vitals where they are poor, make important content accessible, and maintain accurate freshness when the subject changes.
4. Use correlations as hypotheses. If competitors have more relevant referring domains, investigate how they earned them. If they cover important subtopics you omit, improve completeness. Do not copy a number without testing the underlying explanation.
5. Reject unsupported prescriptions. Ask for the source behind any promised ranking threshold, guaranteed traffic percentage, or required word count. If the answer is a tool score or an unnamed “Google insider,” classify it as folklore until stronger evidence appears.
6. Measure over an appropriate timeframe. Technical corrections can appear after recrawling, sometimes within days, while content and authority changes may require several weeks or months. Compare similar queries and pages over at least 8 to 12 weeks where seasonality and demand permit; do not claim causation from a seven-day movement.

What this means for budgets and decisions

Confirmed factors are the foundation, not a guarantee. Fixing a noindex directive can unlock existing value; it cannot make an irrelevant page authoritative. Improving a 4.5-second mobile experience toward the 2.5-second LCP threshold may improve usability and page experience, but it should not displace work on product accuracy or editorial quality.

Inferred factors are best used for competitive diagnosis. A link gap may justify digital PR, partnerships, or genuinely useful research. A coverage gap may justify a new guide, comparison, calculator, or original dataset. Folklore should not receive budget merely because it is easy to report as a percentage or score.

The disciplined question is not “Which ranking factor are we missing?” It is “What evidence supports this proposed change, what alternative explanation exists, and what outcome would make the investment worthwhile?” That approach produces a more defensible SEO program than any universal checklist of supposed ranking secrets.

Related reading

• Technical SEO — scope, method and what we measure.
• Content strategy
• Agentic SEO: Separating the Capability From the Marketing
• Choosing an SEO Course: Criteria, Not a Ranked List