Lists of « 200 Google ranking factors » have circulated for more than a decade. They blend three different things: what Google documents in writing, what studies observe without establishing causation, and what SEO folklore copies from one article to the next. For anyone signing off a budget, that confusion is expensive. You polish details that change nothing while the real problems stay where they are.
This article sorts the list using primary sources only: the Google Search Central documentation, the public How Search Works pages, the quality rater guidelines and, on one point, evidence from the United States antitrust case. Every item gets a label: confirmed factor, observed correlation, or myth. The aim is not another checklist, but a way of judging the next list someone shows you.
What Google actually says about ranking
Google describes three stages: crawling, indexing, then serving results. On the last, the documentation stays deliberately general. Relevance is described as being determined by « hundreds of factors », including the searcher's language, device and location. Google publishes neither the list nor the weighting. Any numbered list you come across is an outside reconstruction, not a Google document. The public How Search Works page names five families and stops there: the meaning of the query, the relevance of the content, its quality, the usability of the page, and the searcher's context.
Ranking is entirely algorithmic, and Google states that it does not accept payment to crawl a site more often or to rank it higher. For 2023, Google reported 719,326 quality tests with external raters, 16,871 live traffic experiments and 4,781 changes shipped (the « Rigorous testing » page, consulted in September 2026). The system moves constantly: the most recent major update on record is the May 2026 core update, launched on 21 May 2026 and rolled out over roughly twelve days.
The factors the documentation confirms
Technical accessibility is a condition, not a bonus
A page has to be crawlable, renderable (JavaScript included) and indexable. Google lists the usual causes of failure: server errors, blocking robots.txt rules, noindex meta robots tags, and content judged to be low quality. This layer does not push a page up; its absence stops it existing at all. It is the first thing to check, and the subject of a technical SEO audit.
Useful, original content
This is the one point on which Google is categorical: creating content people find compelling and useful will likely influence a site's presence in results more than any other suggestion in its SEO starter guide. The attributes it names are concrete: readable, well-organised text, original content, content kept up to date, and expert or first-hand sources where the subject calls for them.
Inbound links
Google confirms PageRank is still part of its core ranking systems, while noting in the same document that Search is much more than links. Links count, but neither volume nor a vendor score predicts a position. A backlink review should therefore be read as a trend indicator, not a points tally.
Core Web Vitals
These are the only page experience metrics Google states in writing that it uses in ranking. The published thresholds: LCP under 2.5 seconds, INP under 200 milliseconds, CLS under 0.1. Two caveats are usually left out: good scores do not guarantee a position, and beyond Core Web Vitals, other aspects of page experience do not directly help you rank.
HTTPS and mobile
HTTPS has been a ranking signal since August 2014, presented from the outset as very lightweight and affecting fewer than 1% of global queries. The real stake sits elsewhere: a site served over HTTP triggers a browser warning. On mobile, since 5 July 2024 Google crawls all sites with its smartphone crawler. A site unusable on a phone is not « penalised »; it is indexed from its mobile version, with whatever that contains.
| Item | Status | Source |
|---|---|---|
| Useful, original content | Confirmed | SEO starter guide |
| Links / PageRank | Confirmed, relative | Ranking systems guide |
| Core Web Vitals | Confirmed | Page experience doc |
| HTTPS | Confirmed, very weak | Google blog, 2014 |
| Mobile compatibility | Indexing condition | Mobile-first indexing doc |
| E-E-A-T | Correlation | SEO starter guide |
| Domain authority scores (DA, DR) | Third-party metric | Absent from Google documentation |
| Click data | Confirmed outside documentation | US v. Google ruling, 2024 |
| Meta keywords, keyword density, keywords in the domain | Myth | SEO starter guide |
Correlations: worth watching, not worth « optimising »
E-E-A-T. The SEO starter guide is unambiguous: asked whether E-E-A-T is a ranking factor, Google answers no. The helpful content documentation sets out the reasoning: the systems combine signals capable of identifying content that demonstrates experience, expertise, authoritativeness and trustworthiness, with trust as the most important element. The weight given to it rises on YMYL subjects: health, finance, safety. E-E-A-T is a way of reading a page, formalised in the Search Quality Rater Guidelines (version of 11 September 2025, 182 pages). It is not a setting; rater scores do not feed directly into the algorithms.
Domain authority. Scores such as DA, DR or Trust Flow are tool vendors' metrics, calculated on their own link indexes. Useful for following a trend, they appear nowhere in Google's documentation and have no predictive value for an individual page.
Behavioural signals. Google does not document bounce rate or time on page as factors. But the ruling of 5 August 2024 in United States v. Google describes a signal called Navboost, which associates queries with documents by memorising user click data, trained on thirteen months of data (eighteen before 2017). The nuance matters: this is data aggregated at query level, not a dial you turn. Working on titles and descriptions is still worth doing, but do it for the traffic.
The myths worth dropping
Google's starter guide keeps a section on things it believes you should not focus on. It settles several arguments at once:
- the
meta keywordstag is not used by Google; - keyword stuffing, including stuffing with variants, falls under the spam policies;
- keywords in the domain name or the URL have almost no effect beyond the breadcrumb;
- the domain extension (.com, .co.uk, .guru) counts only for geographic targeting, and weakly;
- there is no ideal content length, minimum or maximum;
- subdomain or subdirectory: pick whichever is simpler to manage;
- there is no internal « duplicate content penalty »; it is inefficient rather than punished (copying someone else is another matter);
- the number and order of H1 to H6 headings change nothing for Google, though a sensible semantic order helps screen reader users.
Add two undocumented assumptions to the pile: publishing frequency on its own, and domain age.
What has changed recently
Several reference points still quoted in meetings are out of date. The helpful content system, launched in 2022, was absorbed into the core ranking systems in March 2024 and no longer exists as a separate update. The mobile-friendly test and the Search Console mobile usability report were retired on 1 December 2023; the analysis now runs through Lighthouse and the Core Web Vitals report, fed by Chrome UX Report field data.
On the newer surfaces, Google states that generative experiences (AI Overviews, AI Mode) carry no additional requirements and no particular optimisation: the same good practices apply. And on 28 August 2026, Google announced a differentiated application of its site reputation policy in the European Economic Area. Since 30 August, a manual action taken on those grounds no longer affects results shown to searchers in the EEA.
Where to start, in practice
One French data point illustrates the pattern. The 2025 France Num barometer (DGE, Crédoc survey of 11,021 businesses questioned between 26 March and 18 April 2025, self-reported answers) found that 65% of French micro-businesses and small firms have a website presenting their activity, rising to 81% of SMEs. Those figures describe France and no other market. What travels is the behaviour behind them: many of those sites were published once, then left alone. The order of priority is then almost always the same.
- Check indexing. A page missing from the index cannot rank. A few minutes in Search Console settles it.
- Deal with the content. One page per search intent, written to be useful, updated when it ages.
- Measure Core Web Vitals on field data, not on an isolated lab test, and fix whatever sits above the thresholds, which is the work of a site speed project.
- Look at links last. A link profile does not compensate for content that answers nothing.
One last decision rule, useful when choosing an SEO provider: if someone shows you a weighted list of ranking factors, or promises a position, ask for the source. It does not exist. Google itself warns that some changes register within hours and others take months.
Common questions
How many ranking factors does Google use?
Google says the relevance of a result is determined by hundreds of factors, and never publishes the list or the weighting. The lists of 200 factors in circulation are third-party compilations, not Google documents. Only broad families are described publicly: the meaning of the query, relevance, quality, usability and context.
Is E-E-A-T a ranking factor?
No. Google's SEO starter guide answers explicitly that E-E-A-T is not a ranking factor. It is a concept from the quality rater guidelines describing what trustworthy content should demonstrate. The algorithms use signals meant to identify that content, weighted more heavily on health, finance and safety subjects.
Does bounce rate affect rankings?
Google documents no ranking factor based on bounce rate or time on page. The 2024 ruling in United States v. Google does describe a system, Navboost, that memorises aggregated click data over thirteen months. That is query-level data, not a setting a site owner can adjust.
Should you aim for 100/100 in PageSpeed Insights?
No. Google notes that good scores in these reports do not guarantee a good position, and that chasing a perfect score for SEO reasons may not be the best use of your time. The useful reference points are the Core Web Vitals thresholds on field data: LCP under 2.5 seconds, INP under 200 milliseconds, CLS under 0.1.
How many words does a page need to rank?
There is no ideal length. Google states that content length on its own has no bearing on ranking: no magic minimum, no maximum.
How long before changes show up in rankings?
Google says some changes register within hours and others take months, and recommends waiting a few weeks before judging an effect. Not every change produces a visible result. Major algorithm updates usually roll out over one to three weeks, so readings taken mid-rollout are hard to interpret.