What is a ranking factor, and what has Google actually confirmed?
A ranking factor is a signal that Google’s ranking systems use to determine which pages appear first for a given search. Google has never published a list of weights. Google has published two documents. How Search Works describes five general types of signals: meaning, relevance, quality, usability, and context. A Guide to Google Search Ranking Systems lists the systems that process these signals, from PageRank to freshness and spam detection systems. Everything else you read about ranking factors is derived from these two documents, based on Google’s statements or the industry’s own observations.
At InboundCPH, we distinguish between three sources of knowledge about ranking factors: Google’s own documentation, statements from Google employees, and legal cases such as the U.S. antitrust case in 2023, as well as industry observations in the form of correlation studies, leaked documents, and our own measurements. All three may be correct. However, an observation is not a confirmation, and this guide specifies which category each factor belongs to.
The figure “over 200 ranking factors” comes from Google’s statements in the 2000s and is not used by Google today. Google describes its ranking as a system, not a checklist, and a system like the freshness system weights the same factor differently depending on the search. A news story about a corporate merger is rewarded for being new; a definition of a technical term is not.
Which 10 ranking factors will matter most in 2026?
The 10 ranking factors that will matter most in 2026 are: relevance to search intent, helpful and original content, links from other sites, E-E-A-T signals, freshness, Core Web Vitals, mobile-friendliness and HTTPS, crawlability and indexing, absence of spam, and user behavior. Google has confirmed the first nine in documentation or statements. The tenth has only been partially confirmed. Below, each factor is listed with its status and specific actions you can take.
1. Relevance to the search intent
Status: Confirmed. Google states in "How Search Works" that the most basic relevance signal is that the content contains the same words as the search query, especially in headings and body text. Google's language models also understand synonyms and the intent behind the words. This means that the page must answer what the user actually wants to know, not just contain the search term. Here's what you do: Write one page per question, have the first paragraph answer it directly, and tailor the headings to the search terms for which the page is already receiving views in Search Console.
2. Helpful, original content
Status: Confirmed. Google’s “helpful content” system was launched in August 2022 and has been part of the core ranking systems since March 2024, according to Google’s guide to ranking systems. Google describes the requirement as content that is created for people and contributes something that could not simply be generated from what already exists. Here's what you do: Add your own figures, examples, and opinions to each important page. Remove or merge pages that merely repeat what ten other pages say.
3. Links from other websites
Status: Confirmed. According to Google, PageRank is still part of the core ranking systems, and *How Search Works* cites links and mentions from prominent websites as a quality signal. Links from pages that are themselves relevant to the topic carry more weight than many links from unrelated pages. Here's what you do: Build links the way a journalist would: because you have data, a point of view, or a tool worth referencing. The method is described in our guide to link building.
4. E-E-A-T signals
Status: Confirmed as a direction, not as a score. Google states that its ranking systems look for signals that indicate experience, expertise, authority, and trustworthiness, but that E-E-A-T is not a single factor. The framework comes from Google’s guidelines for quality raters and gained its additional “E” for experience on December 15, 2022. Here's what you do: A named author with a background description, a separate page with the address and CVR number, sources for all figures, and a visible revision date.
5. Freshness
Status: Confirmed, but subject to the search. Google’s freshness systems display more recent content in search results where it is expected, such as for news and current events. In a search like “ranking factors 2026,” freshness is part of the search intent itself. Here's what you do: Update the pages where the facts are out of date, and keep the date in the text and in dateModified Follow the actual revision. Never change the date without changing the content.
6. Core Web Vitals
Status: confirmed, with limited weight. Google states in its documentation that Page Experience, that Core Web Vitals are used by ranking systems, but that good metrics do not guarantee top rankings. Google adds that chasing a perfect score for the sake of SEO is rarely a good use of time. The thresholds are LCP under 2.5 seconds, INP under 200 milliseconds, and CLS under 0.1. Core Web Vitals became a ranking factor in June 2021, and INP replaced FID in March 2024. Guides that still say “will become a ranking factor” are five years behind the times. Here's what you do: Fix the pages marked in red in Search Console's Core Web Vitals report, and stop there. You can find the details in our guide to Core Web Vitals.
7. Mobile-Friendliness and HTTPS
Status: Confirmed. In "How Search Works," Google mentions mobile-friendly content that loads quickly as a usability signal, and HTTPS has been a minor ranking signal since August 2014. Google also indexes the mobile version of your site first. Here's what you do: Verify that the mobile version has the same content as the desktop version and that no pages can be accessed via HTTP without a redirect.
8. Crawlability and Indexing
Status: Confirmed prerequisite. A page that Google cannot crawl or index cannot rank, no matter how good it is. The same applies to Google’s AI features. Google states in its AI guide that a page must be indexed and eligible to display a snippet in order to be considered for AI Overviews and AI Mode. Here's what you do: Check the robots.txt file, noindex tags, and canonical tags on the 20 most important pages, and check in Search Console to see if they've actually been indexed.
9. No spam
Status: Confirmed. Google's spam policies describes what triggers demotion or removal: hidden text, purchased links, scaled-up mass content, and misuse of a site’s reputation for third-party content. Google states in its 2026 AI guide that this also applies to pages that mass-produce variations to target fan-out searches. Here's what you do: Review old link agreements and outsourced subfolders before you build anything new.
10. User Behavior and Brand Signals
Status: partially confirmed. During the U.S. antitrust case against Google in October 2023, Google’s vice president of search, Pandu Nayak, confirmed that the Navboost system uses click data from search results. Internal API documents leaked in May 2024 point in the same direction. Google has not documented how the signals are weighted. Here's what you do: Optimize your title and snippet so they deliver on their promises, and track brand searches in Search Console as an indicator of whether your brand is growing.
Which 5 citation factors determine whether you will be cited in AI-search?
The five citation factors in AI search are: definition first, named sources, visible update date, third-party references, and technical accessibility. Google has not published them as a list. They are InboundCPH’s summary of what distinguishes the pages cited by Google AI Overviews, AI Mode, and ChatGPT from the pages they retrieve but exclude. The basis for this is the raw AI responses with source lists that we save in InboundCPH’s AI-SEO Tools every time we analyze a question.
The status of the five varies. Technical availability has been confirmed by Google in The Guide to Generative AI in Google Search, updated July 10, 2026. Original content reflecting our own perspective is also Google’s own recommendation in the same document. The other three are observations from the industry and our own, supported by data but not by an official statement from Google or OpenAI.
Definition first
Pages where the first paragraph answers the question in a single sentence are cited more often than pages that begin with context. A language model selects passages, not pages, and a passage that stands on its own is easier to use. Test: Can your first sentence be copied into an AI response without losing its meaning?
Cited Sources
Statistics that include names, years, and links are reproduced; “most studies show” is not. Language models must be able to stand behind their answers, and a claim with a traceable source is easier to defend. Your own measurements, with a described methodology, are the strongest source because they also demonstrate experience.
Visible update date
Google itself explains that AI responses are based on retrieving updated pages from the index. A page without a date is treated as potentially outdated. In our measurements from 2026, pages with a visible revision date more often win the citation on rapidly changing topics, even over pages with better organic rankings.
Third-Party Coverage
Ahrefs’ analysis of 75,000 brands showed that brand mentions correlate with AI visibility with a strength of 0.66, compared to 0.22 for backlinks. An AI response typically mentions 3 to 5 brands, and those brands are the ones that other sites already mention in the same context. At the same time, Google warns against inauthentic mentions, so it’s mentions with substantive content that count.
Technical Accessibility
Google requires that the page be indexed, be eligible for snips, and not be excluded from AI features in Search Console. ChatGPT, Perplexity, and Claude crawl pages using their own crawlers, which must be granted access in robots.txt. Google also states that llms.txt and special markup are not used by Google Search.
Structured data is not on the list. Google explicitly states that no special schema.org markup is required for the AI features, and that breaking content down into small chunks is also unnecessary. How the five factors are measured across ChatGPT, Perplexity, and Google is explained in our article on Visibility in AI Search.
Where do the two lists overlap, and where do they not?
The two lists overlap in terms of content, authority, and technique, but they are weighted differently: Google ranks pages, while AI responses select passages and brands. InboundCPH’s 2026 Factor Map is our overview of 10 ranking factors and 5 citation factors, with the status of each: confirmed by Google, partially confirmed, or observed. We use this chart when a client asks whether a particular effort will improve their Google ranking, AI citations, or both.
| Factor | Google ranking | AI citation | Status |
|---|---|---|---|
| Relevance to the search intent | Yes, basically | Yes, but at the passage level | Confirmed |
| Helpful, original content | Yes, core system since March 2024 | Yes, Google's AI guide recommends taking a personal perspective | Confirmed |
| Links from other websites | Yes, PageRank | Weaker; media coverage carries more weight (0.66 vs. 0.22) | Confirmed / observed |
| E-E-A-T signals | Yes, as a direction | Yes, author and sources | Confirmed as a direction |
| Freshness | Depends on the search | Yes, visible date | Confirmed / observed |
| Core Web Vitals | Yes, weight limit | No documented effect | Confirmed |
| Mobile-Friendliness and HTTPS | Yes, you can | Only as a prerequisite | Confirmed |
| Crawlability and Indexing | Prerequisite | Prerequisite, plus AI crawlers | Confirmed |
| No spam | Yes, you can | Yes, the same policies apply | Confirmed |
| User Behavior and Brand | Partially confirmed (Navboost) | Brand mentions, observed | Partially confirmed |
| Definition first | Indirectly, based on relevance | Yes, you can | Observed |
| Cited Sources | Indirectly, via E-E-A-T | Yes, you can | Observed |
| Structured data | Rich results only | No, according to Google | Confirmation not required |
We measured a concrete example of this difference on September 6, 2026, in InboundCPH’s AI-SEO Tools, using the question “What are the most important ranking factors in Google in 2026?” Google displayed an AI-generated response, and our article about Google I/O 2026 was one of the six sources cited in the response’s introduction. In AI Mode, the same article was the first source cited. ChatGPT with web search retrieved 20 sources, cited 2 of them, and neither Google’s own documentation nor our article was among those 2. Same question, three interfaces, three different source selections.
"Placement" and "Quote" are two different contests
Google ranks pages based on signals that Google itself has described. AI responses select passages and brands based on patterns that no one has made public, but that can be measured in the raw responses. A page can win one competition and lose the other. The only way to know where you stand is to measure both aspects for the same questions.
How does a Danish B2B company prioritize its efforts?
A Danish B2B company should prioritize ranking factors and citation factors in the order where confirmed prerequisites come first and observed factors come last. The order listed below is the one we use at InboundCPH when a pilot project is expected to deliver measurable results within 3 to 6 months. An e-commerce site with 40,000 pages faces different technical challenges than a consulting firm with 60 pages, but this order holds true for most B2B sites we see.
Assess the baseline with 20 questions
Select 20 questions that your customers ask before choosing a supplier. Track organic rankings, whether Google displays an AI-generated answer, whether you’re cited in it, and what ChatGPT responds. Save the raw responses along with a list of sources. Without a baseline, you won’t be able to distinguish the impact of your work from normal fluctuations.
Clear the obstacles out of the way
Indexing, robots.txt for both Googlebot and AI crawlers, HTTPS, the mobile version, and the pages marked in red in Core Web Vitals. These are confirmed factors with clear solutions, and they can typically be resolved in 2 to 4 weeks. Don’t get stuck here any longer than necessary.
List the pages that can be cited
One page per priority issue, with the definition first, named sources, original figures, a named author, and a visible revision date. This covers relevance, helpful content, E-E-A-T, and three of the five citation factors in a single workflow. We call it “dual-intent content”: semantic depth for Google, citable authority for AI.
Include a description before the links
Place articles, data, and citations in Danish media outlets and industry portals that AI engines have been proven to source from. This coverage directly benefits AI citations and usually includes links, so the Google aspect of the strategy is covered as well. The Azets pilot included 22 placements over 3 months. AI Overview visibility rose from 55 to 88 percent, and the organic ranking for “Interim CFO” went from 12th to 2nd.
Measure again after 30 and 90 days
Run the same 20 questions again. Changes in Google typically occur before changes in ChatGPT because Google’s index is updated faster than what the language models retrieve and weigh. Also note what didn’t change. That will determine whether the next round should focus on more content or more exposure.
Some things are intentionally left for last—or omitted entirely. Structured data beyond rich results isn’t part of the plan, and neither is llms.txt, because Google has confirmed that Google Search ignores that file. What AI-SEO covers as an overall discipline is outlined in our Introduction to AI-SEO.
Frequently Asked Questions About Ranking Factors
How many ranking factors does Google have?
Google has not disclosed the exact number of ranking factors. The figure “over 200” comes from Google’s own statements in the 2000s and is not used by Google today. Instead, Google describes its ranking as a series of systems, including PageRank, freshness systems, spam systems, and the former ’helpful content” system, which became part of the core systems in March 2024. The list is included in Google’s guide to ranking systems, last updated in December 2025.
Will Core Web Vitals be a ranking factor in 2026?
Yes, Core Web Vitals have been a ranking factor since June 2021, and Google confirms in its documentation on page experience that they are used by the ranking systems. Google also states that good metrics do not guarantee top rankings. The thresholds are LCP under 2.5 seconds, INP under 200 milliseconds, and CLS under 0.1. INP replaced FID as a metric in March 2024.
Is E-E-A-T a ranking factor?
No, E-E-A-T is not a single ranking factor with a score. E-E-A-T is Google’s framework for experience, expertise, authority, and trustworthiness from the Quality Rater Guidelines, expanded to include experience on December 15, 2022. Google states that its ranking systems look for signals that align with the framework, such as named authors, sources, and mentions by others. The impact is greatest for topics related to health, finance, and law.
What determines whether you’ll be cited by AI?
Whether a page is cited by AI is determined by InboundCPH’s assessment of five factors: definition first, named sources, visible update date, third-party mentions, and technical accessibility. Only the last factor has been confirmed by Google, which requires indexing and the ability to display excerpts. In an analysis of 75,000 brands, Ahrefs found that brand mentions correlate with AI visibility at 0.66, compared to 0.22 for backlinks. An AI response typically mentions 3 to 5 brands.
Are GEO and AEO different disciplines from SEO?
According to Google, GEO and AEO are not separate disciplines from SEO. In its July 2026 guide to generative AI in Google Search, Google states that optimization for AI Overviews and AI Mode is optimization for the search experience and therefore still SEO. At InboundCPH, we use the term “AI SEO” as an umbrella term for GEO, AEO, and LLMO, because the five citation factors build upon the classic ranking factors rather than replacing them.
See which factors are missing from your most important pages
An AI visibility analysis measures your 20 most important questions on Google, AI Overviews, and ChatGPT, and shows which pages are cited instead of yours. You’ll receive the raw results, including source lists and a factor map for your own site.
Sources
- Google: How Search Works, Ranking Results
- Google Search Central: A Guide to Google Search Ranking Systems (Updated December 2025)
- Google Search Central: Understanding Page Experience and Core Web Vitals
- Google Search Central: Optimizing Your Website for Generative AI Features on Google Search (updated July 10, 2026) and AI Features and Your Website
- Google Search Central: Spam Policies
- InboundCPH's AI-SEO Tools, measurement on September 6, 2026, “Modspil” project”
Last revised on September 6, 2026. Google Docs were accessed in the version available online on the revision date; metrics from InboundCPH's AI-SEO Tools are updated every 30 days.
