Google uses hundreds of signals to decide which pages appear in search results and in what order. For most of SEO’s history, the specific signals were a matter of informed speculation — practitioners tested, observed, and theorized based on limited public guidance from Google.
That changed in May 2024 when 14,014 internal API attributes leaked from Google’s Content Warehouse. The same year, the DOJ antitrust trial forced Google executives to testify under oath about how their ranking systems work. For the first time, we have evidence — not just Google’s public statements, but their internal documentation and sworn testimony — about which signals exist and how they function.
This guide focuses on the ranking factors that are either confirmed by the leak, verified through the antitrust trial, or officially stated by Google. Where a factor is speculated but unconfirmed, that’s noted. The goal isn’t a list of 200 items you’ll never remember — it’s the factors that actually move rankings in 2026, organized by how much evidence supports them.
The Factors Confirmed by the 2024 API Leak and Antitrust Trial
Before covering the full landscape, it’s worth isolating the signals that the leak and trial specifically confirmed — because several of these directly contradict years of public statements by Google engineers.
siteAuthority
Google denied for over a decade that it calculates or uses any kind of domain-level authority metric. The leaked documents contain a siteAuthority integer field within the CompressedQualitySignals module that functions as a persistent, composite score applied at the site or sub-domain level. It serves as a foundational input for preliminary ranking in Google’s Mustang system.
This doesn’t mean Google’s siteAuthority is identical to third-party “Domain Authority” metrics from Moz or Ahrefs. The internal calculation methodology isn’t documented. But the concept — that Google evaluates sites at the domain level, not just individual pages — is confirmed.
NavBoost
NavBoost is Google’s click-driven re-ranking system. It was mentioned 84 times in the leaked documents and independently confirmed under oath by Google VP Pandu Nayak during the antitrust trial, where he called it “one of the most important” ranking signals.
Google publicly denied using click data for rankings for years. NavBoost tracks goodClicks (satisfied clicks with long dwell time), badClicks (quick returns to search results), lastLongestClicks (the final result a user clicked and stayed on), and unsquashedClicks (clicks deemed genuine). It operates on a rolling 13-month window and is region-specific.
NavBoost effectively means that user satisfaction with your page — measured through real click behavior — is a direct ranking input, not just a correlated metric.
chromeInTotal
Google representatives, including Matt Cutts and John Mueller, stated publicly that Google does not use Chrome browser data for rankings. The leaked documents contain a chromeInTotal attribute that captures site-level views from Chrome browsers. Additional attributes track Chrome user behavior including time on page, bounce patterns, and browsing frequency.
With Chrome holding approximately 65% of global browser market share, this gives Google access to a massive behavioral dataset that no other search engine can replicate.
OriginalContentScore and contentEffort
The leak revealed an OriginalContentScore that evaluates content uniqueness, particularly for shorter content (scored 0-512). A related contentEffort attribute appears to use large language models to estimate the effort required to create an article — helping Google distinguish between content requiring genuine expertise and content that was assembled by scraping, translating, or algorithmically remixing existing material.
hostAge (Sandbox)
Google denied the existence of a “sandbox” for new websites since the mid-2000s. The leaked PerDocData module contains a hostAge attribute described as being used to identify and handle “fresh spam during serving time.” This confirms that new domains receive different treatment — reduced ranking capability — during an initial period after creation.
smallPersonalSite
An attribute called smallPersonalSite exists in the documentation. Its function is unknown — it could promote or demote small personal sites. Given that Google’s Helpful Content Update and subsequent core updates significantly reduced visibility for many small niche sites, some analysts speculate it’s currently used for demotion. This remains unconfirmed.
Content Quality and Information Gain
Content quality has been a ranking factor since Google existed. What’s changed in 2026 is how quality is measured and what “quality” means in an environment where AI can generate passable content on any topic in seconds.
The Information Gain Threshold
The most important content differentiation factor in 2026 is information gain — publishing something that didn’t exist before you published it. This includes original data from your own research or business operations, first-hand experience with products, services, or processes, expert analysis that synthesizes information in a new way, proprietary frameworks or methodologies, and primary source reporting.
AI-generated content can competently summarize existing information. What it can’t do is create new information from real-world experience. The contentEffort attribute in the leak suggests Google is actively trying to measure this distinction. Content that clearly required human expertise, original research, or direct experience to create scores differently from content that could have been assembled by an algorithm.
This doesn’t mean AI-assisted content is automatically penalized. Google’s official guidance states that AI-generated content is acceptable if it’s helpful and demonstrates E-E-A-T. But content that adds nothing new to what’s already indexed — regardless of whether a human or AI wrote it — is at a structural disadvantage for ranking.
E-E-A-T: Experience, Expertise, Authoritativeness, Trustworthiness
E-E-A-T isn’t a direct ranking factor — it’s a framework Google uses to train its quality raters, whose evaluations inform algorithm development. But the leaked documents confirm that several E-E-A-T components have direct algorithmic representation.
Experience — the first “E,” added in December 2022 — evaluates whether the content creator has firsthand experience with the subject. Original product photography, staff-written reviews, personal case studies, and “tested by our team” content all demonstrate real experience. The leak’s author-entity recognition attributes connect directly to this signal.
Expertise — demonstrated through depth of knowledge, accurate technical details, and qualifications. Author profiles with verifiable credentials, detailed technical content, and cited sources strengthen expertise signals.
Authoritativeness — earned through external validation. Quality backlinks, citations from trusted publications, industry recognition, and consistent brand presence build authority. The siteAuthority metric confirmed in the leak reflects the algorithmic representation of this concept.
Trustworthiness — the most critical component, according to Google’s documentation. Secure site (HTTPS), clear contact information, transparent editorial policies, accurate sourcing, and visible privacy policies build trust. Google states explicitly: “untrustworthy pages have low E-E-A-T no matter how Experienced, Expert, or Authoritative they may seem.”
E-E-A-T carries extra weight for YMYL (Your Money or Your Life) topics — content about health, financial decisions, safety, or civic information where inaccurate content could cause real harm.
Search Intent Alignment
Search intent — whether the user wants to learn, compare, buy, or navigate — is the foundational relevance signal. A technically perfect page will fail to rank if it doesn’t match what the user is actually looking for.
Google’s algorithms now evaluate intent alignment at three levels (sometimes called the “3 Cs”): content type (is it a blog post, product page, or video?), content format (is it a how-to guide, a listicle, a comparison, or a review?), and content angle (what’s the unique perspective or hook?).
The practical test: search your target keyword and study the top 5 results. If they’re all comparison articles and you’re publishing a product page, you have an intent mismatch. If they’re all long-form guides and you’re publishing a 200-word summary, your format doesn’t match. Aligning with the dominant intent pattern in current results is the baseline requirement for ranking.
Title tags and meta descriptions serve as the interface between your content and the search results page. The leaked TitleMatchScore attribute confirms Google measures how well your title matches user queries. Place your primary keyword near the front of the title, keep it under 60 characters, and write a meta description that accurately previews the page content with a clear reason to click.
Backlinks and Link Authority
The 2024 leak confirmed that backlinks remain a significant ranking factor. The siteAuthority attribute itself is built partly from link signals. Despite Google’s occasional suggestions that links are becoming less important, every piece of evidence — the leak, the antitrust trial testimony, and observable ranking patterns — indicates links still carry substantial weight.
What the Leak Tells Us About Links
Google evaluates links across multiple dimensions: the authority of the linking domain, the topical relevance of the linking page to the target page, anchor text signals (with demotion for mismatched or spammy anchors), link diversity (unique referring domains matter more than total link count), and link freshness (a continuous stream of new links signals ongoing relevance).
The anchorMismatchDemotion attribute penalizes pages receiving links with anchor text that doesn’t match the page’s content. The IsAnchorBayesSpam flag detects spam anchor text patterns. Over-optimized anchor text distribution — too many exact-match anchors — is a demotion trigger.
Natural Anchor Text Distribution
Based on analysis of top-ranking pages, a natural anchor text profile typically includes approximately 50% branded anchors (your company or site name), 25% topically relevant anchors (contextual terms related to your content), 15% generic anchors (“click here,” “learn more,” “this article”), and 10% or less target keyword variations (exact match and phrase match combined).
Pages with an unnatural concentration of exact-match anchor text — especially if acquired in a short timeframe — trigger manipulation flags. Link velocity (the rate at which you acquire new links) should be consistent with your site’s organic growth pattern. Sudden spikes in link acquisition with similar anchor text are a red flag.
Topical Authority and Site Focus
The leaked SiteFocusScore and SiteRadius attributes confirm that Google evaluates topical authority at the site level, not just the page level.
SiteFocusScore measures how concentrated your site’s content is around specific topics. SiteRadius measures how far individual page embeddings drift from the overall site embedding. Together, they reward sites with clear thematic identity and penalize sites that publish widely unrelated content.
This explains why niche sites often outrank larger generalist sites for specific queries — a tighter SiteFocusScore sends a stronger topical signal. It also explains why randomly publishing content outside your core topic area can dilute your ranking power across the board.
The practical application is the pillar-cluster content model: build comprehensive coverage of your core topics through interconnected content, then expand into adjacent topics only when you can create logical thematic bridges.
Internal Linking
Internal linking doesn’t get its own named attribute in the leak, but its importance is well-established through multiple mechanisms. Internal links distribute link equity across your site. They help search engines understand content relationships and hierarchy. They signal which pages you consider most important. And they create navigation paths that affect user engagement metrics.
A SearchPilot split test found that adding internal links to a homepage footer produced a 5% uplift in organic traffic to the linked pages, with desktop traffic jumping 10%. Another test showed that simply changing the anchor text of existing internal links — without adding or removing any links — produced a statistically significant positive result for the destination pages.
For ecommerce and content-heavy sites, internal linking is one of the highest-leverage optimization activities available because it’s entirely within your control and can be implemented at scale.
User Engagement and NavBoost
NavBoost transforms user engagement from a correlated metric into a confirmed ranking input. The system tracks click quality across a 13-month rolling window, which means pages need to satisfy users consistently — a viral spike decays as old click data rolls out, while pages that reliably satisfy searchers accumulate positive signals over time.
The practical implications: your title tag and meta description are ranking assets because they earn the initial click that starts the NavBoost record. Your content quality determines whether that click becomes a goodClick (user stayed, engaged) or a badClick (user bounced back to results). And your page’s position as the lastLongestClick in a session — the result that ended the user’s search — is a particularly strong satisfaction signal.
Pogo-sticking — users clicking your result, bouncing back to the SERP, and clicking a competitor’s result — sends a clear negative signal. The best defense against pogo-sticking is matching content to search intent, answering the query immediately (don’t bury the answer below a wall of preamble), and providing enough depth that the user doesn’t need to continue searching.
Page Experience and Core Web Vitals
Google confirmed Core Web Vitals as ranking signals in 2021. In 2026, the three metrics are:
Largest Contentful Paint (LCP) — loading speed. Target: under 2.5 seconds. The LCP element is typically the largest image or text block in the viewport.
Interaction to Next Paint (INP) — interactivity. Replaced First Input Delay (FID) in March 2024. Target: under 200 milliseconds. Measures how quickly the page responds to user interactions throughout the visit, not just the first one.
Cumulative Layout Shift (CLS) — visual stability. Target: under 0.1. Measures unexpected content shifts during page load.
Core Web Vitals are confirmed ranking factors but carry relatively low weight compared to content quality and links. Google has described them as a “tiebreaker” — when two pages are otherwise similar in quality and relevance, the one with better page experience wins. Their larger impact is indirect: poor page performance causes bad user engagement signals (high bounce rates, short dwell times) that feed into NavBoost.
HTTPS is a confirmed but lightweight ranking signal — Google stated it affects less than 1% of global queries. Mobile-first indexing means Google evaluates the mobile version of your site for ranking purposes. Responsive design is Google’s recommended approach.
Content Freshness
The leaked documents reveal three freshness signals: bylineDate (stated publication date), syntacticDate (date in the URL or title), and semanticDate (estimated date based on content context). A lastSignificantUpdate attribute appears to detect cosmetic date changes versus substantive content updates.
Freshness matters primarily for queries where timeliness is relevant — Google’s Query Deserves Freshness (QDF) algorithm boosts recent content for trending topics, breaking news, and rapidly changing subjects. For evergreen topics (“how does photosynthesis work”), freshness carries minimal weight.
When updating content, aim for roughly 30% substantive change: cut outdated sections, add new data and examples, update recommendations, and include information that didn’t exist when the page was originally published. Changing only the byline date is detectable and may be downweighted.
Brand Signals
After analyzing the full API leak, Rand Fishkin’s strategic conclusion was direct: “If there was one universal piece of advice I had for marketers seeking to broadly improve their organic search rankings and traffic, it would be: Build a notable, popular, well-recognized brand in your space, outside of Google search.”
This conclusion follows from multiple leaked signals. SiteAuthority rewards established, trusted domains. NavBoost rewards sites that generate consistent positive engagement — and branded searches (people searching for your company name) produce some of the highest-quality click signals possible. ChromeInTotal captures direct traffic and repeat visits. Author entity recognition rewards known, credible experts.
Building brand signals means investing in activities that don’t look like traditional SEO: PR coverage, industry events, podcast appearances, social media presence, community building, and customer experience. These generate the brand recognition that feeds the algorithmic signals Google actually measures.
Local SEO Ranking Factors
Local search uses additional ranking signals beyond the standard organic factors.
NAP consistency — your business Name, Address, and Phone number should be identical across your website, Google Business Profile, directories, and social profiles. Inconsistent NAP data fragments your business identity in Google’s view. Businesses with consistent NAP data are 40% more likely to appear in the local pack.
Google Business Profile — your primary local SEO asset. Complete every field, respond to reviews, post regular updates, and ensure your business categories accurately reflect your services.
Customer reviews — reviews account for approximately 9% of local pack ranking factors. Volume, recency, diversity, and sentiment all matter. Responding to reviews increases conversion rates — each 25% increase in response rate produces a 4.1% improvement in conversion.
LocalBusiness schema markup — structured data that tells search engines your business name, address, hours, service area, and other operational details. Properly implemented schema enables rich results in search listings.
Generative Engine Optimization (GEO): The New Visibility Dimension
In 2026, ranking on the traditional SERP is only half the visibility equation. AI Overviews now appear on approximately 48% of all queries. Around 60% of searches result in no click at all. Being cited in AI-generated answers — by Google’s AI Overviews, ChatGPT, Perplexity, or Claude — is a separate discipline from traditional ranking.
Research shows the overlap between top-10 Google ranking pages and AI-cited sources has collapsed from approximately 70% to below 20%. Ranking first organically does not guarantee being cited in the AI Overview displayed above your result.
GEO optimization focuses on making your content citable by AI systems: structuring information in extractable chunks (specification tables, comparison charts, concise definitional statements), building entity clarity through consistent naming and schema markup, earning third-party mentions from multiple independent sources, and publishing original data and analysis that AI systems can reference as primary source material.
GEO doesn’t replace SEO — it builds on the same fundamentals (quality content, authority, technical health). But it adds a new layer of optimization focused on how AI systems retrieve, evaluate, and cite information rather than how traditional search algorithms rank pages.
Factors That Don’t Matter (Despite Common Belief)
Clearing up misconceptions is as important as identifying what works. The following factors are commonly cited in SEO discussions but have no confirmed impact on rankings:
Keyword density — Google does not use a target keyword density percentage. Using your keyword naturally is sufficient. Stuffing keywords harms readability and can trigger spam detection.
Word count — Google does not rank pages based on length. A 500-word article that perfectly answers a query can outrank a 5,000-word article that doesn’t. The leaked OriginalContentScore evaluates originality, not length.
Social media signals — Google has stated that social shares, likes, and followers are not ranking factors. Social activity can drive traffic and brand awareness, which indirectly affects rankings, but the social metrics themselves are not signals.
Exact-match domains — owning an exact-match domain (e.g., bestrunningshoes.com) provided a ranking advantage in Google’s early years. This benefit was largely eliminated by the 2012 EMD Update and carries no meaningful advantage in 2026.
AI content penalty — Google does not penalize content for being AI-generated. It evaluates content quality regardless of production method. Low-quality AI content fails for the same reasons low-quality human content fails: thin information, no expertise, and no original value.
Demotion Signals to Avoid
The leak cataloged several explicit demotion mechanisms. Understanding what triggers demotion is as important as understanding what drives ranking.
anchorMismatchDemotion — inbound links with anchor text that doesn’t match the destination page’s content trigger demotion. Google evaluates both sides of a link for topical relevance.
GibberishScore — detects artificially generated, scraped, or machine-translated content. High gibberish scores risk page removal from the index. Recovery takes 3-6 months or longer.
trendSpam — detects artificial click manipulation. Manufactured click patterns may produce temporary ranking improvements, but gains reverse once artificial clicks stop, and persistent manipulation risks permanent demotion.
Nav Demotion — penalizes sites with poor navigational experience. Important pages buried deep in submenus may not be indexed at all.
Panda — the leaked documentation confirms Panda uses embeddings for content quality assessment. It functions as a site-level quality modifier — a Panda demotion affects your entire domain, not just individual pages.
Frequently Asked Questions
What is the single most important Google ranking factor?
There isn’t one. Google uses a layered system where multiple signals interact. But if forced to prioritize, the evidence points to content that matches search intent and satisfies the user (confirmed through NavBoost click signals) as the foundational requirement. Without intent alignment and user satisfaction, no amount of technical optimization or link building produces sustainable rankings.
Does Google use AI-generated content as a negative ranking signal?
No. Google evaluates content quality regardless of how it was produced. AI-generated content that provides genuine value, demonstrates expertise, and satisfies user intent is treated the same as human-written content of equivalent quality. The risk with AI content is that it typically lacks original information gain — the one quality signal that AI currently can’t replicate at scale.
How important are backlinks in 2026?
Still very important. The leaked siteAuthority attribute is built partly from link signals. The antitrust trial confirmed links remain a core ranking input. What’s changed is the emphasis on quality over quantity: link source authority, topical relevance, anchor text naturalness, and referring domain diversity all matter more than raw link volume.
How long does it take for a new website to rank?
The leaked hostAge attribute confirms that new domains face a period of reduced ranking capability. Practitioners estimate this period typically lasts 6-18 months for competitive queries. The timeline depends on how quickly you build authority signals: quality content, authoritative backlinks, brand recognition, and positive user engagement patterns.
Are Core Web Vitals still ranking factors?
Yes, but their direct weight is relatively low — Google has described page experience signals as a “tiebreaker” between otherwise similar pages. Their indirect impact through user engagement is larger: slow, unstable pages produce poor NavBoost signals (high bounce rates, short dwell times) that compound into ranking decreases over time. Note that INP (Interaction to Next Paint) replaced FID (First Input Delay) as a Core Web Vital in March 2024.
What is GEO and do I need it?
Generative Engine Optimization is the practice of making your content citable by AI systems — Google AI Overviews, ChatGPT, Perplexity, and similar platforms. With AI Overviews now appearing on approximately 48% of queries and 60% of searches producing no click, GEO is increasingly important for maintaining visibility. It doesn’t replace SEO — it builds on the same fundamentals — but it adds optimization for how AI retrieves and cites information, which is a different mechanism from how traditional search ranks pages.
