How Google Actually Ranks Pages

An honest look at ranking signals, confirmed factors, and the gap between what Google says and what practitioners observe.

Google's Honesty Problem

Google claims there are over 200 ranking signals. But Google does not publish the list, the weights, or the algorithm. This creates an awkward truth: we have to infer how Google ranks pages by observing patterns, running tests, and listening to occasional public hints from Google employees.

This page separates what Google has explicitly confirmed, what practitioners have strong evidence for, and what is genuinely uncertain. Avoid anything in the "myths" column — chasing unproven factors wastes time.

Confirmed Ranking Signals

These are factors Google has publicly acknowledged influence ranking. This is the safe ground to build on.

Content Relevance

Your page must answer the user's query. This sounds obvious, but it is constantly done wrong. A page about "how to train a dog" that does not mention training techniques will not rank, even if it has 500 words and a domain authority of 50.

Relevance is about semantic matching: your page should cover the main subtopics users are searching for. Google uses BERT and neural matching to understand not just keywords, but meaning. If a user searches for "best laptops for video editing," a page needs to discuss editing-specific performance needs (RAM, GPU, CPU), not just list expensive laptops.

Backlinks

Google has consistently maintained that links are a vote of confidence. A page with 50 links from quality sites will outrank a page with 5 links from low-quality sites, all else equal.

The nuance: link quality matters far more than quantity. A link from a relevant, authoritative site (like a major publication or educational institution) is worth 100+ links from low-quality directories. Google also weights anchor text — a link with text "best SEO practices" signals more about the linked page than a link with generic text like "click here."

Most sites do not have a link-building problem — they have a content quality problem. When your content is genuinely valuable, earning links is much easier. Focus on content first; links follow.

Core Web Vitals (Page Experience)

Google has explicitly stated that page speed and responsiveness affect ranking. Core Web Vitals measure:

• Largest Contentful Paint (LCP): How quickly the main content loads. Target: under 2.5 seconds.
• First Input Delay (FID): How long until the page responds to user interaction. Target: under 100ms. (Replaced by Interaction to Next Paint in 2024.)
• Cumulative Layout Shift (CLS): How much the layout shifts during load. Target: under 0.1. Stable layouts rank better.

Poor Core Web Vitals are not a death penalty, but they are a ranking disadvantage. Most sites can improve LCP by optimising images, reducing JavaScript, and using a CDN. CLS often improves by adding dimensions to images and avoiding pop-ups that shift content.

Mobile-Friendliness

Google prioritises mobile experience. Mobile-first indexing means Google crawls and ranks the mobile version of your site first. If your mobile site is broken, missing content, or slower than desktop, you will lose ranking.

Test your site with Google's Mobile-Friendly Test. If it fails, this is a high-priority fix.

HTTPS

Google has stated that HTTPS is a ranking signal (though a minor one compared to content and links). If you are running HTTP, migrate to HTTPS. It also builds user trust and is now practically mandatory.

Domain Authority

Google does not publish a "domain authority" score, but it clearly considers the overall authority of a domain. Older, established sites with many backlinks tend to outrank new sites on the same topic. This is not unfair — authority is earned over time through consistent quality.

New sites can still rank, but they need better content or different keywords (less competitive, more specific). Targeting low-competition terms initially, then building towards competitive terms, is a sound strategy.

Likely But Unconfirmed Signals

These factors correlate strongly with ranking, but Google has not explicitly confirmed they are direct ranking signals. They may be correlated with ranking rather than causal.

User Engagement Metrics

It is commonly believed that click-through rate (CTR), dwell time, and bounce rate influence ranking. Google has denied this in various public statements. However, practitioners consistently observe that pages with high engagement rank well.

The truth: these metrics probably do not directly influence ranking. Rather, they correlate with pages that satisfy users, which Google rewards through other signals. Focus on satisfying users, and engagement will follow naturally.

Content Freshness

Google has confirmed that freshness is a ranking signal for time-sensitive queries (news, current events, trends). For evergreen topics (how to write a blog post), old content can rank indefinitely if it is still relevant.

Updating old content does not guarantee a ranking boost, but it can help. If you update a page and get re-crawled quickly, you may see a small lift as Google re-evaluates relevance.

Structured Data

Structured data (JSON-LD, microdata) helps Google understand your content. It enables rich snippets, which may improve CTR. Whether structured data directly affects ranking is uncertain, but it clearly helps Google process your content better.

Implement structured data for your content type (Article, Product, Recipe, Event, etc.). Make sure it is accurate — incorrect structured data can hurt trust.

Brand Mentions and Authority

Practitioners observe that sites with strong brands rank well. This may be because brand authority itself influences ranking, or because strong brands naturally accumulate links and citations. Probably both.

You cannot overnight build a brand, but you can be consistent. Author bios, consistent branding across channels, and earned mentions in industry publications all contribute.

Neural Ranking and Machine Learning

Google's core ranking algorithm has shifted from explicit rules to neural networks and machine learning. RankBrain (Google's machine learning system) processes queries and learns what results satisfy users.

This matters because machine learning is not easily gamed. You cannot study a formula and figure out the optimal values. Instead, the algorithm learns by observing billions of search interactions. If your page genuinely satisfies users (they click, stay on the page, do not return to search), the algorithm learns to rank you higher for similar queries.

Practical implication: create content that genuinely helps users. Avoid trying to reverse-engineer the algorithm.

Common Ranking Myths

Myth: Exact Match Keywords in the Title Tag Are Required

False. While titles should be relevant, they do not need to match the query word-for-word. "The Complete Guide to Training Your Dog" ranks for "how to train a dog" without the exact phrase. Relevance is what matters, not exact matching.

Myth: Keyword Density Matters

False. Google does not have a target keyword density. Stuffing keywords at 2-3% density does not help. Natural language and semantic coverage is what matters. If you naturally mention your topic from multiple angles, keyword density will take care of itself.

Myth: Bounce Rate Is a Ranking Signal

Google has explicitly denied this. Bounce rate is a metric in Google Analytics, but it is not a ranking signal. Some pages naturally have high bounce rates (single-page documents, quick answers) without being low-quality.

Myth: The Keyword Must Be in the First 100 Words

False. A page can rank for a query even if the exact phrase does not appear in the first paragraph. Google understands context and synonyms. A page about "how to housetrain a dog" ranks for "dog potty training" without that exact phrase.

Myth: More Content Is Always Better

False. A focused, well-written 2,000-word page beats a padded 5,000-word page. What matters is depth, clarity, and usefulness. Adding filler just to hit a word count target is counterproductive.

Personalisation: Your Ranking Is Not The Real Ranking

Google personalises results for each user based on location, search history, device, and other signals. This means you cannot trust your own search results to gauge ranking performance.

Test your actual ranking using Google Search Console, rank tracking tools, or an incognito browser in a different location. Your local ranking is not representative of real-world performance.

Signals Table

Why Ranking Is Not a Formula

Some practitioners treat ranking like a checklist: include keyword in title (check), get 50 backlinks (check), improve Core Web Vitals (check), publish 3,000 words (check), then profit. This approach rarely works.

Ranking is the output of a neural network trained on billions of search sessions. The algorithm learns the pattern of "what page makes users happy for this query." You cannot game a neural network by hitting arbitrary targets. You can only build genuinely good content and let the system recognise it.

What You Can Control

• Content quality and relevance to user intent.
• Page speed and Core Web Vitals (measurable and improvable).
• Mobile-friendliness and responsive design.
• Author credibility and E-E-A-T signals on the page.
• Earning quality backlinks by creating linkable content.
• Structured data and on-page SEO basics.
• User experience (reducing friction, improving clarity, adding value).

What You Cannot Control

• The exact ranking algorithm or how much weight each signal receives.
• How Google personalises results for individual users.
• Whether your domain receives crawl budget (you can influence it, not control it).
• How other sites rank relative to you (but you can be better).

Focus your energy on what you control. The results will follow.