How Google Search works: crawling, indexing, ranking, in plain words

• Crawling: Googlebot discovers URLs from links, sitemaps and past crawls, then fetches them. It renders JavaScript with an evergreen Chromium in a second wave.
• Indexing: Google analyses text, images and video, understands the topic, groups duplicates, picks a canonical, and stores the page. Not every crawled page is indexed.
• Serving: for a query, systems rank pages by relevance, quality, usability and context. The results page adds features around them.
• There is no fee and no submission needed. A public page linked from a known page is found. A sitemap and Search Console speed discovery.
• For a new site, discovery and indexing take days to a few weeks. Ranking for competitive queries takes months. There is no confirmed sandbox.

1. 1

   You publish

   You ship /guides/getting-started on yourproduct.com. It is linked from /guides and the homepage.

2. 2

   Google discovers

   Googlebot follows the link from your homepage or reads it from your sitemap.xml. If neither exists, discovery waits for a link from another site.

3. 3

   Google crawls

   Googlebot fetches the HTML and static assets it sees. Heavy scripts or blocked assets delay or prevent full understanding.

4. 4

   Google renders

   In a second wave, Google renders the page with evergreen Chromium to load client-side content. If key text only appears after user actions, Google may not see it.

5. 5

   Google indexes

   Google decides if the page is worthy of indexing. It picks a canonical if there are duplicates and stores the page with its signals. Some pages are crawled but not indexed.

6. 6

   Google serves

   When someone searches “getting started yourproduct” in your country, the ranking systems may show your page with a title and snippet. Features around it depend on the query.

A healthy path for /guides/getting-started: discovered from your sitemap on day 1, crawled on day 1 or 2, rendered within days, indexed within a week, first impressions within two weeks. There is no promise on timing.

• Link the page from a page Google already knows. Add it to your main nav, footer and any relevant page. Use clear anchor text like “Getting started”.
• Submit a clean sitemap.xml. Include only canonical 200 URLs. Update it on publish. Point to it in robots.txt and in Search Console’s Sitemaps report.
• Keep robots.txt open. Do not block CSS, JS or images you need for rendering. Never block your whole site on production.
• Use URL Inspection’s Request indexing on one key URL. There is a small daily quota. It asks for a recrawl, not an instant ranking boost.
• Make the server fast. Return 200 for live pages, 301 for moved, 404 or 410 for gone. Avoid chains. Keep Time to First Byte low.
• Ensure mobile-first crawlability. Serve the same primary content on mobile and desktop. No interstitials that hide content on load.

What a fixed page looks like: /pricing is linked from the header, is in the sitemap, returns 200, loads key text without client-only rendering, and is not blocked by robots.txt.

• Give each page a single canonical URL. Use rel=canonical to point to itself. Only point away if you truly want Google to index the other URL instead.
• Avoid thin duplicates. Do not ship empty tag pages or parameter variants that add no value. Consolidate with canonicals or redirects.
• Set indexability on purpose. No noindex on live pages. Avoid conflicting signals like a canonical to A but a redirect to B.
• Use a unique title and primary content that states the topic. A page with the same pitch as five others is likely to be skipped.
• Check Search Console’s Page indexing report. “Alternate page with proper canonical tag” is fine if that is your intent. Fix it only when the canonical is wrong.
• Fix “Discovered, currently not indexed” by adding internal links and improving content quality. Fix “Crawled, currently not indexed” by making the page more useful and unique.

If /guides/getting-started and /guides/start-here are the same, redirect one to the other and keep one canonical. After that, Google keeps one indexed URL.

Google lists ranking systems it uses as of 2026. They include BERT, crisis information systems, deduplication, exact match domain system, freshness systems, helpful content guidance folded into core, link analysis systems and PageRank, local news, MUM, neural matching, original content systems, removal-based demotion, page experience system, passage ranking, reviews system, RankBrain, reliable information systems, site diversity system, and spam detection systems like SpamBrain.

• What you can influence on a small site:
• Match intent. Write the page a searcher wants. If the query is “yourproduct pricing”, show plans, limits and a clear next step on /pricing.
• Be helpful, original and trustworthy. Show who wrote it, how it works, evidence, and first-hand experience where relevant.
• Earn links. Links remain a signal among many. Publish pages worth citing and earn mentions from sites in your space.
• Page experience. Serve over HTTPS, be mobile friendly, avoid intrusive interstitials, and hit Core Web Vitals: LCP about 2.5 seconds, CLS about 0.1, INP about 200 ms.
• Technical basics. Make pages crawlable, indexable, canonical, and titled. Use clear headings. Use descriptive alt text for meaningful images.
• Target the right queries. A new site will not rank for “project management” soon. It can rank for “yourproduct project templates” faster.

• What does not matter the way myths claim:
• There is no single “domain authority” score at Google. Third-party scores do not feed Google directly.
• Keyword density, meta keywords and word count targets do not drive rankings.
• Social shares are not a direct ranking signal. They can help discovery and linking.
• Lists of “200 ranking factors” are folklore. Focus on the user, not a checklist of numbers.

A fixed /pricing page ranks for “yourproduct pricing” with a clear H1, honest details, fast load and real plans. It earns links from reviews and comparisons.

• Discovery and indexing take from days to a few weeks for a new site. There is no promise.
• Ranking on competitive queries takes months. A 2017 Google video said four months to a year for SEO work to show its benefit.
• Nothing in Search Console until the property is verified and crawled. Data lags about two days. A new property shows the last 16 months of what Google already had for the site.
• There is no confirmed sandbox. New sites struggle for competitive terms because others already match intent with stronger signals.

• First, search site:yourproduct.com. If nothing shows, Google may not have crawled it or it is blocked.
• In Search Console, check the Page indexing report. Then use URL Inspection on a sample URL.
• Check robots.txt and the meta robots tag. Remove a stray noindex, and never ship Disallow: / on production.
• Submit a sitemap in Search Console. Ensure it lists clean, canonical 200 URLs.
• Get at least one link from a known site. A profile or a simple mention can trigger discovery.
• If your brand name is generic, expect to add a qualifier to your query. Rankings for brand terms without uniqueness are hard.

Classic case: a staging noindex shipped to production. Fix the tag, request indexing on the homepage, and wait for the next crawl to restore listings.

• Use Search Console’s Performance report. It is Google’s own source for average position by query and page.
• For a spot check, search in an incognito window in the right country. Results are personalised by location and history otherwise.
• A rank tracker can monitor terms at scale. For traffic, use Search Console for Google clicks and your analytics for all sessions by channel.
• Similarweb or a suite can estimate another site’s traffic, with limits.
• Search Console definitions: an impression is when a link to your page appears on a results page a user loaded. For web search, scrolling is not required.
• Position is the highest link to your site on that page. Average position is the mean over impressions. A page can average 8 while appearing at 3 and at 40.
• CTR is clicks divided by impressions. The queries table omits rare and anonymised queries, so its sums are below the page total. The report holds up to 16 months. The default comparison is the last 28 days against the 28 before.

1. 1

   Find the drop

   In Search Console’s Performance report, compare the period to the one before. Split by page and query to see where it fell.

2. 2

   Check Google’s side

   Look at the Google Search Status Dashboard for a core update or spam update on those dates. Check Manual actions and Security issues.

3. 3

   Check your side

   Review deploys. Look for redirects, noindex, canonical or robots changes. Check server errors and CDN rules.

4. 4

   Check seasonality

   Compare the same period last year. Many niches dip in holidays or off seasons.

5. 5

   Check the results page

   Search the queries. See new rivals or new features that push organic links down, such as visual units or an AI Overview.

If /guides/getting-started lost clicks but kept position, a new feature above it may have reduced CTR. Rewrite the title to fit the query better and test again.

• Results pages change by query. News, images, videos, shopping, local and AI Overviews can appear and shift clicks.
• Impressions can rise while clicks stay flat if you moved to page one but below new features. Expect lower CTR on crowded pages.
• Titles and descriptions matter for CTR. Keep titles under about 60 characters. Write a clear meta description of 120 to 155 characters.

A fixed /guides/getting-started shows a concise title and a meta description that matches “getting started with yourproduct”, winning a higher CTR at the same rank.