How Google ranks a page

Two businesses publish the same kind of page on the same day. A month later one of them is on the first page of Google and the other has never been seen. The difference is rarely the writing. It is almost always one of three steps that happen before a page can rank at all, and most of the advice you'll read skips straight past them.

Every page goes through the same pipeline: Google has to find it, it has to understand and store it, and only then does it decide where to show it. Fix the pipeline and ranking becomes a contest you can win. Ignore it and nothing else you do counts.

Step one: Google has to find the page

Google discovers pages by following links and by reading the sitemaps you submit. Googlebot, its crawler, starts from pages it already knows, follows the links on them, and queues up whatever it finds. A page that nothing links to and that isn't in your sitemap is invisible, however good it is.

Two things decide whether your page gets crawled at all:

1. Something points to it. A link from your own navigation, from another page on your site, or from another site. Internal links are the ones you control, so a new page should get a link from a page Google already visits regularly, typically your homepage or a hub page.
2. You haven't told Google to stay away. Your robots.txt file can block crawlers from entire folders, and a page can carry a noindex tag that says "don't store this one". Both are useful on purpose and disastrous by accident. A surprising number of "why doesn't my page rank" problems end with a developer finding a stray block left over from a site launch.

Google also doesn't crawl everything at once. Large sites get visited on a budget, so deep pages that are many clicks from the homepage get seen less often. Keep the pages that matter close to the top of the site.

Step two: Google has to understand and store it

Once Googlebot fetches a page, Google renders it (runs the JavaScript, roughly the way a browser would), reads the content, and decides what the page is about. Then it stores that understanding in its index. Only indexed pages can appear in results.

Indexing fails quietly more often than people expect. Google's own Search Console reports a page as "Crawled, currently not indexed" when it fetched the page but decided not to keep it, which usually means the page looked thin, duplicated, or too similar to another page already in the index. Two product pages that differ by one word, a tag archive that lists the same posts as a category page, a location page copied from another city with the town name swapped: Google sees all of these as versions of the same thing and keeps one, or none.

What helps Google understand a page:

• A single clear topic per page, stated in the title tag, the main heading and the first paragraph, in words a searcher would use.
• Text that a crawler can read. Content loaded only after a click, hidden in an image, or rendered by JavaScript that fails on the server is content Google may never see.
• A canonical tag that names the page's preferred URL, so duplicates created by parameters, filters or trailing slashes all point back to one page.
• Structured data where it describes something real: an article, a product, a local business. It doesn't make a page rank, but it removes ambiguity about what the page is.

Step three: Google decides where to show it

Only now does ranking happen. For every search, Google pulls candidate pages from the index and orders them using a mix of signals. Google's own documentation groups them into a few families: whether the page's content matches the meaning of the query, whether the page and site are seen as trustworthy and authoritative for the topic (links and mentions from other sites are a large part of this), whether the page is usable (it loads quickly, works on a phone, isn't buried under interstitials), and the context of the searcher, including their location and language.

Two of those families are where you can move fastest:

Relevance to the query. Not keyword density, but whether the page actually answers what the searcher meant. A page targeting "best running shoes for flat feet" that opens with the history of running shoes loses to one that names three shoes and explains why in the first screen. Search intent is the whole game here.

Trust from the rest of the web. Google treats links and mentions from other sites as votes of confidence, weighted by how relevant and trusted the voting site is. This is slow to build and it is why a new site with better content can still sit behind an older one for months. The later chapters on authority cover how to earn it.

Ranking is also not a single list any more. Depending on the query, the results page mixes classic blue links with a map pack, shopping results, videos, and AI-generated overviews that cite a handful of pages. Each of those is its own contest with its own rules, which is why a local business and an online store need different playbooks.

Where sites quietly fail

After auditing enough sites a pattern shows up. The pages that don't rank are rarely bad pages. They are:

• Pages that exist but nothing links to, so Google visits them rarely or never.
• Pages Google fetched but refused to index because they duplicate something else on the site.
• Pages that target a search their content doesn't actually satisfy.
• Good pages on a site the rest of the web has never mentioned.

Each of those has a fix, and each fix belongs to a different chapter of this academy. The point of starting here is so that when a page isn't ranking, you check the pipeline before you rewrite the words.

What to take away

• A page must be found, indexed and then ranked, in that order, and most ranking problems are actually finding or indexing problems.
• Internal links and your sitemap decide whether Google sees a page at all; robots.txt and noindex can switch it off by accident.
• Google indexes pages it can read and understand as distinct; duplicates and thin variations get dropped.
• Ranking rewards pages that answer the searcher's real question on a site other sites trust, and it's a different contest for each result type.

Next

Now that you know how Google decides, see how AI assistants make a similar decision with different sources in How AI search picks its sources.