How Google ranking works, explained without jargon

Most people who ask "how does Google ranking work?" are really asking one of two much narrower questions: why isn't my site showing up, or what do I have to do to get it higher. Those are different problems with different fixes, and you cannot tell them apart until you understand that Google does four separate jobs, in order, and any one of them can be where your problem lives.

The four jobs are: finding your pages, reading them, filing them, and choosing between them. In the industry those are called crawling, rendering, indexing and ranking. People blur all four together and call the whole thing "SEO", which is why so much advice misses. Telling someone to write better content when their pages are not being filed at all is like recommending a better window display to a shop with the shutters down.

A rough analogy, which I will not stretch too far: imagine a very large second-hand bookshop. Someone walks the streets collecting books nobody has catalogued yet. Someone else opens each book and actually reads it, including the pages that only appear when you unfold a flap. Someone writes an index card describing it and files it. And when a customer walks up to the counter and asks for "a short book about beekeeping for a nervous beginner", someone answers, on the spot, using the cards. Four people, four jobs. Your book can fail at any desk.

The four stages, and what goes wrong at each

| Stage | What Google is doing | What goes wrong here | How you would notice | |---|---|---|---| | Crawling | Sending an automated program to fetch your pages, following links it already knows about and URLs you submit | No links point at the page, the page is blocked in robots.txt, the server is slow or intermittently down, the site is behind a login | The page has never appeared in Google at all, and a site: search for your domain does not show it | | Rendering | Running the page the way a browser would, so text that only appears after scripts run is still seen | Content is loaded by JavaScript that fails, times out, or needs an interaction to fire; text lives inside an image | The page is indexed but Google's cached or tested version is nearly empty compared with what you see | | Indexing | Working out what the page is about and storing it, deciding whether it is worth storing at all, and picking one version of duplicate pages | Near-duplicate pages compete, thin pages are crawled but not stored, the wrong version of a URL is chosen as the canonical one | Search Console reports the page as discovered or crawled but not indexed, or a different URL of yours shows instead | | Ranking | For one specific query, ordering the stored pages it considers eligible | Nothing is broken — you are simply being beaten by pages that answer the query better, are trusted more, or have been around longer | The page ranks, just not high enough, and its position moves around |

Read that table as a diagnostic sequence rather than a description. If you are not being crawled, nothing you do to your writing matters. If you are indexed but sitting on page four, crawling advice is a waste of your afternoon. The single most useful habit for a business owner is to find out which desk you are stuck at before spending anything.

The first three stages are largely mechanical and largely fixable. Get the server reliable, make the important text present in the raw HTML rather than assembled by scripts afterwards, do not accidentally block your own pages, and give every page at least one link from somewhere Google already visits. That is not glamorous work and it does not need a retainer. It also does not, by itself, get you ranked — it only gets you eligible.

Ranking is per-query, not a score for your site

This is the idea that changes how people think, so it is worth stating flatly: there is no single number attached to your website that determines where you appear. Ranking is computed per query. The same page can be first for one phrase, thirtieth for a near-identical phrase, and absent for a third.

A concrete case. A two-person plumbing business in one town publishes a page about emergency boiler repairs. That page might sit first for "emergency boiler repair" plus the town name, because there are four local competitors and the page is specific, has an address, and loads fast. The same page might sit nowhere for "boiler making strange noise", because that query attracts national brands, diagnostic guides and video results, and the page does not actually answer it. Nothing about the business changed between those two queries. What changed was the question, and therefore the set of pages Google considers reasonable answers.

This is also why "we rank number one" is close to meaningless as a claim unless the query is stated. Ranking first for your own company name is normal and tells you nothing. Ranking first for the phrase people type when they have money in hand and no supplier is the thing that is worth money.

Practically, it means your unit of work is the query, not the site. Before you write anything, you need a defensible view of which questions your buyers actually type — which is a research problem, and one you can do without spending anything, as covered in keyword research without a paid subscription. Guessing at this stage is where most wasted effort starts.

Relevance, quality and usability are three different questions

When Google orders the eligible pages, it is weighing several distinct things that people mash together into "quality". Separating them helps because the fixes are different.

Relevance is: does this page address this specific query? A page can be excellent and irrelevant. A long, careful guide to commercial boiler servicing is not a good answer for someone who typed a domestic query at eleven at night. Relevance failures are usually failures of intent — you wrote the page you wanted to write rather than the page that answers the question asked.

Quality, in the sense Google's documentation describes, is roughly: is this a source worth showing? Google's own guidance talks about content that demonstrates experience, expertise, authoritativeness and trustworthiness, and about signals from other sites pointing at yours. That second part is where links come in — a link from another site functions partly as a vote, and how much a given vote is worth varies enormously, which is the subject of how backlinks actually work and what one is worth. Quality is the slowest of the three to change and the hardest to fake, which is exactly why it carries weight.

Usability is: once someone lands, can they use the thing? Loading speed, layout that does not jump around while you read, a page that works on a phone, no interstitial covering the content you came for. Google has confirmed page experience signals matter, while also saying repeatedly that great content on a slow page can still outrank mediocre content on a fast one. Usability is a tie-breaker and a floor, not a lever.

The trap is that usability is the easiest of the three to measure, so it gets disproportionate attention. Tools produce a score out of a hundred, the score feels like progress, and it is possible to spend months moving that number while remaining irrelevant to every query that matters.

Why two people searching the same thing see different results

There is no single set of ten blue links. Results vary, and it is not a conspiracy.

Location is the biggest factor for anything with a local dimension. Search for "accountant" from two different towns and you get two different first pages, largely built from a different map result set. Device matters: mobile and desktop results differ, partly because the eligible pages differ and partly because the layout of the results page does. Language and country settings change the pool entirely. Time matters for anything news-adjacent or seasonal. And some personalisation exists based on your own recent activity, though it is generally weaker than people assume.

Two consequences for a business owner. First, checking your own rankings by searching from your own office, on your own logged-in browser, in the town where your business is registered, will flatter you. You are the most likely person in the world to be shown your own site. Second, a rank tracking tool reporting "position 6" is reporting one sample from one configuration; it is a useful trend line and a poor absolute truth.

What Google has confirmed, and what the industry has inferred

This distinction is worth internalising because a large amount of confident advice sits on the second pile while sounding like the first.

Google publishes documentation about how Search works, publishes spam policies describing what will get you penalised, and its representatives answer questions publicly. From that, some things are genuinely stated. Google says there is no preferred word count and no ideal keyword density. Google's spam policies state plainly that buying or selling links for ranking purposes is link spam. Google has confirmed that links are used as a signal, that page experience is considered, and that its systems aim to reward content made for people rather than for search engines.

Almost everything else you read — the weighting of any particular factor, the existence of a specific threshold, how long a penalty lasts, whether a given change caused a given movement — is inference. Some of it is careful inference from large-scale observation by people who have watched thousands of sites. Some of it is one person's site moving and a theory attached afterwards. Both are presented in the same confident tone.

A practical filter: when you read a ranking claim, ask what would have to be true for the author to know it, and whether they could have known it that way. "Google confirmed X in their documentation" is checkable. "Google's algorithm gives 20% weight to Y" is not something anyone outside Google can know, and the specificity is the tell.

A short diagnostic for your own site

Work down this list in order. Stop at the first stage that fails, and fix that.

1. Search for your exact page URL in Google. If it does not come back, it is not indexed. Go to step 2. If it does, skip to step 4.
2. Fetch the page in a browser with JavaScript disabled, or view the raw source. If the main text is not there, you have a rendering problem, and no amount of writing will fix it.
3. Check that the page is linked from somewhere on your own site that Google already visits, and that your robots file and page-level directives are not excluding it. A page linked from nothing and submitted nowhere is a book on the floor of the stockroom.
4. Indexed but not ranking: search the exact query you care about, in a private window, from a device in your target area. Look at the pages beating you and ask honestly what they have that you do not — is it specificity, is it that they are five years old, is it that twenty other sites reference them?
5. If the honest answer is "they have been established for years and dozens of sites link to them", you are not in a content problem. You are in a time-and-authority problem, and the realistic shape of that is set out in how to rank a brand new website, a realistic timeline.

Step 5 is where most small businesses actually land, and it is the least comfortable answer because it cannot be solved this month.

Common misconceptions worth dropping

Submitting your site to Google is not a ranking action. It helps with discovery, once. It does nothing for position.

Meta keywords are not used. They have not been for a very long time.

There is no "SEO score" that Google holds. Tools that produce one are producing their own opinion, which is fine as a checklist and misleading as a target.

Publishing frequently does not, in itself, raise anything. Fresh content matters for queries where freshness matters. For "what size boiler for a three-bed house", a good page from three years ago is still a good page.

Paying Google Ads does not affect your organic positions. They are separate systems. Ads buy placement in the advertising slots and stop the moment you stop paying — the difference in economics between renting and building is worked through in paid versus organic, an honest cost comparison.

Questions people actually ask

How long before I see anything? For a new site competing on anything commercial, longer than you want. What the available evidence supports is set out in how long SEO takes and what the evidence supports; the short version is that indexing can happen in days and competitive ranking usually cannot.

Can I pay Google to rank higher organically? No. You can pay for ad placements, which are labelled as ads. Organic positions are not for sale from Google, and buying links from other sites to influence them is described in Google's spam policies as link spam.

Does my site need to be big? No. It needs the specific page to be the best available answer to the specific query, and to be trusted enough to be believed. A six-page site can beat a six-hundred-page site on a narrow query and routinely does.

My rankings dropped overnight. What happened? Usually one of: a broad algorithm update, a technical change on your side, or competitors improving. Check whether your pages are still indexed first — an accidental noindex or a blocked robots file produces a drop that looks like a penalty and is not one.

Where a paid marketplace fits, and where it does not

We sell paid visibility, so treat what follows with the scepticism it deserves — but the honest position is narrower than you might expect us to argue.

Nothing on this page is a route around the four stages. If you need to be found for "boiler making strange noise" at eleven at night, you need to earn a position in Google's results, and that means relevance, quality and time. A listing on a marketplace like ours is a different thing entirely: a position on a public leaderboard, held because it is paid for, with the amount visible to anyone who looks. It is useful when there is a specific audience that reads that leaderboard, and it is honest about what it is — you can see every position and every figure behind it on the public rankings.

It is the wrong purchase if what you actually need is to be findable by strangers typing questions into Google. It does not build the thing that makes organic ranking possible, and it is not a shortcut to it. We have written the limits down properly in what pay-to-rank cannot do, and if after reading this the honest conclusion is that your money is better spent on three genuinely good pages and eighteen months of patience, that is a conclusion we are content for you to reach.