Access
Robots rules, canonicals, redirect chains, rendered content and crawl budget. This is the stage where a site can lose everything while looking perfectly designed to a human.
Google what is SEO is a question about how one company's search engine decides what to show, and it has a concrete answer: SEO is the work of getting a page crawled, indexed, ranked and shown for a query it can satisfy. Every tactic you have read about belongs to one of those four stages. Nothing enters the index because you asked nicely, and nothing stays on top because you repeated a phrase.
In Google's terms, SEO means making a page crawlable, indexable, relevant and clickable. Google sends crawlers to fetch the URL, stores and understands the rendered content in its index, ranks the indexed pages against a query, and shows the result as a title, snippet and link. SEO is influencing those four steps, in that order.
SEO is not one system. Baidu weighs different signals for a Chinese audience, Bing behaves differently again, and marketplaces run their own ranking inside their own walls. Asking about Google narrows the answer to one engine with one documented process, which is why the four-stage model is useful: Google has published enough about crawling, indexing and ranking that the stages can be described without guessing.
The label also sets the audience. A page written for Google's English-language results competes with the entire English web. A page written for a regional engine competes with a smaller set and often rewards different things, such as local citations or a locally hosted site. None of that changes the sequence, only the weight of each stage.
One caveat matters before the table. Google does not publish its ranking formula, and any supplier who claims to know it is inventing. What is public is the pipeline: a crawler, an index, a set of ranking systems, and a results page that assembles titles and snippets. Work on the pipeline is provable. Work on a secret formula is not.
A page travels through these stages in sequence, and it can fail at any one of them. Fixing stage three when a page never cleared stage one is the single most common waste of budget in this channel.
| Stage | What Google does | What it needs from you |
|---|---|---|
| CrawlStage 01 | Discovers URLs from links, sitemaps and past visits, then fetches them with a crawler. | Reachable URLs, no accidental blocks, sane internal linking, and a sitemap that matches reality. |
| IndexStage 02 | Renders the page, extracts the text and stores it with a canonical decision. | Server-rendered content where it matters, self-referencing canonicals, no duplicate competing versions. |
| RankStage 03 | Compares indexed pages against a query and orders the candidates. | One intent per page, depth that matches the result set, and evidence that the page satisfies the query. |
| ShowStage 04 | Builds the result: title, snippet, sitelinks, rich features. | Titles and descriptions that promise something specific, plus structured data where it applies. |
Notice that only stage three is about being "the best". Stages one, two and four are about being legible. A mediocre page that is fully legible will out-earn an excellent page that is invisible, every time.
Translate the pipeline into four kinds of work. Each has an owner, a deliverable and a way to check it landed.
Robots rules, canonicals, redirect chains, rendered content and crawl budget. This is the stage where a site can lose everything while looking perfectly designed to a human.
Read the current result set before writing. The format Google already rewards for a query is a brief, not an opinion: guides for guides, comparison tables for comparisons.
Speed, layout stability and intrusive interstitials affect whether the visit survives the first seconds. They rarely rescue a bad answer, but they routinely sink a good one.
Search Console impressions, click rate at a position, and the enquiries that name a page. Without this you cannot tell which stage is failing, only that the number moved.
A sitemap is a suggestion, not a command. It helps discovery and tells Google which URLs matter. Indexing is still a decision the engine makes after fetching and rendering the page.
Length is not a signal. Coverage of the questions behind a query is. A 600-word page that answers four sub-questions beats a 2,000-word page that answers one and repeats it.
Most sudden drops trace back to a template change, a redirect, a canonical mistake or a competitor publishing a better answer. Check the pipeline before blaming an update.
Send a URL. We will tell you whether it is a crawl problem, an index problem, a rank problem or a show problem, and what to fix first.
A fix is not finished when it ships. It is finished when the stage it targeted shows a change, which is why we sequence the work instead of batching it.
Check the response code, robots rules and rendered HTML. If a crawler cannot see the main copy, nothing downstream matters.
Look for a competing canonical, a duplicate parameter version, or a noindex inherited from a template.
Match the depth and shape the current results use, then answer the first question in the first paragraph.
Title and description state the outcome. Structured data is added only where it is honest.
Impressions for the cluster, click rate at a stable position, enquiries by page. One sheet, reviewed monthly.
We start at stage one because that is where the cheapest wins hide. Crawl access, canonical logic and internal links are unglamorous, measurable and usually neglected for years. Then the intent map: one page per question, merged duplicates, redirects that point at the survivor. Then the pages, written to answer the query before selling anything. Then the snippet work, so the result that appears is worth clicking.
For an English-language site aimed at Google's results, the scope is set out under Whats Is SEO?. If you want the definition without the pipeline, Google SEO Services covers it in plain language.
A laboratory equipment supplier had a category page that ranked on page two for its main product term and never moved. The instinct was to rewrite the copy. The pipeline said otherwise: the page's specification table was injected by a script that the crawler skipped, so the indexed version contained a headline and a phone number.
The fix was server rendering, not prose. Once the table was in the HTML, the page was re-crawled within days and the indexed version finally matched what buyers saw. Only then did rewriting the introduction and the title produce a change in click rate. Had the team started with copy, they would have paid twice for the same lesson.
Google publishes guidance on how search works, on crawling and indexing, and on creating helpful content. It does not publish a formula. Treat its documentation as the description of the pipeline and treat everything else as inference.
Crawling can happen in hours or weeks depending on the site's authority and update frequency. Indexing follows the crawl. Ranking changes are evaluated over repeated queries, so real movement is usually measured in weeks or months, not days.
You do not have to, but you should. A sitemap and a verified property in Search Console make discovery and diagnosis faster. Neither guarantees indexing.
In practice it is the same discipline aimed at one engine. The stages are identical; the weighting and the documentation differ. Businesses that sell mainly in one market usually mean Google when they say SEO.
That is a stage four problem. The page is eligible and ordered, but the title and description are not competitive, or the position is low enough that few people scroll to it. Rewrite the promise before rewriting the page.
Sometimes, if the site is already crawlable and indexable. On a site where content is rendered by script or canonicals are misconfigured, editorial work alone will not surface. Check stage one and two before buying content.
One URL and the query it should win. We will reply with the stage, the evidence, and the smallest change that moves it.