Check this page with an assistantOpens a chat asking it to summarise this article and name the evidence behind each claim.
Claude opens with the prompt on your clipboard: Anthropic does not support prefilled prompts on the web, and we would rather copy it than ship a button that drops it.
One Pipeline, Seven Places to Get Stuck
Google documents three stages: crawling, indexing, and serving results. Splitting those into seven is more useful diagnostically, because each substage fails in a different way and the symptom from outside is always the same: your page is not there.
That identical symptom is why so much SEO effort lands in the wrong place. A page missing because nothing has ever told Google it exists and a page missing because it is outranked look the same in a report. One is fixed by a sitemap and a link; the other by writing something better. Choosing wrong costs a quarter.
The stages run in order, which gives you a diagnostic sequence. There is no point improving content on a page that was never crawled, and no point checking your sitemap for a page that is indexed and sitting at position forty. Work forward from discovery and stop at the first stage that fails.
The Stages, and How Each One Fails
Read this table as a checklist rather than an explanation. For each of your problem pages, identify which row you are stuck on, and the right work becomes obvious.
| Stage | What happens | Failure and fix |
|---|---|---|
| Discovery | Google learns a URL exists, via links or sitemaps | "URL is unknown": submit a sitemap, link to the page, per the first 90 days |
| Crawling | The page is requested and the response read | Blocked by robots, server errors, or budget spent elsewhere |
| Rendering | JavaScript executes to produce final content | Content only exists after JS; see JavaScript SEO |
| Indexing | Google decides whether to store and understand the page | "Crawled, currently not indexed": thin or duplicate |
| Canonical selection | One URL is chosen to represent a group of similar pages | The wrong URL appears; see duplicate content |
| Ranking | Eligible pages are ordered for a specific query | Indexed and invisible: usually intent mismatch |
| Serving | The result is assembled, possibly inside a generated answer | Ranking without clicks; see zero-click search |
The indexing row contains the fact that surprises people most. Google states directly that not every page it crawls will be indexed. Crawled is not indexed, indexed is not ranked, and ranked is not clicked. Four different states, four different problems, and the same empty analytics report for all of them.
Running the Diagnosis
The URL Inspection tool answers most of this in thirty seconds per page, and almost nobody uses it before forming a theory. Its coverage state tells you which stage you are at, which tells you what to do.
- Inspect the URL and read the coverage state.
Unknown, discovered but not crawled, crawled but not indexed, or indexed. Those four states map directly onto the first four rows of the table and eliminate most possibilities immediately.
- If indexed, check the canonical Google chose.
The inspection shows both the canonical you declared and the one Google selected. When they differ, your page is being represented by another URL, and that is the problem rather than anything about content quality.
- If indexed and correct, look at impressions.
Zero impressions means you are not being shown for anything, which is a relevance problem. Impressions without clicks means you are being shown and passed over, which is a snippet or intent problem.
- Compare the rendered page to the source.
View the rendered HTML in the inspection tool. If your main content is absent from what Google fetched, you have a rendering dependency regardless of what the page looks like in your browser.
- Only then consider the content itself.
Content quality is the right answer to exactly one of these stages. Reaching for it first is why so many rewrites change nothing: the page was never in the competition to begin with.
Where AI Surfaces Fit
They sit at the serving stage, drawing on the same index. There is no separate AI crawler for Google's generative features and no separate AI index. Everything upstream, discovery through ranking, works the same way and remains the prerequisite.
This is the most useful thing to understand about AI visibility, and it removes a great deal of noise. A page that is not indexed cannot be quoted in a generated answer, because there is nothing to quote from. No special file, markup, or optimisation substitutes for being in the index, which is why the levers that work are the ordinary ones.
What does change at the serving stage is the unit being selected. Generated answers assemble passages from multiple documents rather than presenting a ranked list, and a single question may be decomposed into many searches first, as described in query fan-out. That affects how you structure a page, not whether the pipeline before it applies.
Why This Explains the Timescales
"How long does SEO take" has no single answer because each stage has its own clock. Discovery can be days or never. Indexing can follow immediately or not at all. Competitive ranking settles over months as evidence accumulates.
Knowing which clock you are on makes waiting rational rather than anxious. A page indexed within a day and sitting at position thirty after three weeks is behaving normally and needs time and links, not another rewrite. A page still unknown to Google after a month is not waiting for anything; something is broken and no amount of patience fixes it.
It also explains why new sites are slow in a way that is not mysterious. Every stage has to happen for the first time, and the final one requires evidence that does not exist yet. That is the sequence set out in the first 90 days, and the expectation-setting version is in how long SEO takes for a small business.
Questions People Ask About How Search Works
- What are the stages of Google Search?
Google documents three: crawling, indexing, and serving results. In practice it helps to split them further into discovery, crawling, rendering, indexing, canonical selection, ranking, and serving, because each of those can fail independently and each failure has a different fix.
- Does Google index everything it crawls?
No, and this is the stage people are most surprised by. Google states plainly that not every page it crawls will be indexed. Pages can be crawled and then not selected because they duplicate something else, because the content is thin, or simply because the system did not judge them worth storing. Crawled is not the same as indexed.
- Why is my page indexed but not ranking?
Because indexing and ranking are separate stages. Being in the index makes you eligible to be considered; ranking is a competition against everything else eligible for that query. A page that is indexed and invisible is usually losing on relevance or intent match rather than suffering a technical problem.
- Does Google render JavaScript?
Yes, but as a separate step that can happen later than the initial crawl. Content that only exists after client-side JavaScript executes depends on that rendering step completing, which introduces delay and a further point of failure. Content present in the initial HTML response avoids the dependency entirely.
- How long does the whole process take?
There is no fixed answer, and Google does not commit to one. Discovery can take days to weeks on a new site with no links or sitemap, indexing can follow quickly or not at all, and competitive ranking takes months to settle. Each stage has its own timescale, which is why 'how long does SEO take' has no single number.
