How to Get Your Site on Google: Find the Stage Where It Drops Out

You published the page. You put real care into the copy. You added the site to Search Console, submitted the sitemap, and waited.

It never showed up.

Nobody told you why. No warning came, no receipt, nothing at all. You stared at the silence and reached the usual conclusion: Google did not like your content.

Google has no opinion about your page. Between the publish button and the searcher’s screen there is a production line. It has stages, each with a name and a place in the order. And every stage has a side exit where a page can disappear without a sound.

Almost every article on this subject hands you the same to-do list. Set up, submit, write, wait. You have done all of that. The question left unanswered is a different one: why it did not work. This piece swaps the list for a diagnosis.

The kitchen you never saw

Picture a restaurant with every table full. You know the dining room well: the table, the menu, the plate that arrives or never does. The kitchen sits behind a swinging door, and nobody ever let you walk through it. That is where your dish was plated, checked, set aside, or thrown out. The dining room only shows you the outcome.

Opening that door comes with a first shock. When someone asks Google a question, the system checks a copy of the internet. That copy was built before the question existed. It was processed, cataloged, and filed ahead of time.

Your page does not compete on the internet. It competes in the copy that was made of it.

Imagine a library with a fleet of roving photocopiers. They head out across the city, walk into every house, and copy every written page. The copies come back to the building. Inside, a team reads each sheet, underlines it, sorts it, and files it. When you walk up to the desk with a question, the librarian turns around and checks the files.

If your sheet was never copied, or was filed in the wrong place, the librarian cannot find you. Your writing may be excellent. It is simply missing from the files being checked.

The four doors, and which one is yours

Google itself describes three stages in its public documentation. Crawling: automated programs download text, images, and videos from the pages they find. Indexing: the system analyzes what it downloaded and stores it in a huge database, the index. Serving search results: given a query, the system returns what it considers relevant. ¹

On the same page, Google states it outright: “not all pages make it through each stage.” ¹ That is the side exit, confirmed by the people who run the kitchen.

I work with four stages, and I flag the difference before using the fourth. I split the third stage in two. On one side sits retrieval and ranking: the system picks the candidates and puts them in order. On the other sits presentation: whatever survives becomes the screen the searcher reads.

The three stages are Google’s. The split into four is mine. ² I make that split because the third stage is where a page vanishes without you seeing it. The diagnosis depends on looking inside it.

I am careful with this distinction, because it holds for everything you will read about search. Anyone who blends Google’s documentation with personal interpretation and presents it all as official is inflating the evidence in front of you. Now walk through the doors in order, looking for yours.

Door 1, crawling: did the crawler reach the page, and did it find the content?

This is the most basic door, and the one most people skip because it seems too obvious to be the problem. The crawler needs a path to the page. Google discovers new URLs through links from pages it already knows, or through the list you submit, the sitemap. ¹

A page with no links pointing to it, and missing from the sitemap, may never be found. Even with a path, the visit can fail: server problems, network issues, or a robots.txt rule blocking access. ¹ In those cases, the page never made it into the building. Crawling has its own pace and priorities, and understanding that comes before blaming the copy.

This door has a second half, and it quietly takes down good sites. After the download, Google opens the page in a browser with no screen and runs the scripts, as your own browser does. ^{1 3} This is called rendering.

Every page that responds normally, with a 200 status code, goes into the rendering queue, JavaScript or not. The wait can be a few seconds, or longer. The page gets its turn once Google’s resources allow. ³

The queue is the same for everyone. What changes is what the page hands over before it gets there. A page with its text ready in the HTML shows the content on the first read. A page assembled by script arrives nearly empty, and its content only appears after rendering. ³

Now put this in an online store. The product description loads by script, a few seconds after the screen opens. For the shopper, the experience is great. For the machine’s first read, that page is a bare skeleton, and its content waits its turn in the queue. The same documentation calls server-side rendering or pre-rendering “still a great idea,” because not all bots can run JavaScript. ³

Door 2, indexing: did the page make it into the files?

Once the copy is made, the filing work begins. Google analyzes the page’s text, images, and markup, and stores what it understood in the index. This is called indexing.

Being crawled does not guarantee a place in the files. Google writes it plainly: “Indexing isn’t guaranteed; not every page that Google processes will be indexed.” Among the common reasons, it lists three. Low-quality content, robots meta rules that disallow indexing, and a site design that makes indexing difficult. ¹

There is also a filter that catches many sites without the owner noticing. Google groups pages with similar content and picks the one that best represents the group. That page is the one that can appear in results. The others become alternate versions, which may be served in specific contexts, such as a search from a phone. ¹ The chosen page has a name: the canonical page. If you have three nearly identical pages, Google picks one, and it may not be the one you wanted.

And here lives the most misunderstood fact in all of search. Part of the judgment on your page happens at this point, and the result is stored. Google’s documentation says it in its own way. During indexing, Google collects signals about the page and its contents. Those signals “may be used in the next stage,” where the page is served in search results. ¹

The internal documentation that leaked in 2024 shows the same design from the inside. One of its modules stores quality signals per document, compressed to save indexing space. ⁴ Google confirmed the material was authentic in May 2024. ⁵

I need to be honest about the limits of what I just said. It is documented that signals are calculated in advance and stored. How much of the judgment happens in advance, nobody outside Google knows. ² That part is my own reading of the architecture.

The practical consequence is still a hard one. When your customer types the question, part of the verdict on your page is already sitting in the drawer. The optimization that counts happens before the search, while the files are being built.

Door 3, retrieval and ranking: did it become a candidate, and survive the reshuffle?

When the query comes in, the system searches the index for matching pages. Then it returns the ones it believes are the highest quality and most relevant. ¹ The list takes shape, and its order can still change after that.

Mike King, of iPullRank, analyzed the leaked documentation. He described re-ranking functions that run after the primary ranking algorithm. They can adjust a document’s score or change its position. ⁶ The same material holds a family of demotions: named attributes, each tied to a kind of problem. ^{4 6}

The names are there for anyone who wants to read them. One points to a mismatch between the link and its destination. Another points to navigation. Another, to a domain whose name is the exact keyword. Another carries the results page in its own name. ⁴

Here I will give you the pattern and hold back the detail, and I will tell you why. The document gives the attribute’s name and a short description, often a single technical line. ⁴ Almost everything circulating about what triggers each demotion is industry interpretation. King himself writes that nobody knows how these attributes are weighted in the scoring functions. ⁶

An attribute name does not define behavior. Anyone who hands you the exact criterion for a demotion is selling reverse engineering dressed up as documentation. When Google confirmed the material was authentic, it cautioned against assumptions based on out-of-context, outdated, or incomplete information. ⁵ That warning applies to the whole industry, and it applies to me.

Read this without being afraid to publish. My reading of these names, and I flag it as mine, points to a persistent pattern of poor experience. ² Promising in the link what the page does not deliver. Wearing out the people who arrive. Existing only to match a search term. The risk lives in the habit that keeps repeating.

Door 4, presentation: did it reach the screen of the person you wanted?

The last door is the one that most often fools people who test their own site. The screen changes from one person to the next. Google writes that relevance is determined by hundreds of factors. They could include the searcher’s location, language, and device. ¹

The elements on the results page also change with the query. In Google’s own example, a search for “bicycle repair shops” will likely show local results. A search for “modern bicycle” is more likely to show images. ¹

Think of the owner who searches for their own site on a phone, cannot find it, and panics. That screen is just one screen. Search Console Help warns that an indexed page is not guaranteed to show up in your own searches. Results are customized by search history, location, and many other variables. ⁷

The same help page finishes the thought. If Search Console says a page is indexed, “you can assume that it is indexed and eligible to appear in search results.” ⁷ When it is indexed and shows up for nobody, the documentation names three suspects. The content is irrelevant to what people search for, the quality is low, or a robots meta rule prevents serving. ¹

The ten-minute exercise that gives you the answer

You do not have to guess which door your page is stopping at. The first snapshot comes from one dashboard, and it costs nothing.

Open Google Search Console on your site’s property and go to the Pages report, under Indexing. Google’s help calls it the Page indexing report. It splits your site into two groups: indexed and not indexed. It also lists the reason for each exclusion, with the number of URLs per reason. ⁷

Do three things, in this order:

1. Write down how many pages are indexed and how many were left out.
2. Read the list of reasons for the pages not indexed, and write down the top three.
3. Write out by hand the ten most important pages of your business, the ones that sell. Paste each URL into the URL Inspection bar at the top of Search Console, and check whether it is indexed. ⁸

Item 3 is the most revealing. If a page that carries your revenue is not indexed, it fell at one of the first two doors. It never even competed for a position. Before you touch its copy, read the reason URL Inspection gives. That reason tells you whether the job is technical or about content.

One correction up front, because you will likely be tempted. You have probably seen someone search site:yourdomain.com on Google and count the results. That count is misleading. The official documentation for the operator says the list of URLs “is not always exhaustive.” It also says the operator does not necessarily return every URL that is indexed. ⁹

John Mueller, of Google, put it more bluntly in the Ask Googlebot video series. He said a site: query “is not meant to be complete, nor used for diagnostics purposes.” He also told anyone who wants a count of indexed pages to use Search Console. ¹⁰ The help page for the Pages report settles the matter. Its indexed and not indexed totals are “complete and accurate from Google’s perspective.” ⁷

The operator is still useful for a quick look, when you are hunting for one important page. For sites with fewer than 500 pages, Google’s own help suggests starting with searches like this. ⁷ For counts and reasons, use the report.

The question you ask from now on

The kitchen door stays open now, and it will not close again.

When one of your pages does not show up, you now have a better question. The old question was whether Google liked the content. The new one is at which stage you dropped out.

Did the crawler reach the page? Was the content in the HTML, or did it wait its turn in the rendering queue? Did the page make it into the index, or was another one chosen in its place? Did it become a candidate and survive the reshuffle? Did it land on the screen of the person you were trying to reach?

Each question has a different answer and a different job. Anyone who does not know the production line treats them all as one problem, and calls that strategy. The technical fronts of that line are covered in the technical SEO section, in the order you should audit them.

One last thing, which I would rather say now. Being indexed is the ticket. Ranking is the line. Being recognized as one specific thing in the world, with a name, attributes, and relationships, is another job with a section of its own. It starts with understanding how Google records companies, people, and places.

That work assumes your pages have made it through all four doors. Start with the Pages report today, with your list of ten pages in hand.