Chapter 3: Ranking

Last updated: November 1, 2026.

A page being indexed by Google doesn’t mean it will rank near the top in searches. The real competition starts after that.

The question Google is trying to answer is roughly this:

“Which page should I show first to the user who made this search?”

For example, a user searches for “best coffee machine.” Google’s index may have thousands or even millions of pages on this topic. Among them, Google tries to understand which one:

• Is more relevant to the search
• Better answers the user’s need
• Is more trustworthy
• Is more useful
• Offers a better user experience

And it ranks the results based on that.

So why does Google show one page above another?

This is exactly where the foundation of SEO begins.

Because there’s no single answer. Google weighs a large number of factors together when ranking. When we do SEO, we try to understand what those factors are, which ones matter how much and how they affect a site. As we figure them out and develop our site in the right direction, we move higher in the search results.

Among them are many factors like these:

• How relevant the content is to the search
• Whether it answers the user’s search intent
• The quality and originality of the content
• The authority of the site and the page
• Backlinks
• Internal linking
• Whether the page works properly from a technical standpoint
• Mobile friendliness and user experience
• How fresh the content is
• The language, country and location of the search

How does Google group ranking signals?

Google doesn’t share its ranking formula, but on its “How Search Works” page it describes the kinds of information it looks at in five main groups (Google):

When I look at this table, I see that it says the same thing as the list above. Google just groups them in its own words. The phrase “when everything else is roughly equal” in the usability row is especially important, I think. Site speed and mobile friendliness matter, but on their own they won’t push an irrelevant page to the top. I cover speed in Chapter 24, and why everyone doesn’t see the same rankings because of context and settings in Chapter 6.

Google’s ranking systems

Google also publishes a separate guide describing which ranking systems it uses (Google). It still doesn’t give the formula, but it shows which things it pays particular attention to. I think these are the ones worth knowing at the start:

• RankBrain, BERT and neural matching: AI systems that try to understand the meaning of the words in a query and how they relate to each other. They make it possible to find a relevant page even when the searched words don’t appear on it exactly.
• Passage ranking: Looks at individual sections of a page, not just the page as a whole, to judge relevance.
• Freshness systems: Bring fresher pages forward for queries where users expect newer content.
• Link analysis and PageRank: Look at how pages link to each other. Google says PageRank is still part of its core ranking systems.
• Deduplication: When there are many very similar pages, shows only the most relevant ones. This is the ranking side of the “I already have something like this” situation I described in the indexing chapter.
• Site diversity: Generally doesn’t show more than two results from the same site in the top results.
• Spam detection systems: Try to catch spam content and behavior with systems such as SpamBrain.

The guide also lists systems that no longer run separately. For example, the helpful content system became part of the core ranking systems in March 2024. Panda and Penguin had been folded into the core systems in the same way years earlier.

In SEO, the truths change too

One of the systems on this list reminds me of an old article of mine. In 2012 I wrote an article (in Turkish) on how domain names affect SEO. In it, I explained that exact match domains appeared more often in search results, and argued that choosing a memorable domain meant starting the race ahead. I used indir.com, a Turkish software download site, as an example: we launched the site at the end of December, and about 2.5 months later it reached the first page on Google.

Today, Google’s ranking systems include an exact match domain system. Google counts the words in a domain name as one of many factors it looks at when deciding whether content is relevant. But it uses this system to limit how much credit goes to domains bought just to match a search.

So something I saw as a strong advantage in 2012 doesn’t carry the same weight today. That’s worth keeping in mind as you learn about ranking factors: a signal that matters today may not carry the same weight tomorrow. That’s why it’s better to look at what Google is showing today than to rely on memorized rules.

Looking at a single metric

But there’s a common mistake here:

Reducing SEO to a single metric.

• “I have more backlinks, so why am I not number 1?”
• “My content is 3,000 words and my competitor’s is 1,000 words. Why is the competitor above me?”
• “My site speed is better, so why am I not moving up?”

Because Google doesn’t rank by looking at a single signal. Sometimes a shorter piece of content can rank above a longer one because it gives the answer the user is looking for more effectively. Sometimes a page with far fewer backlinks can get ahead of a strong site because it’s much more relevant to what’s being searched.

A very good example of this came from Yandex. When Yandex’s source code was leaked in January 2023, a list of 1,922 ranking factors came out of it. At the time, I wrote a long thread on X going through that list. Yandex isn’t Google. We can’t say those factors work the same way in Google. But it was a very good example of how a search engine doesn’t rank results by looking at a single signal.

That’s why in SEO, rather than asking “Which of my metrics is higher than my competitor’s?”, you need to ask “Why is Google showing my competitor ahead of me for this search?”

I think most of SEO is trying to find the answer to that question.

Google doesn’t give us the formula for its ranking algorithm. But by studying the search results, competitors, Search Console data, the site and the results of the changes we make, we can piece the clues together.

In a sense, we’re constantly doing observation → hypothesis → implementation → measurement. And as we figure out the right things, we move up.

Observation, hypothesis, implementation, measurement: a coffee machine example

Let’s think through this cycle with an example. Say an e-commerce site’s coffee machine category page is stuck on page two for the search “best coffee machine.”

• Observation: We search for the keyword on Google and look at the first page. The top positions are mostly guides comparing models. There are almost no category pages.
• Hypothesis: Google thinks the user wants to compare options for this search. Our category page, on the other hand, lists products directly. The problem may not be backlinks or speed, but the page type.
• Implementation: We create a guide that explains the types of coffee machines, which one suits which user and how the models differ. We link from the guide to the category page.
• Measurement: A few weeks later, we look at the guide’s impressions, clicks and average position for this query in Search Console. If we didn’t get what we expected, we go back to observation.

Our hypothesis in this example could turn out to be wrong. What matters is basing the change on a hypothesis and measuring the result. That way we know what to look at in the next step. I cover how to study the search results page in detail in Chapter 5.

Why do rankings fluctuate?

Rankings also aren’t fixed. A page that’s in position 3 today can drop to position 5 tomorrow. Because competitors are working too, new content is being published, existing content is being updated and Google’s ranking systems are constantly changing.

That’s why SEO isn’t something you do once and leave.

Small daily movements are normal. A page moving up or down a position or two over a few days usually doesn’t indicate a problem. Bigger changes, on the other hand, usually come with Google’s core updates.

A few times a year, Google makes broad updates to its search systems. These updates don’t target a specific site or page. They change how Google evaluates content in general. According to Google’s own explanation, losing rankings after an update doesn’t always mean something is wrong with the site. Sometimes other pages now simply answer the user better (Google).

Google’s recommendations on this, in short:

• Wait for the update to finish before starting your analysis.
• After the update is done, wait at least a week and compare that period in Search Console with a week before the update.
• If there’s a small drop, don’t make big changes to pages that are performing well.
• If there’s a big, lasting drop, review the content from the user’s point of view, starting with the most affected pages.
• Know that the effect of improvements can sometimes take months.

I think the most important point here is not to make decisions in a panic. When rankings drop, the first job is to see when the drop started and which pages and queries it affects. Changes made without seeing this usually make it even harder to understand the problem.

Practice

• Search for your 3 most important keywords on Google. How are the pages in the top 3 positions different from yours?
• If you have an explanation based on a single metric, like “My competitor has more backlinks,” check whether there’s a page above you for the same search with fewer backlinks than you.
• Write down the hypothesis behind your most recent SEO change and the date you’ll measure it. Check the result in Search Console.
• Look at the last 16 months of the performance chart in Search Console. Do any big drops or rises line up with a core update Google announced?
• When rankings drop, first list which pages and queries were affected, then make changes.
• Do you have very similar pages on your site targeting the same search? In cases like this, Google usually shows only the most relevant one.

Previous: Chapter 2: Indexing · Contents · Next: Chapter 4: Search Intent

This chapter was first published in Turkish on X (August 31, 2026).