Does CTR Affect SEO Rankings? What NavBoost Shows and How to Test It

Clicks can affect Google rankings, but not as a simple dial you turn up. Google says it uses data on how people interact with results to judge relevance. Court testimony and a 2024 leak show a system called NavBoost that learns from clicks. Nobody outside Google knows how much CTR counts, and Google filters out clicks that look fake.

CTR means click-through rate. It is the share of people who see your result and click it. If 1,000 people see your page in Google and 50 click, your CTR is 5 percent. Our guide on how to calculate click-through rate has more worked examples, and our guide to organic CTR by position has the benchmarks.

Does CTR affect SEO rankings? The short answer

Clicks are one ranking input among many, and their weight is not public. Three sources point the same way.

For years, Google staff said clicks were not used to rank pages. SparkToro’s write-up of the leak lists several of those denials. The three sources above changed the debate.

What Google says about clicks

Google’s own page on ranking says it uses “aggregated and anonymized interaction data” to check whether results are relevant. It turns that data into signals for its machine learning systems. You can read it on Google’s How Search works page.

Interaction data is what searchers do on the results page, such as which result they click. Google does not list which parts it uses. It does not say how much they count either.

What the antitrust trial showed: NavBoost

In October 2023, Google search executive Pandu Nayak testified in the US government’s antitrust case against Google. He said NavBoost is “one of the important signals that we have.” Semrush’s NavBoost explainer sums up the testimony and the later leak.

NavBoost uses past clicks on search results to help rank pages. Reports on the trial and the leak say it:

• keeps about 13 months of click data;
• splits that data by the searcher’s location and device;
• treats some clicks as good signs and others as bad ones.

So the same page can get a different lift on a phone in Chicago than on a laptop in London. That is one reason rankings differ by city and device.

What the 2024 leak showed

In May 2024, more than 2,500 pages of Google’s internal API docs became public. Rand Fishkin of SparkToro published the first look at them. The docs list 14,014 attributes. Several deal with clicks:

• goodClicks and badClicks: clicks Google seems to sort into good and bad results for the searcher.
• lastLongestClicks: likely the last result a searcher clicked and stayed on.
• Squashed and unsquashed clicks: raw clicks versus clicks after a filter. Fishkin’s source said this helps fight click spam, both manual and automated.

The leak shows these fields exist. It does not show how much each one weighs, or if Google still uses it. Semrush notes that some of the meanings are guesses.

What this means for your pages

Three practical points follow from the evidence.

1. Clicks only count where people see you. NavBoost learns from clicks on results people see. A page on page five gets almost no clicks to learn from.
2. A better title pays off either way. A higher CTR at the same position means more visits today, even if rankings never move.
3. Fake clicks are what Google filters. The squashed click fields exist to discount clicks that look manipulated. Buying clicks to rank is a bet against that filter.

Blocks above your result, like AI Overviews and ads, also pull clicks away. Our post on SERP features shows what each block does to CTR.

Why most CTR tests prove nothing

A common story goes like this. Someone sends extra clicks to a page. Two weeks later it moves up three spots, and they call it proof.

But rankings move on their own all the time. New content, competitor changes, Google updates and demand all shift them. Our post on why Google rankings fluctuate covers the causes.

Without a control group, you cannot tell if the clicks worked or if the page would have moved anyway. A control group is a set of similar keywords that get no extra clicks. If the test keywords move and the control keywords stay put, you have a signal. If both move together, you saw normal noise.

How to design a CTR test you can trust

Step 1: pick pages that already rank

Choose keywords where your page sits around positions 4 to 15. Clicks there look normal, and there is room to move. Check the current position from a fixed city and device, signed out. Our guide on checking your Google rankings shows four ways to do it.

Step 2: split keywords into two groups

Take 10 to 20 similar keywords from the same site. Split them at random. The test group gets the clicks. The control group gets nothing new. Both groups should have the same intent and similar positions.

Step 3: write the success rule first

Decide what counts as a result before you start. For example, the test group must gain at least three average positions more than the control group. It must also hold that gain for two weeks after the clicks stop. Writing this down first stops you from explaining away a weak result later.

Step 4: measure with first-party data

Track impressions, clicks, CTR and average position for each keyword in Google Search Console. Check rank for both groups from the same location and device each week. Write down every change you make to the pages.

Step 5: run it for weeks, then report what you found

Give the test weeks, not days, since NavBoost is reported to use months of data. Many tests end with “no clear change.” That is a real result. It tells you to spend the money on titles, content and links.

To run the click side, you can set up the test in Serplify step by step. That guide covers keyword groups, daily volume, the control group and stop rules.

What CTR testing cannot do

• It cannot promise rankings. Clicks are one input among many. Anyone who promises a ranking from clicks is guessing.
• It does not replace a good page. The surest way to earn good clicks is a page that answers the search.
• It carries risk. Google’s spam policies count attempts to manipulate rankings as spam. Sites that break them can rank lower or drop out of results. The same page lists automated queries to Google as machine-generated traffic.

So test only on your own pages, keep the volume small, and treat the result as one data point.

Where Serplify fits

Serplify sells CTR test clicks for the click step. Each click is an automated session in a real Chrome browser on a residential or mobile connection. It searches Google for your keyword, finds your page, clicks it, and stays for the time you set. These are test clicks, not customers.

Clicks cost $0.04 each, and only delivered clicks are billed. Click projects start after a first top-up of $25. The $1 signup credit covers the SERP API, which you can use to check rank for both groups at $0.005 per SERP. If you are comparing click tools, our list of SerpClix alternatives sets out prices and click types side by side.