How Are Click-Through Rate, Dwell Time, and Search Rankings Related? Which Evidence Still Falls Short of Establishing Causation?

Click-through rate, the time between a click and a return to search results, and engagement metrics such as time on page and bounces all feature in discussions of search rankings. But being studied or changing alongside rankings does not mean that improving a metric will improve rankings. The supplied material includes research leads on click data and metrics potentially related to Navboost, but it does not establish each metric’s role in rankings—much less justify treating analytics values directly as search engine ranking signals.

First, distinguish behavior on search results from on-site engagement

These metrics have different starting and ending points. They are not interchangeable simply because they all relate to “engagement”:

• Click-through rate (CTR):Measures the proportion of search result impressions that lead to clicks, not simply the number of clicks.
• Dwell time:The time between a user clicking a search result to enter a webpage and returning to the search results page.
• Average time on page:Describes how long users spend on a page on average; returning to search results is not a required endpoint.
• Bounce rate:Harris’s 2018 article used “leaving without viewing a second page” to explain a common understanding at the time, while noting that different analytics tools may define it differently. This should not be applied as a universal definition for today’s tools.

For example, knowing how long a visitor spent on an article does not tell you when they returned to search results. Time on page and dwell time therefore cannot be equated.

Research and system clues about click data do not amount to a ranking formula

Neil Patel’s 2015 article cited research by Radlinski and Joachims indicating that click logs can provide useful training data for ranking. This supports the research value of click data, but that account alone cannot establish how Google’s ranking mechanisms actually work.

A 2025 Semrush article listed names from leaked documents that may be related to Navboost, including clicks, goodClicks, badClicks, and lastLongestClicks. However, the article explicitly stated that the precise meanings of most metrics remained a matter of speculation. Names alone therefore cannot establish what counts as a “good click” or support the conclusion that increasing clicks will improve rankings.

Historical observations and denials must be read with their qualifications

In 2015, Neil Patel described new pages initially ranking relatively high, with rankings potentially rising or falling after a week, and treated daily clicks and average time on page as determining factors. His premise was that no large-scale link building had been done for the page. These were the author’s observations and explanations; the material did not provide causal evidence that ruled out other factors.

Harris’s 2018 article noted that remarks made in 2017 about Google using machine learning to study clicks, time spent, and return behavior still did not prove that these were ranking factors. The article also reported that Google’s Gary denied in 2015 that bounce rate was used as a ranking metric. These statements concern particular periods and metrics. They cannot be generalized to mean that “all engagement is irrelevant,” nor can they support the opposite claim that “longer visits always improve rankings.”