Dwell time is one of those SEO terms that refuses to die. Every few months a new guide resurfaces claiming it’s a “hidden ranking factor” you need to optimize for, usually paired with a screenshot of average session duration in Google Analytics. The reality is messier, and more useful once you understand it.
The short version: dwell time is the interval between a user clicking your result on a Google search results page and returning to that SERP. It’s a Bing-coined concept that SEOs imported, applied loosely, and then attributed to Google without much evidence. Google has said publicly, more than once, that dwell time is not a direct ranking signal. But, and this is the part most blog posts skip, the engagement patterns dwell time tries to approximate (long clicks, pogo-sticking, return-to-SERP behavior) absolutely do appear inside Google’s quality systems, based on the May 2024 Content Warehouse leak and Pandu Nayak’s DOJ antitrust testimony.
So the question isn’t “how do I increase dwell time.” The question is: what signals is Google actually using, what do they correlate with on your site, and how do you build pages that satisfy them honestly? That’s what this post is for.
Key takeaway: Dwell time is not a confirmed direct Google ranking factor. John Mueller and Gary Illyes have stated this on the record. However, engagement signals that resemble dwell time, particularly click satisfaction patterns like long clicks and pogo-sticking, are referenced in leaked Google documentation and DOJ testimony. Optimize for genuine engagement, not the metric.
What dwell time actually means
Dwell time, in its original and technically correct definition, is the amount of time that passes between a searcher clicking a result on Google (or Bing) and returning to the SERP. If you click a result, read it for 90 seconds, then hit back to try another result, your dwell time on that page was roughly 90 seconds.
Two things make this definition distinctive:
- It only applies to traffic that came from a search engine. A direct visit, a social click, or an email click has no dwell time.
- It ends when the user returns to the SERP, not when they close the tab, navigate elsewhere, or convert. If a user clicks your result and never goes back to Google, dwell time is effectively undefined (often interpreted as “satisfied”).
The term was popularized by Duane Forrester at Bing in a 2011 blog post discussing how Bing thought about result quality. Google has never adopted the term in its public communications, which is why every conversation about “Google dwell time” is, strictly speaking, an interpretation.
Dwell time vs. the metrics it gets confused with
The biggest source of confusion in SEO conversations about dwell time is that it gets used as a synonym for metrics that measure different things. Here’s how they actually differ.
Time on page
Time on page is calculated by analytics tools (historically Universal Analytics, now GA4 through engagement time) by measuring the interval between pageviews. It applies to all traffic sources and is famously unreliable for single-page sessions, if a user views one page and leaves, classic Universal Analytics recorded a time on page of zero. GA4 changed this by using engagement time, which tracks active time the page is in focus.
Session duration
Session duration is the total time of a user’s visit across all pages until they go idle or leave. It includes every page they viewed, not just the entry page. Useful for understanding site-wide engagement, useless as a proxy for “did this specific page answer their query.”
Bounce rate
Bounce rate (in the classic Universal Analytics sense) is the percentage of sessions with exactly one pageview and no interactions. A high bounce rate is often misread as bad, but if your page answered the user’s question in 20 seconds and they left satisfied, that’s a bounce and a win at the same time. GA4 replaced bounce rate with engagement rate, which is a more honest framing.
The key distinction
None of your analytics metrics are dwell time. Dwell time exists in Google’s systems, not yours. You cannot see it in GA4. The best you can do is build proxies, which I’ll get into below.
What Google has actually said
Let’s separate the public statements from the leaked and litigated evidence, because they tell different stories.
The public line: not a ranking factor
John Mueller, in multiple Search Off the Record episodes and office hours hangouts going back to 2017, has said dwell time, time on site, and bounce rate are not direct ranking factors used by Google’s algorithms. Gary Illyes has echoed this, pointing out that these metrics are too noisy and too easy to manipulate to use as raw inputs. Google’s stated position is that user behavior on a third-party site isn’t directly fed into rankings.
The DOJ testimony: click signals matter
In the 2023 United States v. Google antitrust trial, Pandu Nayak, Google VP of Search at the time, testified about a system called Navboost, which uses 13 months of click data to influence rankings. Navboost specifically distinguishes between “good clicks” and “bad clicks.” A long click (the user clicks a result and stays for a while) is treated differently than a short click followed by a return to the SERP. That is, structurally, the same behavior dwell time tries to capture.
The 2024 Content Warehouse leak
In May 2024, internal Google API documentation leaked, exposing thousands of ranking-related attributes. Several referenced click-quality signals, including attributes named goodClicks, badClicks, lastLongestClicks, and unsquashedClicks. These attributes live inside modules associated with Navboost and Glue. They strongly suggest that Google quantifies whether searchers were satisfied with a result, partly through how long they spent on it before returning.
So: dwell time per se is not a ranking signal. But the behavioral pattern dwell time describes, searcher satisfaction inferred from click duration, is demonstrably used by Google. The distinction is real, and it matters for how you should think about optimization.
Why optimizing for dwell time directly is the wrong move
If you go looking, you’ll find blog posts recommending you “increase dwell time” with things like: hiding the conclusion behind a click-to-expand, splitting articles across pagination, embedding long videos at the top, or pumping content with filler so people scroll longer. All of these are bad ideas.
Two reasons:
- They optimize for the metric, not the goal. A user trapped on your slow-loading, padded article isn’t satisfied, they’re frustrated. The “dwell time” you generate is the bad kind.
- Google’s quality systems are explicitly designed to distinguish satisfied dwell from frustrated dwell. Navboost rewards long clicks where the user doesn’t return to search and try a different result. Filler doesn’t help with that, it often hurts, because pogo-sticking goes up when content disappoints.
The honest framing is this: build pages that genuinely answer the query better than the next result on the SERP. Dwell time, engagement time, scroll depth, and every other behavioral proxy will improve as a byproduct. Inverting that, chasing the proxy and hoping satisfaction follows, doesn’t work.
Levers that actually improve engagement
If you accept that the right target is real searcher satisfaction, here are the levers that move it. These are the same things that improve every other engagement metric you care about.
Match search intent precisely
Look at the current top 10 for your target query. Are they listicles, in-depth guides, product pages, comparison posts? That format is what Google has learned satisfies the intent. If you publish a 3,000-word essay for a query that wants a quick comparison table, you’ll lose every behavioral signal Google measures, including the one that resembles dwell time.
Front-load the answer
The user who clicks your result is impatient. They want to confirm within the first few seconds that they landed in the right place. Lead with a clear definition, a key-takeaway box, a TL;DR, whatever’s appropriate for the format. Counterintuitively, answering quickly increases dwell time, because the user trusts the page and keeps reading rather than bouncing back to find a better one.
Fix page speed and Core Web Vitals
A slow page erodes engagement before the content has a chance. LCP under 2.5 seconds, INP under 200 milliseconds, CLS under 0.1, those aren’t arbitrary targets, they’re the thresholds at which user behavior changes measurably. Every second of additional load time corresponds to higher bounce probability in Google’s own research.
Improve readability and scannability
Short paragraphs, descriptive subheadings, lists where appropriate, sensible line height. Most people don’t read web content, they scan it. A wall of text causes immediate disengagement regardless of how good the content is.
Internal linking that’s actually useful
When a user finishes your post, the next logical action shouldn’t be hitting back. Internal links to deeper or related content keep the session going and signal a healthy site architecture. The keyword is useful, linking for the sake of linking is noise. Link where you’d genuinely send a curious reader.
Earn the title and meta description
A clickbait title produces a short click. The user feels misled and pogo-sticks back to the SERP, which is precisely the signal Navboost penalizes. Write titles that promise what the page actually delivers.
How to measure proxies for dwell time in GA4
You can’t see dwell time directly. You can build a reasonable picture of engagement quality from your own analytics. Focus on these four reports.
Engaged sessions and engagement rate
GA4 defines an engaged session as one lasting at least 10 seconds, with a conversion event, or with two or more pageviews. Engagement rate is the percentage of sessions that qualify. Filter to organic search traffic and look at engagement rate by landing page, pages with low rates are your dwell-time problem children.
Average engagement time per session
This measures active time the tab was in focus, which is a much better proxy for genuine attention than the old time-on-page metric. Segment by landing page and source/medium to find pages that fail organic visitors specifically.
Scroll depth
Enable enhanced measurement (or set up custom scroll triggers) and you’ll see what percentage of users reach 90% page depth. A page where most organic visitors never get past the first viewport is failing, either the intent match is wrong or the opening doesn’t earn the scroll.
Exit page analysis
Look at your top organic landing pages and see where users exit. If they exit from the landing page itself, you have a satisfaction problem. If they exit deeper in the funnel, the entry page did its job.
Pair this with Google Search Console click-through rate and position data. A page with strong CTR but weak engagement is winning the click and losing the user, which is precisely the bad-click pattern Navboost is built to detect.
The honest summary
Dwell time is a useful concept and a misleading SEO target. The metric itself is not something Google has ever confirmed using directly, and Google representatives have said as much for years. But the underlying behavior, searchers clicking, staying, and not coming back to try a competitor, is unambiguously part of how Google evaluates results, based on DOJ testimony about Navboost and the click-quality attributes exposed in the 2024 Content Warehouse leak.
The takeaway for anyone doing SEO in 2026 is simple. Stop trying to game a metric you can’t see. Build pages that match intent, load fast, answer quickly, read well, and link sensibly. The behavioral signals Google does use will follow, your GA4 engagement metrics will follow, and your rankings will follow. That’s the only sustainable version of “optimizing for dwell time” that actually exists.