Dwell time
The interval between a searcher clicking a result and returning to the results page. It is an SEO-industry term, not a metric Google publishes, documents, or exposes in any tool.
Karl-Gustav Kallasmaa, Founder & CEOLast updated Dwell time is the interval between a searcher clicking a search result and returning to the results page. It is a term of the SEO industry, not of Google's documentation: Google does not publish it, does not name it among its ranking systems, and does not expose it in any tool a site owner can open.
That is the fact most pages about dwell time omit, and it changes what you should do about it.
Why nobody outside the search engine can measure it
The measurement requires two timestamps taken on opposite sides of a boundary. The first is the click that leaves the results page; the second is the return to it. Only the search engine observes both.
Your analytics sees the middle of that journey and neither end. It knows a session started and roughly how long the page held focus, but not that the visitor came from a results page they then went back to — and a visitor who leaves for a competitor, a bookmark, or lunch is indistinguishable in your data from one who bounced back to the SERP dissatisfied. Search Console reports the click and the impression, and stops there.
So dwell time is not a metric you can improve deliberately, because it is not a number you can read. It is an inference about satisfaction, drawn from something you cannot see.
What Google actually documents
Google's guide to its ranking systems lists the systems it says are at work in ranking: BERT, crisis information systems, deduplication, the exact-match domain system, freshness systems, link analysis and PageRank, local news systems, MUM, neural matching, original content systems, removal-based demotions, passage ranking, RankBrain, reliable information systems, the reviews system, site diversity, and spam detection. No user-behaviour signal appears in that list — not dwell time, not bounce rate, not click-through rate.
That is not proof no such signal exists anywhere in the stack. It is proof that acting as if dwell time were a documented, controllable lever means acting on something Google has not said.
The measurable thing that is not dwell time
The closest documented analogue is engagement time in Google Analytics 4, and it is a different measurement. GA4 defines user engagement as the time a page was in focus or an app screen in the foreground, and counts a session as engaged when it lasts longer than 10 seconds, has a key event, or includes two or more page or screen views. Bounce rate in GA4 is simply the inverse of that.
Note what this measures: activity on your site, across all traffic, on your own instrumentation. It never sees a results page. Using it as a stand-in for dwell time is defensible as a proxy for reader satisfaction, and indefensible as a claim about how a search engine perceives your page.
Failure modes
- Optimising the proxy. Splitting one answer across three pages raises page views per session and makes the reader's life worse. Any tactic that increases time on page by delaying the answer is optimising against the reason the metric was ever interesting.
- Reading a low number as a verdict. A page that answers a question in twelve seconds and is never seen again did its job, and zero-click behaviour makes that outcome more common, not less. Reference pages, opening hours, and error-code lookups all look identical to failure by this measure.
- Attributing it to a ranking system. Building a content strategy on an undocumented signal means you cannot tell whether a change worked, because you were never able to observe the input.
- Assuming it transfers to AI answers. In a retrieval-based answer there is no results page and no return trip. Nothing about the reader's session is visible to the system that chose the passage, whether that is an AI Overview or an assistant.
What to do instead
Measure what you can see, and treat it as a proxy that you own. Time in focus, scroll depth, and whether the reader reached the section that answers the question are all instrumentable on your side. Fix the causes the intuition points at — a slow page, an answer buried under preamble, a headline that promised something the page does not contain — and measure the fix with Core Web Vitals and your own analytics, not with a number no tool reports.
Frequently asked questions
Does Google document dwell time as a ranking system?
No. Its published ranking-systems guide lists no user-behaviour signal.
Can I see it in a report?
No. The measurement needs both the departure from and the return to the results page.
Is GA4 engagement time the same measurement?
No. It is time in focus on your site, with no visibility of any results page.
Does it apply to AI answers?
No. There is no results page to return to; fetchability and passage self-containment are what matter.
Terms related to Dwell time
Google's three published field metrics for loading, responsiveness and visual stability, with the exact thresholds and the percentile they are judged at.
The two separate stages that decide whether a page can be retrieved at all, and the reason a serving rule on a blocked page is never read.
A search that ends without the reader visiting any website, and the measured gap between sessions that show an AI summary and those that do not.
Google's AI-generated summary at the top of a results page, and the snippet controls that decide whether your page can appear inside one.