What is a "good" dwell time?

[nicole.healthline]

I know there isn't any official documentation from Google about exact number of seconds a user should spend on a site, but does anyone have any case studies that looks at what might be a good "dwell time" to shoot for?

We're looking on integrating an exact time on site into or Google Analytics metrics to count as a 'non-bounce'--so, for example, if a user spends 45 seconds on an article, then, we wouldn't count it as a bounce, since the reader likely read through all the content.

[IOSC]

I have not seen any studies indicating such a thing,

(but my guess is that is that dwell time seems to be such a strong signal of relevance that google would never release that info, I could be totally wrong though)

An idea to improve UX... if you have a page with 2 paragraphs of text, take the average  time it takes for 10 ppl in your office to read it and set the 'bounce rate' accordingly.  Then you'll know if ppl are reading it.

If you have a page with 2000 words, avg. that time, etc.

If visitors bounce too soon, edit the text until your office avg. meets visitor avg.  That would equal relevance right?

[ThompsonPaul]

The answer to this is really going to be dependent on the page content, Micelleh. A simple page with a clear call to action could result in a user getting exactly what they want from a page within a few seconds and then leaving. A 350 word page might mean 45 seconds, but a 1500 word page might need 2 minutes to prove a user actually got value.

At best, if you insist on a value, get several users to use a good number of your pages, record their on-page time, then create a site-specific average from that.

However, you might be even better off using events for this process, instead of something nebulous like dwell time.

You could add event tracking to the amount of the page a user scrolls, and if they scroll more than half a page (for example), an "interactive" event triggers. "Interactive" events have an effect the same as another pageview (without screwing up your pageview metrics) so a single page visit that scrolled at least half way down the page would no longer be recorded as a bounce.

You could also create interactive events for things like pdf downloads, form submissions, sending emails, viewing a video etc that you consider appropriate for your site to negate what would be considered a bounce.

The biggest benefit to this events-based approach is that it would be vastly more accurate. It would  track visitors' actual actions, as opposed to just assuming a dwell time meant a valuable interaction. (For example, we all know that the habit of opening multiple tabs at once for sequential reading significantly over-inflates time on page for many users.)

Perhaps that idea would work better for what you're trying to accomplish?

Paul