Average session duration: what it is and what's good

What is average session duration?

A session is one visit: a group of things a visitor does on your site in one go. Google Analytics 4 ends a session after 30 minutes of inactivity by default. Average session duration adds up how long every session lasted and divides by the number of sessions (the bounce rate calculator does that sum, and bounce rate, from your own counts). Google's reporting documentation describes the metric as "the average duration (in seconds) of users' sessions".

A worked example, with made-up numbers. Five sessions last 0, 0, 30, 90 and 300 seconds. Together that is 420 seconds, and 420 / 5 = 84 seconds, or 1 minute 24 seconds.

Two things to notice. One long session (E) drags the average up, so the average says little about the typical visit. And if a tool leaves the two zero-second sessions out, the average becomes 420 / 3 = 140 seconds. Including or excluding zero-second sessions changes the average.

How tools measure session time differently

Time on a site is not recorded by a stopwatch. A tool only knows when it receives a signal, such as a page view or an event, so each tool makes choices about what to do when there is no second signal. Two documented examples:

Both come from each tool's own documentation, checked on 17 October 2026 (links at the bottom). So a site with mostly one-page visits will look very different in a tool that counts those visits as zero. Settings change this too, so check the current documentation for yours.

Average session duration in GA4

GA4's standard reports lead with average engagement time. Google's reporting documentation lists Average session duration as a metric, so if you can't see it in a report, check the metrics you can add when you customise a report or build an Exploration. Reports and menus change, so look at your own property.

The two are not the same measure. Google defines engagement time as "the amount of time someone spends with your web page in focus or app screen in the foreground". Session duration is the length of the session as a whole. Because the measures differ, don't expect an engagement time figure to match a session duration figure, from GA4 or any other tool.

For the per-session version, the name says what it is: GA4's user engagement figure is "the total amount of time (in seconds) your website or app was in the foreground of users' devices". Average engagement time per session spreads that across sessions. Average engagement time per user spreads it across users, so the two answer different questions. When a number looks too low or reads 0, the first suspects are tag problems: the tag firing twice, not at all on some pages, or a single-page app that doesn't report page changes.

What is a good average session duration?

A useful target depends on the page's job and how your tool measures time. Compare similar pages within the same tool.

The comparisons that help are your own: the same page last month, the same page by traffic source, or two pages that do a similar job. A page that falls from 90 seconds to 30 after a change is a better lead than any average. Few sessions make the number jumpy, so wait for enough visits before reading a change into it. Related: bounce rate is the other half of this picture, and it has the same problem of being defined differently in every tool. Splitting sessions into new vs returning visitors often explains a duration that looks off.

How to increase average session duration

First decide whether you should. A longer session is not always a better one: a visitor who spends four minutes hunting for a phone number has had a worse visit than one who found it in four seconds. Raise the number only where the page is meant to lead somewhere. Then work through these:

1. Check it's measured properly. Open your own site and see what the tool records for one visit. A tag that fires twice or never on some pages makes every figure wrong.
2. Split it by source and page. One traffic source with very short sessions often means the link or ad promised something the page didn't deliver. Tagged links make this possible, see UTM parameters explained.
3. Make the next step obvious. Link to the page a reader would want next, in the body text, not only in the footer.
4. Open the page as a visitor would. Load it on your phone on mobile data. Is it slow? Does a banner cover the content?
5. Change one thing and wait. Give it enough visits to mean something, then look again.

Looking at sessions per page and per source in Stickybeak

Stickybeak is live analytics for every site you own, and it's in early access. The dashboard shows it as Session time: the average from a session's first to its last activity, including presence pings (see site pages). It also lets you do the splitting in the steps above: click to filter by referrer, page, country or device.

For sites that sell something, each order keeps its trail: the entry page, the referrer, every UTM tag and the pages read before buying. So instead of asking how long visitors stayed, you can ask which entry pages and sources led to a sale. That is often the question you wanted answered. See also SEO and Google Analytics for which tool answers which question.

Stickybeak doesn't have a public price yet. If you'd like to try it when there's a spot, leave your email.

Sources, checked 17 October 2026: Google Analytics Help, Session (30-minute default timeout) and User engagement (engagement time). Google Analytics Data API, API dimensions and metrics (average session duration, user engagement). Plausible Analytics documentation, Metrics definitions (visit duration).