Average engagement time is the GA4 metric for how long your site or app was actually in view for users, on average. It counts only the time a page was in focus in the browser, or an app was on screen, so a tab left open behind other windows adds nothing.
How average engagement time works
While a page is in focus, the Google tag keeps a running count of milliseconds. When the visitor switches tab, moves to another page or leaves, the tag sends that count with a user_engagement event, and GA4 adds the time up and divides it.
The divisor matters. At the time of writing (October 2026), GA4 offers average engagement time per active user, which spreads total engagement time across the people active in the period, and average engagement time per session, which spreads it across visits. A loyal visitor who returns five times in a week raises the per-user figure far more than the per-session one, so check which version a report shows before comparing it with anything.
This is a different measurement from the old Universal Analytics figures, average session duration and average time on page. Those measured the gap between one recorded hit and the next, so a visitor who read a single page for four minutes and left was counted as zero seconds. GA4 measures foreground time directly, so it is usually closer to reality, but it cannot be compared with Universal Analytics data, which standard properties stopped collecting in July 2023. The same timing also decides whether a visit counts as an engaged session, the basis of engagement rate.
Why it matters
Engagement time is one of the few signals in GA4 that tells you whether people read what you published. On a guide for UK landlords explaining a change in tenancy rules, a page that holds attention for two or three minutes is probably being read. If visitors average eight seconds, the page is probably answering the wrong question, or failing to load properly on a phone.
It is most useful as a comparison: one page against similar pages, this month against the same month last year, or visitors from one channel against another. Paid social traffic with far lower engagement time than organic search often points to a mismatch between the ad’s promise and the landing page, rather than a weak page.
For content planning, it separates pages that attract clicks from pages that actually serve the reader. That distinction is what content SEO depends on when deciding which pages to rewrite, merge or leave alone.
Common mistakes
- Treating longer as always better. A contact page where visitors find your phone number in eight seconds and ring you is doing its job perfectly.
- Reading an average without looking at the spread. On a page with little traffic, one or two visitors who leave it open in the foreground can lift the average a long way.
- Comparing per-user with per-session figures, or GA4 engagement time with old Universal Analytics time on page.
- Ignoring sites built as single-page applications, where page changes may not register and all the time piles up against the first page. Single page application tracking has to be set up deliberately.
- Forgetting consent. With a cookie banner in front of GA4, engagement time describes visitors who accepted analytics, who may behave differently from those who declined.
How to act on it
Add average engagement time to the Pages and screens report if it is not already there, and sort your most-visited pages by it. Look for pages with plenty of visits and unusually little engagement. Check which searches bring people there in Search Console, read the page on a phone, and ask whether the opening paragraph answers the question people arrived with.
Then compare like with like. Group pages by type, such as service pages, guides and blog posts, and judge each against its own group. Read engagement time alongside bounce rate and key events, because a page that holds attention but never leads to an enquiry needs a clearer next step, not more words.