Average Session Duration
How long an average visit lasts, from arrival to the last thing measured.
What it means
Average Session Duration is the total time across every visit divided by the number of visits, shown in minutes and seconds. Time is only measured between recorded events, so a visit that ends on the page it started on records close to nothing. Bounces are counted in the average and drag it down, which makes this a rough signal rather than a precise one.
Show the math
Formula and a worked example
Total session time adds up every visit from its first event to its last recorded one. Whatever someone does after that final event — including reading the page they're on — is invisible.
Sessions counts every visit in the period, bounces included, and a bounce contributes roughly zero seconds.
Worked example. 40,000 sessions totalling 4,200,000 seconds gives 105 seconds — 1m 45s.
Now take the bounces out: the 25,600 Engaged sessionsEngaged sessionsEngaged sessions: sessions lasting 10s+, with a key event, or 2+ page views average 2m 44s. Same site, same month, and that's the number that describes a real visit.
It answers the question
When someone visits, how long do they stay? Short can mean a quick path to checkout or a page that lost them, and this number alone won't say which.
Why it matters
It's a cheap early signal that something on the site changed. A theme update, a slow script or a broken template shows up here almost immediately, usually before anyone notices it in revenue.
It also separates two kinds of weak conversion. Short visits with few orders means people never engaged, which is a traffic or landing problem; long visits with few orders means they engaged and couldn't finish. Time in the shop works like time in the aisles — someone lingering is either interested or unable to find what they came for.
What good looks like
There's no universal standard, and chasing a target here does more harm than good — a one-product repeat-purchase shop should run short, a considered €400 purchase should run long. Judge it against your own trend rather than against any outside figure. A sharp move with nothing else on the page changing is more often a site or tracking change than customer behaviour.
How to improve it
| Lever | What you do | Expect | How long | Watch out for |
|---|---|---|---|---|
| Fast Answer the first question on the first screen | Put sizing, delivery time and returns where people look first instead of three scrolls down | Fewer visits ending in seconds, duration up | 1 week | It pushes photography and the buy button lower, and impulse products can lose the add to cart you already had. |
| Fast Fix the templates that load slowest | Compress hero images and defer render-blocking scripts on the pages taking the most traffic | Duration up as more visits survive the first few seconds | 1–2 weeks | Deferred scripts can silence reviews, personalisation and tags — check what still fires before you trust the lift. |
| Slow Give a visit somewhere to go next | Related products, collection links and reviews below the fold | More second page views, duration up | 3–5 weeks | Every extra link is another way off the buying path. Watch Conversion Rate on those templates — a longer visit that ends without an order is worth nothing. |
| Slow Change the source mix, not the pages | Cut the placements that send visits which leave immediately and reinvest in the ones that stay | Site-wide duration rises without touching the site | 1 quarter | Sessions and reach fall with them, and the cheap placements you drop were often filling the retargeting audiences the rest of the account depends on. |
Every lever costs something somewhere. The last column is the one to read twice.
Read it with
Duration is an average, and averages hide their shape. It takes a second number to say whether a move is real.
Reading and buying
People are spending time and finishing. Whatever you changed is worth repeating — write down what it was while you still remember.
Hunting, not buying
Long visits that end empty. That's missing information, confusing navigation or a checkout that's hard work — the time is being spent looking for something, not enjoying it.
A shorter path
Not a problem. People are finding what they came for and paying faster, which is what a good product page is supposed to do.
Arriving and leaving
Visits are getting shorter and worse. A slow template or a change in where the traffic comes from explains most cases of this.
Bounced visits record almost no time, so they sit inside this average pulling it towards zero. A duration that drops in the same week Bounce Rate climbs is one event, not two problems — and reading them together stops you rewriting page content when the real fault was load speed.
Depth against breadth. Plenty of engaged sessions with a falling average means more visits are clearing the 10-second bar and stopping just past it, which looks like improvement in one column and dilution in the other.
Common misreads
Only up to a point. Time spent because someone can't find the size chart or the delivery cost is time you'd rather they didn't spend. Judge it against orders, never on its own.
Nothing after the last recorded event is counted, so the true figure is higher and unknowable. It's a comparison tool between rows and periods, not a fact about attention.
A shift towards mobile, a new broad campaign or a change in which pages take the traffic all move this without a word changing on the site. Check whether it fell on every row or only on the new ones.
Also called
Avg. Session Duration · time on site · average visit length
See yoursYour average visit length for the period, and the change on the period before.
Open Google Analytics Overview →