What is a good ecommerce bounce rate?

The question usually arrives wanting a number. Something like 45%, so you can hold your own figure against it and find out whether you have a problem. The honest answer is that the number you will find measures a different thing from the number your analytics reports, and the gap between those two definitions is wider than any difference between your store and the average.

Bounce rate was redefined, and the benchmarks were not

Universal Analytics counted a bounce as a session with a single interaction: one page, no event, however long the visitor stayed. GA4 counts the opposite thing and subtracts it. A session is engaged if it lasts longer than ten seconds, or fires a key event, or includes two or more page views, and bounce rate is the share of sessions that were not engaged. Universal Analytics stopped processing data for standard properties on 1 July 2023, so most stores have spent three years reading the second definition while comparing themselves against tables written for the first.

The size of the gap depends on your traffic, so the arithmetic is worth doing on your own. Take a store with 10,000 sessions, 6,200 of them single-page visits with no event. Universal Analytics reports a 62% bounce rate. If 1,800 of those 6,200 lasted longer than ten seconds, GA4 counts them as engaged and reports 44%. Nothing about the store changed. Those inputs are illustrative rather than measured, and the mechanism is the point: the longer your visitors linger before leaving, the further apart the two numbers sit.

A single number could not answer it anyway

Bounce rate is a property of the traffic as much as of the site. Two identical stores with different traffic mixes report different bounce rates, and neither of them is wrong.

• Branded search bounces least. Someone who typed your name arrived on purpose and already knew what they wanted.
• Broad display and paid social bounce most. Nobody was looking for you when the ad interrupted them.
• Mobile bounces more than desktop on the same pages, partly because the connection is worse and partly because the session is shorter by nature.
• A blog post should bounce more than a category page. It answered the question and the reader left satisfied. That is the article working, not failing.

A site-wide figure averages all of that into one number and hides every fact worth knowing. Move budget from branded search into display and the number gets worse while the business gets bigger.

The comparison that works is against yourself

The useful question is never whether 55% is good. It is why one page bounces at twice the rate of its neighbours when both were reached by the same kind of visitor. Four cuts find almost everything.

1. By landing page. This is the cut that finds problems, because it separates the page from the traffic that arrived at it.
2. By device. A page bouncing at twice the rate on mobile has a mobile problem, not a content problem.
3. By source. If one campaign bounces far above the others, the mismatch is between the ad and the page rather than inside either.
4. Against the same segment last month. Not against a benchmark from a study you cannot inspect.

What a bounce is usually a symptom of

Three causes account for most of it, and none of them is fixed by aiming at bounce rate directly.

• Speed. Google puts a good Largest Contentful Paint at 2.5 seconds or under, needs improvement up to 4 seconds, and poor above that. On a slow page the visitor leaves before there is anything to bounce off.
• Intent mismatch. The ad or the search snippet promised something the page does not open with. The visitor is not disappointed by the page so much as by the difference between the two.
• Nothing legible above the fold. A hero image and a slogan is not an answer to the query that brought somebody there.

The middle one is a paid search and copy problem more often than a design problem, and it is the same failure described in why a site gets visitors and no enquiries.

What to do with the number

1. Confirm which definition you are reading. If the property was migrated, nothing before mid-2023 is comparable to anything after it.
2. Segment before concluding anything. Landing page first, then device, then source.
3. Fix the pages where bounce and intent disagree, rather than the pages with the highest number.
4. Then stop watching it. Bounce rate points at a page worth fixing. It is not a target, and a store can push it down without selling anything more.

The figures that decide whether the work paid for itself sit further down: enquiries, orders, and what an order is worth. Bounce rate is useful for finding the page that is losing them, and for very little else.