Bounce rate

What the metric measures

The bounce rate shows the share of visitors who leave a page without viewing a second page or triggering an action. The addition matters: a session with a single page counts as a bounce even when the visitor found exactly what they were looking for. The metric measures behaviour, not satisfaction.

The most important rule follows from that. A high value is not automatically a problem. On a contact page, a store location or a glossary entry, bouncing is the norm, because the goal is reached on a single page. On a listing or in the checkout it is a signal.

Where the value is meaningful

• Category pages: Here the visitor should click onwards. A high value suggests missing hits, unclear filters or weak product images.
• Product pages: A bounce can be a purchase, but also an abandonment before the cart. The comparison with the share of cart additions on the same page is what matters.
• Cart and checkout: Nobody should leave on purpose. A value above your own average points at a hurdle, often shipping cost or forced registration.
• Search pages: A bounce after a search usually means the results did not fit, even when hits were displayed.

What the causes can be

Four causes cover most cases. First, wrong expectation: the title promises something different from what the page shows. That happens often with paid ads using broad copy. Second, technical hurdle: slow loading, blocking notices or a cookie banner covering the content. Third, no match: the customer wants a product the category does not carry. Fourth, intent: the page answers the question completely, which is not a defect.

Two values are enough to tell them apart. Time on page shows whether the visitor engaged. Session quality, the share of sessions with genuine interaction, shows whether the metric is high only because of a technical effect. Reading both alongside the bounce rate separates the causes clearly.

What is worth doing

Small adjustments usually work better than a redesign. A clearer title matching the expectation. A notice in the listing when no hits exist for a selection, plus a suggestion. A cookie banner that does not cover the content. And a check of loading time on mobile, because that is where the share of abandoned page views is highest.

What does not work is a blanket target. Anyone trying to push the bounce rate to a given value ends up optimising against their own purpose, because part of the bouncing is intended. Comparing similar pages makes more sense: if two category pages with a similar range show very different values, the search for a cause is worthwhile, because a difference exists that needs explaining.