A simple bounce-rate example
Suppose a report contains 100 visits. Forty have one matching pageview and no custom event. The other sixty have another matching pageview or a custom event. The bounce rate for that dataset is 40%.
Keep the selected period and filters in mind. They determine which activity is included in the calculation. Do not compare a filtered landing-page report with an unfiltered whole-site report and assume they describe the same behaviour.
40 bouncing visits ÷ 100 visits × 100 = 40% bounce rateA single-page visit can be a successful visit
A reader may find the answer to a question, read it carefully and leave. Someone visiting a contact page may find the phone number immediately. Those visits can be useful even when there is no second pageview.
A product page intended to lead to a signup has a different job. If few people continue, investigate the offer, the call to action and the mobile experience. The useful comparison is with similar pages and consistent measurement, alongside the action you want readers to take.
Custom events change what counts as a bounce
In our measurement engine, a custom event makes a visit non-bouncing. Adding an automatic event to every page can therefore reduce bounce rate even if visitors behave exactly as before. A lower number after a tracking change is not proof that the page improved.
Track actions that help you answer a real question, such as a deliberate button click or a completed form interaction. Name them consistently and document when you introduce them. Avoid sending personal information in event names or properties.
How to investigate a page with a high bounce rate
First verify that the tag is present on the next page in the journey. Missing tracking can make a multi-page visit look like a one-page visit. Next compare devices and sources: a campaign promising something different from the destination page can bring traffic that has little reason to stay.
Check the page yourself on a phone. Look for a slow or broken experience, unclear wording, distracting layout, or a next step that is hard to find. Change one meaningful thing, then compare equivalent periods and the relevant action counts.
- Confirm the metric definition and event setup have stayed the same.
- Check that the page’s intended next step actually works.
- Compare similar traffic sources and device groups.
- Judge the outcome using useful actions as well as bounce rate.