29.06.2026
Digital Marketing

Definition of bounce rate in GA4, calculation formula, differences from Universal Analytics and levers for improvement: the complete guide to interpreting your non-engaged sessions and improving your visitors' experience.

In Google Analytics 4 (GA4), bounce rate measures the proportion of sessions that were not considered engaged. It helps analyse certain visitor behaviours and identify pages or traffic sources that deserve particular attention.
A high bounce rate can reveal an irrelevant page, a navigation problem or technical difficulties. But it can also correspond to a very short visit during which the user got the information they were looking for.
That is why a good bounce rate is not necessarily a low one. It must be interpreted according to the site's objectives, the type of page and the expected actions.
How do you calculate bounce rate in GA4? Where can you find it in Google Analytics reports? How does it differ from engagement rate and how can you improve the results?
ID Interactive is a digital agency based in Arradon, near Vannes (Morbihan), since 2006, and a Google Partner, Meta Business Partner and Shopify Partner. It supports companies in analysing and optimising their digital performance.
This guide explains how to interpret this indicator and identify the most relevant actions to improve the user experience.

Bounce rate is an audience analytics indicator used to measure the proportion of sessions that do not meet the engagement criteria defined by Google Analytics.
Since Google Analytics 4, its definition has been based on engaged sessions.
A session is a set of interactions carried out by a user on a website or app during a given period.
In its usual configuration, GA4 considers a session engaged if it meets at least one of the following conditions:
Meeting just one of these criteria is enough for the session to be considered engaged.
The 10-second time threshold is the default value and can be changed in the GA4 data stream settings.
A visitor lands on a blog article, reads its content for 30 seconds and then closes their browser.
Even though they viewed only one page, this session is normally considered engaged, since it exceeds the defined time threshold.
Conversely, someone who opens a page and leaves after a few seconds, without any other page view or key event, may generate a non-engaged session.
In GA4, a bounce therefore does not simply mean that an internet user viewed a single page before leaving the site.
Bounce rate is the percentage of non-engaged sessions out of all sessions measured.
The formula is:
Bounce rate = (non-engaged sessions ÷ total sessions) × 100
It can also be calculated from the engagement rate:
Bounce rate = 100% − engagement rate
Imagine a website records 1,000 sessions over a given period.
Of these sessions:
The bounce rate is therefore:
350 ÷ 1,000 × 100 = 35%
The engagement rate is 65%.
This result alone does not allow you to conclude whether the site is performing well or badly.
You need to know in particular the objectives of the pages and the actions carried out by visitors.
The historical definition of bounce rate changed with the move from Universal Analytics to Google Analytics 4.
In Universal Analytics, a bounce was mainly associated with sessions containing a single interaction sent to Analytics.
In GA4, it corresponds to sessions that meet none of the engagement criteria.
Not directly.
The two tools rely on different definitions and data collection methods.
A 65% bounce rate once observed in Universal Analytics does not mean that a 35% rate in GA4 necessarily represents an improvement.
To analyse changes, it is better to compare periods using consistent measurement methods.
Changes to the GA4 configuration must also be taken into account.
Bounce rate does not necessarily appear in the default Google Analytics 4 reports.
Depending on the property's configuration and the permissions available, you may need to add this metric.
In GA4:
Customising reports requires the appropriate permissions, generally an Editor or Administrator role.
If you cannot see the customisation option, check your account's permissions.
The landing page is the first page viewed during a session.
It is a particularly useful entry point for analysing bounce rate.
For example, a visitor may enter on a service page, then view two other pages, and then leave the site.
In this case, the session should not be interpreted as a bounce simply because the person eventually leaves the site.
To identify the entry pages that generate the most non-engaged sessions, favour the report Landing pages.
The report Pages and screens answers other questions about the content viewed. It should not be confused with a manual bounce calculation for each page viewed.
There is no ideal bounce rate valid for all websites.
The result depends on the sector, the objective of each page, the type of traffic and the analytics configuration.
A showcase site, an e-commerce store and a blog do not have the same journeys.
An informational article can quickly answer a specific question.
For example, someone looking for the definition of a term may find their answer and leave the page without any further action.
This does not necessarily constitute a bad experience.
However, if the visit is very short and the page does not match the search intent, a high bounce rate may reveal a relevance problem.
On a service or quote request page, the objective is generally to encourage an action.
A high bounce rate may therefore deserve a closer look, particularly if the number of contact requests is low.
You need to check how well the offer is understood, the quality of the information and the simplicity of the journey.
For an online store, visitors may land directly on a product page or a collection page.
Bounce rate must be studied alongside add-to-carts, orders, internal searches and other commercial interactions.
A product page generating many non-engaged sessions may need improvements.
However, the decision must be based on real results rather than an isolated rate.
These examples show that the same bounce rate value can correspond to very different realities.
No.
A high bounce rate can correspond to several situations.
An ad or a search result may promise information that the page does not clearly provide.
The visitor then leaves the site because the content does not match their need.
An overly long loading time, an impractical mobile interface or a form that doesn't work can prevent the user from continuing.
Some pages meet immediate needs, such as finding a phone number, an address or a short piece of information.
A visit with no recorded engagement therefore does not always indicate dissatisfaction.
A visitor may click an external link, open a messaging app or start a phone call.
If the action is quick and not recognised as a key event, the session may not meet the engagement criteria.
You therefore need to analyse the real journey before drawing conclusions.
Not necessarily.
A bounce rate close to 0% may seem excellent, but it can also reveal a configuration problem.
For example:
In these situations, the tool may consider too many sessions as engaged.
You should never change events solely to lower the bounce rate.
The priority must remain the faithful measurement of the interactions that really matter to the company.
Before changing pages, you need to identify the origin of the observed result.
Start by spotting the pages that receive many non-engaged sessions.
Focus on those that are important for the business or that generate a lot of traffic.
An anomaly on a quote page may be more of a priority than a similar variation on a rarely viewed article.
Visitor behaviour can differ depending on where they come from:
For example, an advertising campaign poorly aligned with its landing page can generate many short visits.
Conversely, brand traffic may behave differently from people discovering the company.
You therefore need to compare comparable audiences.
A high bounce rate only on mobile may signal a display or interaction difficulty.
It is useful to check menus, buttons, content and forms on real smartphones.
Track how bounce rate changes over time.
A change after a redesign, a new campaign or a technical change deserves analysis.
However, make sure to take into account seasonality and any changes to the GA4 measurement plan.
Bounce rate is more useful when combined with other data.
The aim is to identify the problems that really hold visitors back, without focusing solely on an overall average.
Improvement actions must address the observed causes.
A site does not necessarily need to apply all the following recommendations.
Internet users must find the information or offer announced in the search result, the ad or the link they have just followed.
Check consistency between:
A page that answers the visitor's need quickly and clearly reduces some frustrations.
Visitors must understand where they are and how to access useful information.
This means working on menus, content hierarchy and the readability of actions.
Work on UX/UI design makes it possible in particular to organise interfaces around users' needs and the company's objectives.
Simpler navigation can make it easier to discover additional content.
However, it does not guarantee a systematic drop in bounce rate, particularly on pages meeting a very one-off need.
A slow site can discourage visitors before they have been able to view the content.
Several factors can be involved:
Google recommends tracking in particular the Core Web Vitals: LCP, INP and CLS.
PageSpeed Insights helps identify certain technical points to improve.
A page must remain easy to read and use on mobile.
Check in particular text size, menu behaviour, buttons and forms.
A responsive site is not automatically user-friendly.
To find out more about this subject, see ID Interactive's guide to website ergonomics.
Visitors do not always read an entire page in order.
Clear headings, well-organised paragraphs and relevant tables make information easier to access.
Favour:
Above all, an article must correctly answer the search intent, rather than trying to artificially hold on to its reader.
Internal linking directs visitors to content related to their need.
For example, an article explaining the cost of a website can link to the corresponding website creation services.
Links must be placed in a relevant context, with an explicit label.
The aim is not to provoke an extra click at all costs.
It is about making navigation easier when visitors want to explore a subject further.
A button must clearly describe the proposed action: request a quote, view case studies or book a service.
A page can include several calls to action when this meets its visitors' needs.
There is no universal rule imposing a maximum of three CTAs per page.
It is more relevant to prioritise actions, avoid repetitive prompts and observe interactions.
Some elements can prevent internet users from continuing their visit:
These problems must be checked first when they concern pages that are important for contact requests, bookings or sales.
A poor analytics configuration can lead to misleading conclusions.
In GA4, key events are used to measure actions that are particularly important for the business.
Depending on the site, this may be a submitted form, a booking or an order.
Marking an event as a key event can also influence how sessions are classified as engaged.
This status should therefore be reserved for genuinely relevant actions.
For example, marking every page view as a key event would distort the interpretation of engagement.
After adding a tag management tool, a redesign or a change to Analytics tracking, check that a page view is not recorded several times.
Two page_view events during a session can be enough to meet the two-view criterion, even if this result comes from a technical error.
Duplicates can then cause an artificial drop in bounce rate.
The 10-second threshold can be adjusted in GA4.
Changing it alters the operational definition of some engaged sessions.
It can therefore affect comparisons with previous periods.
This parameter must be chosen according to measurement needs, not to display a more favourable result.
Audience measurement can be affected by the consent mechanism, privacy settings, tracking blockers or analytics configuration choices.
GA4 data therefore does not necessarily correspond to all real visitors.
These limits must be kept in mind when comparing periods, sites or traffic sources.
No. The bounce rate measured in Google Analytics 4 is not used as a direct ranking factor in Google Search.
Two elements must be distinguished.
The first concerns the performance observed in GA4: sessions, engagement, key events and measured behaviour.
The second concerns Google Search's ranking systems, which assess in particular the relevance of content and various aspects of page experience.
Improving ergonomics, technical performance and editorial quality can be useful for SEO.
But a drop in GA4 bounce rate is not in itself proof of an SEO improvement.
It guarantees neither more impressions, nor a better position, nor an increase in clicks from Google.
To check organic results, you need in particular to analyse Google Search Console data.
The ID Interactive team supports companies in analysing these issues as part of its services for SEO.
An internet user arrives from Google on an article giving the definition of a term.
They find their answer quickly and then close the page.
If the session meets no engagement criteria, it may be counted as a bounce.
This behaviour does not necessarily indicate dissatisfaction.
The analysis must focus on the relevance of the answer and the usefulness of the content.
A company notices a high bounce rate on a sales page, along with a low number of contact requests.
Several hypotheses can be explored: an unclear promise, missing information, poor mobile adaptation or insufficiently qualified traffic.
User tests and analysis of the conversion journey will help identify the relevant fixes.
A new Google Tag Manager configuration is deployed.
The bounce rate suddenly drops to a very low value, with no visible change to the site and no increase in contact requests.
Before concluding that things have improved, you need to check the events sent to GA4, any duplicates and the key event configuration.
The change may come from the measurement set-up rather than from user behaviour.
ID Interactive works on analysing website performance and optimising user journeys.
This approach can combine audience data, business results, content quality and the site's technical characteristics.
Bounce rate analysis can in particular be part of several steps:
1. Check the data. Make sure that events, sessions and key events match the measurement objectives.
2. Identify priority pages. Spot the entry pages that matter for the business.
3. Analyse behaviour. Compare traffic sources, devices and the actions carried out.
4. Examine the user experience. Identify obstacles in content, navigation or forms.
5. Prioritise fixes. Choose the interventions with real operational value.
6. Measure results. Compare performance after the changes, keeping a consistent measurement method.
This approach is part of ID Interactive's digital marketing services, linked to visibility, visit quality and conversion.
Not necessarily.
Bounce rate is a useful indicator for identifying non-engaged sessions according to GA4 rules.
It can reveal a mismatch between traffic and content, a usability difficulty or a technical problem.
But it is not enough to measure visitor satisfaction, commercial performance or a website's SEO.
The priority is not to get the lowest possible bounce rate, but to improve the quality of journeys and the results that really matter to the company.
Would you like to analyse your site's performance and identify areas for improvement? Contact ID Interactive to discuss your needs.
Article originally published on 19 May 2021.
Dans GA4, un événement clé est une action particulièrement importante pour l’activité : formulaire envoyé, réservation ou commande, selon le site. Un seul événement clé suffit à rendre une session engagée. Ce statut doit donc être réservé aux actions réellement pertinentes : marquer chaque vue de page comme événement clé fausserait l’interprétation de l’engagement.
Le taux d’engagement est la part des sessions qui répondent aux critères d’engagement de GA4. Il est le complément du taux de rebond : taux de rebond = 100 % − taux d’engagement. Dans l’exemple de l’article, 650 sessions engagées sur 1 000 donnent un taux d’engagement de 65 % et un taux de rebond de 35 %.
Pas nécessairement. Dans GA4, une session sur une seule page est engagée si elle dépasse 10 secondes (valeur par défaut) ou comporte un événement clé : une lecture de 30 secondes est ainsi une session engagée. Dans Universal Analytics, une visite d’une seule page était en revanche souvent comptabilisée comme rebond.
Oui. Les 10 secondes sont la valeur par défaut et peuvent être modifiées dans les paramètres du flux de données GA4. Un changement modifie la définition des sessions engagées et peut affecter la comparaison avec les périodes précédentes : le seuil doit répondre au besoin de mesure, non à l’envie d’afficher un résultat plus favorable.
Non. La mesure d’audience peut être affectée par le dispositif de consentement, les paramètres de confidentialité, les bloqueurs de suivi ou les choix de configuration analytique. Les données GA4 ne correspondent donc pas nécessairement à l’ensemble des visiteurs réels, une limite à garder à l’esprit pour comparer des périodes, des sites ou des sources de trafic.