The bounces (or “rebounds”) are the hits that count only one page view , or in other words browsing sessions of the site which consist of a single page view.
Consequently, the ” bounce rate ” (or “rebound rate”) should represent the percentage of surfers who, having reached the site, immediately abandon it. The overall bounce rate of a site is given by the percentage of visits that end already after viewing the first page viewed: a high bounce rate is commonly a negative figure which should be interpreted as poor user satisfaction with the contents found in the site he came across while browsing.
However, this figure must be taken with due caution : let's see why.
What is a visit to a website? According to the current definition, a visit is a sequence of http requests coming from the same ip and agent, or a series of web pages viewed by the same user , where there is no pause longer than 30 minutes between them. Clearly this is a purely arbitrary definition , although widely accepted in common usage. Most modern systems of statistics are based on this concept, and consequently suffer from some limitations with respect to some key points of the analysis. Some examples to better explain us:
if a user leaves the PC and then resumes browsing (of the same site) after 30 minutes, the two sessions are considered two separate visits; this also applies when the page is left in another window, and consulted later (just think how often we leave tabs open in the browser, even for a few hours);
if a user interacts in any way with the site without leaving the page for more than 30 minutes, and then continues browsing (for example in the interaction with a flash or ajax object, even complex: think of the webmail interface or an online game), two separate visits are considered again;
moreover in both of the above cases the duration of the visit is counted up to the arrival at the last page before the “break”, while it is clear that in the second case it lasted much longer.
In this scenario it is clear that the results of the analyzes, based on shared assumptions and standards, may vary if the definitions change.
Returning to the bounce rate, for the reasons explained it is not so easy to establish with certainty what the “visits consisting of a single page view” actually are.
But an even more serious problem inherent in the definition itself is the determination of which of the visits of a single page are actually abandoned visits (= unsatisfied user ) and which are visits that have ended successfully , for which the user has immediately found that he was looking for, without having to browse the site (= satisfied user). The technical limitations of the http protocol imply that the only way to understand if a page has actually been abandoned or not, depends exclusively on the possibility of tracking any navigator action after arriving on the page. In short, there is no way, using only the criterion of the pages viewed, to distinguish the navigator who immediately leaves a page, as he is not satisfied, from the one who no longer gives news of himself because precisely on that page he found everything he wanted from life:)
Just think of users from search engines, who land on a specific blog page for a relevant search. They read the post, maybe they even save us in the bookmarks or in the aggregator of the feeds … and they leave. The analytics system has a bounce, but in reality it can be a valid and successful visit. The same goes for news sites (what matters is the news itself), forums , information sites or sites that provide multimedia content or gamesin flash, in short, in all cases where the content of the site can be enjoyed with a single page visit. Very often visits to this content start from the search engine, and end as soon as the need has been satisfied. In some cases even a too low bounce rate (and a high average of page views per visit) can be worrying and symptom of a problem: think of the technical support site for a product, we want users to find what they are looking for faster possible.
On the other hand, the bounce rate is a reliable parameter in evaluating the effectiveness of an e-commerce site (where it is hoped that the browser does not stop on the first page), or in any case in which there is a good precise “expected action” by users, in other words a conversion .
In conclusion, in the case of the bounce rate as with most web analytics metrics, to determine whether a value is good or not, it is necessary to specifically evaluate the type of site and the underlying business model. Only by comparing the parameters with the objectives that have been set will it be possible to decide whether 70% of bounce rates is an alarming figure or the signal that we have excellent content!;-)
Recently Google Analytics, like other paid tracking systems, has also introduced native tracking for dynamic objects , such as flash. Tracking interactions with pages that don't involve loading a new page is already a great way to solve the problem for sites with this type of content.