What is the definition of bounce rate for web analytics?

Bounce rate is a more interesting measure in website analysis, and also a measure that Avinash Kaushik individual prefers, and there are more discussions on it online. Individuals prefer Sidney translations-failure rates, transliteration and transliteration. It was always thought that the definition of bounce rate is to browse only a single page of the total access rate, whether it is Google Analytics "single PV visits/total Visits" or "the Omniture" one page Visits/total Visits ", the calculation methods are more similar. But recently, while flipping through the Web Analytics of Avinash Kaushik, it was found that Avinash Kaushik's definition of bounce rate recommended the use of time on Site, which is the proportion of visits that are less than 10 seconds or 5 seconds, I do not know whether it is because the book has a little time to publish the relationship, do not know the "Web Analytics 2.0" in the definition of bounce rate.

However, regardless of the number of visits pages or time on site to define bounce Rate, in fact, there are traps, will cause bounce Rate too high.

Traps in the bounce rate

Let's start by saying why Avinash Kaushik recommends using time on Site, as described in the Bounce rate section of the Web Analytics, Avinash Kaushik is based on practical experience that the user's stay time in the site is less than 10 seconds, generally unable to complete an effective interaction (engagement). Of course, depending on the type of site, Bounce's time on site can select an appropriate value within the 5-10-second interval.

The reason why you use the stay time instead of browsing the number of pages is that it is inaccurate to define bounce with a single page of access. Just like WordPress blog, many users can only visit the first page to browse the latest published articles, that is, although only visited a webpage but achieved the desired goal-to browse the blog recently published articles, RSS based subscription is the case, Then we cannot think of such visits as bounce. To give a more intuitive example, Twitter includes looking at recent tweets or tweets that are actually done under the same url--twitter.com, if simply using single page access to define bounce Rate, It is estimated that the bounce rate of twitter.com will reach more than 98%, which can not really embody the meaning of bounce rate index.

Since the use of page or pv=1 to measure the disadvantages of bounce, then why so many web analytics tools or use this standard? First of all, we must be clear Avinash Kaushik recommended use of time on site is based on a certain premise, that is, the site's stay can be accurately measured, and now the general calculation method of residence time is flawed, you can refer to my previous article- a description of the duration defect in the function and defect of the Web log. Of course, there are some special ways to get users to leave the site point of time, if the user clicks on the link page redirection, close the browser pop-up hidden window, etc., these technologies because of its implementation in some degree will affect the user's use, or let the user feel unwell, is considered an indecent means, Therefore, generally do not recommend the use.

So generally we are very difficult to get users to really leave the site of the point of time, some of the current popular analysis tools are also the case, which will lead to time on site can not be accurately measured, the result is that all the single page of the visits on site are 0, and those 0<time On Site<10s's visits is actually just those visits that are less than 10 seconds from the entry page to the left page, and not the real access time is necessarily less than 10 seconds (we don't know where he stayed on the last page). The time on site in this case will introduce us to a deeper trap, so most of the current analysis tools are based on their own ability to get data Select "The total number of accesses per page" as bounce rate is understandable, or more like a helpless one.

So, for bounce rate to be cautious and optimistic, maybe your site doesn't analyze the data to show it's so bad. So is there a more effective way to define bounce rate this interesting metric? I haven't found it yet, welcome back to the discussion.