Consider the--gomez case of website effect with the measure of speed

Intermediary transaction http://www.aliyun.com/zixun/aggregation/6858.html ">seo diagnose Taobao guest cloud host technology Hall

"Lead" site traffic, bounce rate, and other user satisfaction pointers will be affected by your page opening speed? Gomez's website speed analysis will give you a new perspective from the time dimension.

Gomez this name is not unfamiliar to everyone, have been to the August waw friend must be the Gomez of China's general manager of Mr. Cheng's wonderful speech left a deep impression. Gomez is an American company, is the industry-leading (Internet/mobile) site user experience Management service provider, can provide users with customized technology platform to help test and evaluate the user experience of using the site (including usability and usability, etc.), and thereby help users optimize the site, improve performance.

The Gomez approach is to simulate the user environment from the client, accessing the Web server through a large number of clients distributed around the world to  data related to the access. The client itself is configurable, so it is possible to configure a variety of environments based on the needs of the test, such as different computer hardware, different operating systems, different browsers, and so on, to completely simulate the various real end-user access scenarios.

Case】

I received the customer's task, an important task, to study why the customer each country's site, the same design and content of the page, only the Chinese site bounce rate the highest? We all know bounce rate is one of the most annoying metrics, so when I take this job, I don't feel relaxed. But eventually, when the closing report was made, I felt full of harvest. Here, I would like to share with you the part of the study that utilizes Gomez services.

For web analytics, the usability and usability of the site, we can actually through the site analysis of the data to obtain indirectly. For example, if a page, we want users to click "Register", but a lot of users do not click, but click on the other links on the page, then we can actually know the availability of the page or usability problems. But when I analyze why bounce rate is so high, I can't just blame it for the poor design or bad content of the page, and I need to take it into account at the speed of the client download, especially when this page is made up entirely of flash elements The Code for Web site monitoring has been executed while the HTML page is being downloaded, but Flash may take about 10 seconds to appear, and this long wait is entirely likely to allow some visitors to give up.

Since the Web site's server unification is located in the United States, and the Chinese version of the page is almost all the readers in China, which may also cause Chinese visitors need to spend more time downloading. The next thing we need to do is validate the idea.

[Copyright belongs to author Sidney Song All, welcome reprint, but please inform the author in advance and indicate the origin]

"Testing with Gomez"

My authentication method is to select different countries and different cities, statistics on the average time that visitors in these regions spend on the page, and then compare with the average bounce rate in each region (bounce rate's specific subdivision data requires a Web analytics tool statistic, Site analysis tools for geographical division is not completely accurate, but still good reference data, the author note. Now, on the page download speed, Gomez can help.

Nonsense do not say, directly on the map (because of customer confidentiality requirements, the following data is not entirely from the real data).

The average response time of the test page in each country (China is very red)

As you can see from the graph, Chinese visitors are significantly slower to access this page than other countries, and only 1/2 of visitors from the United States do not have access speed. Then we compare the same page to these few countries between the bounce rate:

Graph: The relationship between page loading time and bounce rate--breakdown by country

Judging from the results, it seems that bounce rate is likely to be slower due to China's load time, but the conclusion cannot be too early. Germany (Germany) is clearly a special case.

The comparison between countries is obviously too extensive and we need to be more subdivided. Gomez provides us with further data, as shown in the following figure:

Figure: Test the average response time of a page in each city

Now, what we are going to do is to further pageload time with the bounce rate of each city.

Graph: The relationship between page loading time and bounce rate--breakdown by City

From the data, most cities do show faster downloads and lower bounce rate, but Beijing and Frankfurt (Frankfurt) are the exception.

It's our course to find the conclusion of the phenomenon. What conclusions can be drawn from the above phenomena?

Overall, bounce rate does have a connection with the speed at which the site downloads. But we know that the download speed of the website is only one of the important reasons that affect bounce rate. The preliminary conclusion we can draw is that increasing the download speed of the page helps to reduce bounce rate.

But, the careful you certainly can see, I said is "The preliminary conclusion", but this conclusion actually already some "perfect". However, the abnormal "exception" is the truth that we want to dig further, this is not the theme of today's article, save until later. and about the page download speed on the impact of bounce rate, we can actually continue one step to get more accurate conclusions.

[Copyright belongs to author Sidney Song All, welcome reprint, but please inform the author in advance and indicate the origin]

"Further testing with A/b test"

With the help of Gomez, we have learned the relationship between the load speed of web analytics and bounce rate. So how do we further validate it? Of course, A/b test, this invincible weapon, my favorite.

Omniture and Google both provide A/b test tool. I suggest choosing Google because Omniture is more expensive. The test method is simple, we prepare two "exactly the same" page, but one of the pictures and flash files slightly larger, so it will take more than 1-5 seconds to be downloaded by the browser. The rest, give it to A/b test tool chatter.

Figure: A/b Test

Finally, the results of A/b test have not come out yet, but we can theoretically predict whether bounce rate will be affected. Because the page's code will be executed before the entire Flash page, the speed will certainly affect the final bounce rate.

[Copyright belongs to author Sidney Song All, welcome reprint, but please inform the author in advance and indicate the origin]