How to make data speak--analysis of website example

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

Editor: s++ Group

Data is seen in many websites as one of the basic metrics for measuring a product or a design. Data indicators have also been pressed for a long time I was breathless. But now think about it. Sometimes the data tells you a lot. It's not necessarily the only standard to measure the quality of a product, but it does tell you a lot.

So what can the data tell us?

– Does your traffic work?

-How to find a leaky hole?

– Do you really know your visitors?

-page, is the architecture reasonable?

– Is advertising useful?

-What does the revision bring?......

Let's look at the short name of the data first.

In the following example, we will do some analysis for these abbreviations.

–PV (that is, page view, integrated pages view)

For example: When I clicked on a banner, then from the click of the page to start, after all the clicks appear on the page is the volume of PV.

–click (page clicks)

–display (Single page view)

–UV (i.e. unique Visitor, independent visitor)

Visit a computer client on your site for a visitor. Within 24 hours, the same IP, multiple visits, only counted once.

–buyer (Purchaser)

Visit a computer client on your site for a visitor. Within 24 hours, the same IP, multiple purchases, only one time.

–CTR (click conversion rate, i.e. Click/display)

-AD bit conversion rate (i.e. Pv/click)

Here's the example.

L Banner Article

-Case A

A Advertising Position conversion rate: 1.9

B Advertising Position conversion rate: 10.8★

This data tells us that in a page is not obvious position of advertising, if the text form will be very easy to be a lot of other information interference can not see, this time if put on eye-catching pictures, it is likely that the effect will be very good.

-Case Two

The purpose of the test is: the optimal performance of the quantity and size of the goods in the banner size.

The data tells us that the number of items as a banner block is not as much as the better, it will have a most suitable size and number of regular, as shown in the picture, 4 items Although the image is large, but because the choice is relatively small, the data performance is really poor, but 8 products relative to the same size of the banner picture will be smaller , the attractiveness of the user will also be affected, so 6 items in the quantity and size of goods for optimization. Of course, the test also has some drawbacks, when 6 or 8 products in the picture of a particularly popular users will affect the performance of the data. So there are only a few attempts to see the optimal combination for this test.

– Case Three

The purpose of the test is: banner in the process of revision of the various categories of performance contrast.

The test is based on the same page three revision process, Center banner data performance to achieve. So it is not through the same time conditions and the same commodity conditions to test, we took a lot of data to compare, as far as possible to avoid the different products caused by the error of banner data.

In the revision process, we first launched a version, when a version of the revision to version B, the size of the banner is relatively wide, and the contents of the product also increased, so we believe that the performance after the line will be better than a version, the results found that is not the case, B version of the conversion rate than the original a a lot of decline, At first we thought that because the new version of the online led to the old users are not accustomed to the new design caused some operational problems, but found that after 1 months of data is still not a big improvement. So we line up the C version, simply increased the size of a single banner to see the data performance, and sure enough this data is better than B is better than a performance.

The reasons for our careful analysis, think it may be because, B version to the user's choice to increase, for the overall banner so that users do not have the option to start, for the advertising category of banner may be a large single advertising will be more excellent.

L Navigation Chapter

– Case Four

The purpose of the test is: the performance contrast of each category in the revision process.

The test is based on the same page navigation three revision process, we not only compared the entire navigation conversion rate, for the page data than the fairness of the navigation and the entire page click ratio.

A version is the old version, the data for the home click:9%,ctr:4.6%. b version of the line, the data performance dropped to account for the first click:2%,ctr:0.5%. C version of the online, the data has rebounded slightly on the home page click:9%,ctr:2.7%, although the conversion rate is not ideal, but click accounted for has risen and a. D version of the online, the data finally rise to account for the home click:12%,ctr:5.2%.

We analyzed the whole navigation of the click Allocation ratio found, the most important reason is because of the channel navigation, a channel entrance is very obvious; b Delete the channel navigation entrance; C Although added back to the navigation entrance, but did not do it clearly; D adds the entry back and deletes other jamming information, making it more obvious.

In terms of navigation, the D version of the performance is excellent, but if the channel home page to increase the entry is really effective. So we looked at the Channel page buyer conversion rate, found that the channel home buyer conversion rate performance, although not better than the process page, but relatively good performance, relative to some other places on the home page, such as the community and some advertising position, channel navigation conversion rate is relatively high, So finally we decided that for navigation, the entrance of the channel's homepage needs not only add, but also obvious. So the D version is a better choice.

L Page Chapter

– Case Five

The purpose of the test is: Home revision of the market to promote the position of the performance than the

The above data table can be seen after the revision, CTR up the module is 2, 3, 6, so we looked at 2, 3, 6 of the buyer conversion rate, the results found in addition to 3 of the higher performance, the other two are data buyer conversion rate is very close, and many buyer conversion rate of the high module, But the Ctr drop is very obvious, so the market promotion module for this revision is a failure.

It is more obvious from the following summary table:

Although HP's page display grew, other data dropped. But in the end we don't have a fallback version, for a variety of reasons, one of which is the addition of the automatic maintenance module in the new version and the buyer growth of the module is very obvious, and the new version of the keyword part of the adjustment also makes the entire page performance growth is very obvious, As a result, the design and location adjustments were made to some important locations, but the entire page was not rolled back.

– Case Six

The purpose of the test is: This is a multiple version test with the same time, with the merchandise, with the link page to compare the performance of the four versions.

The test is for the marketing of landing page to do, in terms of CTR, a version of the conversion rate is superior to other versions, due to the focus of the commodity area, other LP BTU information is relatively weak, less interference. The D version not only uses two-column structure, but also increases the visual effect of LP BTU, which makes the click of the D version of LP BTU larger than version a. As shown in the following illustration:

(LP BTU refers to the landing page and other landing page link)

In terms of Design CTR performance, A is really better than D, but because the page is done for the marketing, and marketing department is not only to see the good or bad of a page, more important is to see the effect of the overall promotion, so the D version on a single page performance although the general, enhanced LP BTU part of the increase in the impact of the mutual chain, Therefore, the overall page to bring the promotion of the page conversion rate increased, and make its performance within 24 hours of the PV effect is not general. Finally, the D version is used.

L How to treat data correctly

-Case A

When the boss set an outrageous data index, the departments in order to achieve the target each of the various means of coping. These instruments, however, have led to a final false indicator that does nothing to help the company's performance.

This explains: The data is not intended to create unrealistic metrics.

In the development of data indicators should have a reasonable increase in the analysis, and should not be shot in the way the head of the decision.

-Case Two

When a more creative project on the line, and three months of data performance is not very good time, whether need to immediately remove the dismount it? Although the quality of the data needs to be looked at, but may not necessarily look at the absolute value, and the growth trend is also very important link. And a large project in the online before should do more adequate preparation, but once the line, if not good enough, can also be adjusted by some users of the research again to revise it, immediately removed dismount is not necessarily the best way.

This explains: Many projects to see long-term data trends, in the process can do some other uer research to improve it.

– Case Three

Each department has a different set of indicators for the department, this is the emergence of contradictions, such as the marketing department feel good, because they have their own data analysis, and the design department will have their own set of data analysis, the conclusion may be completely different, business is the same. This brings a lot of bickering, so how to avoid these quarrels. There must be a maximum target value, such as the end of the buyer as the final indicator value, then on this basis for other data optimization.

This explains: The data needs to be considered in an integrated way, and different angles will have different conclusions.

– Case Four

This is one of our former leaders often hung in the verbal words, he always said that with the data, Uer can not be all. But finally one day, one of us, a super bull-fork, Uer retorted on his remark, which led him to never again dare to mention this argument. The data tells us where the problem is, and user research can tell us where it is. Here are two where I feel very classic.

The end: The data is a double-edged sword. Although the data analysis is very important, but also the owner of each company very much the part, but it needs our reasonable analysis, comprehensive consideration. If its analysis is wrong, it can lead to a fatal error in decision making. So the data need our careful and comprehensive thinking.

This article author: s++ reprint please specify from: Ctrip ued