The algorithm problem in the process of email marketing data analysis

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

Last week and a brand customer on the mail marketing after the data analysis issues launched a series of discussions, customers put forward a lot of email marketing indicators of the question, aroused my thinking. Now put these questions and some of their own thinking and share with you, I hope that in the mail marketing process to correctly analyze the data to play a certain help.

I take this client as an example, the customer is a one-month delivery volume is maintained at about 500,000 of FMCG companies. Here are some of the data from his email marketing analysis in the last one months:

Post total: 500578

Post success: 495000

Open message: 200345

Effective clicks: 81650

Usually when we see these key numbers we come to a few conclusions:

1. The arrival of the mail is also good, in addition to some mail full, or DNS error, or some other recipients of the problem, basically arrived, it should be said very good.

2. Mail opening rate of 44%, for the mail marketing is a very good result, this is mainly due to the user sent the object is mainly his members, loyalty is very high, at the same time the promotion of the Mail has joined the membership of the user name set the dynamic title. So the opening rate will naturally be greatly improved.

3. Effective clicks in 16%, this ratio is not high, indicating that the content of the message and template has yet to be improved, or the user's interest is not very large.

This analysis should be in line with the actual situation, but the client has raised a lot of new questions, these questions are not before I have to think deeply.

1. How to be regarded as a delivery success, is the system to send out even if successful, or the mail service provider there is a successful response, if it is to send the system, if the mail service providers there filtered some, this number is certainly not accurate. If it is based on the data feedback from the server vendor, the final message sent out is in the user's Inbox or spam box or ad mail box. In any case the resulting figures should be getting bigger.

2. Open rate 44%, because now the message open analysis mostly or use analysis log log method to solve, such as in the template with a transparent picture of the picture was read on behalf of the message was opened. This method of analysis is no doubt inaccurate, because many client versions of the mail receive software or Web version of the shield of the image display, which is often a small number.

3. Mail clicks 16%, this is based on each individual click to calculate or as long as the user to click on a link in the message to record once, if it is clicked on a link on the record once the number is much larger.

In the end you'll find that the results of the analysis and the depth of your thinking come down to a huge difference, and the key to the difference is how these numbers come out, but it's often easy to take you to another direction by showing the analysis.