A brief talk on web analysis

Remember a long time ago to see such an introduction:

Imagine a bookstore on the edge of a shopping mall, and if you end up leaving without buying any books, the store chief won't know you've been here. But if you buy a book, the staff of the bookstore will know that they have sold some goods (of course, if you leave contact information there to book another book, they can get more).

Looking back, we look at the site from the point of view of information gathering, and that's another very different world. Whether you buy something or not, you always leave a lot of records when you visit the site, and by collecting a lot of the data left by these visitors, our network experts can get a lot of conclusions about the user experience of the site.

From the site's record, you can know each channel you walk through, click on each link, what products you take, what topics you are interested in, and so on. Even if you know what you've been searching for, the online advertiser will give you the "likely baby" based on the product you're interested in.

This is the simplest web analytics concept, so what is the ultimate motivation for web analytics?

What does your customer need? He does not need the "usability" and "user experience" itself, what he really needs is to achieve his goals and achieve his motivation to visit.

In a narrow sense, web analytics means analyzing the behavior of visitors to a Web site; In a broad sense, web analytics refers to evaluating, adjusting, and adapting the various aspects of the site to meet the company's business objectives.

In other words, web analytics, the ultimate motivation is not "reporting", nor is it figuring out how to send spam messages full of data to policymakers. The real purpose of it is to gain an awareness and measure of action.

Avinash Kaushik, in the book "Web Analytics:an Hour a Day", mentions the fact that some traditional web analytics, sometimes with YY data reports, such as "Exit most pages", "Guest screen resolution", "site interaction", etc. One common denominator of these metrics is that they claim to say something, but barely understand anything.

But in turn think about it, web analytics based on log data is really useless? The answer is, of course, negative. Log analysis, which is still the cornerstone of web analytics, provides us with the broadest range of phenomena that we can "know" from. After the "Know Why" part, we need to combine qualitative analysis and research to find out.

So what basic data can we get from the log? Take a look at the graph below, analyze and classify the collected data vertically, and this diagram also describes one of the most widely used statistical analysis steps based on logging: Log file->pv-> session-> users-> customers-> loyal customers, We can clearly see that the higher the pyramid the more the data is commercially valuable.

An explanation of some of the terms in the diagram:

A hit (Hit) is the same term as a request. In order to obtain a resource on the server (which can be text, images, or any element that can be included in the page), a single connection is made between the browser and the server to which it is connected. A record in the log file is a request.

The number of accesses (Visit) and user sessions are the same terminology. From the CNNIC definition of this term, there is no detailed definition of what is visit, what is loss, at present, a visit must download at least a full page to the client, if not full download on the user closed the window that is the end of the request, then is a loss, Rather than a visit or called session. General metrics: Visitors who interact with the site within 20 minutes are considered to be on the same site, does not record the number of new user sessions; When a visitor lasts 20 minutes without interacting with the site, the visitor is considered to have entered the site again and recorded the number of new user sessions when he visits the site again.