How to identify website and app false traffic

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

The nature of traffic cheating

1. What is the difference between man and machine?

Real traffic is a record of real human behavior.

This is a bit of a detour, let me explain: the real people are very well understood that our target users are of course the real people; is the real network behavior with real purpose and motive! Their browsing, registration, access, departure ... The behavior of these columns is true.

We can tell whether the traffic data is cheating by observing the data and inferring whether the behavior behind the data is real and natural behavior.

The essence of cheating is to imitate human behavior!

Yes, if the cheater simulation succeeds, then we can not identify its cheating! Cheating is always cheating, but the means are different! and the identification of false traffic with the cheat means to improve and increasingly difficult!

2. What data do customers look at when they settle?

How do cheaters cheat? This depends on what data the customer is looking at in the checkout!

If it is the primary flow data, such as IP, PV and so on. Then the cheater will focus on the brush base flow;

If the customer wants to see the event metrics. For example, the advertising clicks, the cheater will be against these event indicators to cheat;

If it is some more in-depth, long-term indicators. For example, during the cooperation period of return visit, retention, active, etc., then cheaters will be against these to cheat.

3, cheating four elements!

IP, cookies, browsers, Click is cheating four elements!

Cheaters cheat means, no outside around the four elements of the fuss!

IP: Primary cheaters will not think of to replace the IP, this time only to detect the IP situation can detect cheating;

Cookie: The beginner cheats are also not going to continue to erase cookies, this time as long as the detection of cookies can detect cheating;

Browser: Recently Zen is doing app cheat analysis. Found that a lot of cheaters directly with the computer UA to brush the flow, even analog phone UA do not bother to do. At this time as long as the test to see if the phone UA, you can detect cheating;

Click: If the cheater replaced the IP, clear the cookie, analog mobile phone UA, how to do? Combined with click and three other elements to analyze. When the combination of standards more complex, the longer the analysis of the time, cheaters cheat behavior the more hiding!

Ii. examples of traffic cheating and how to identify and prevent

1. Primary Cheating

Tao Chan is only to observe the characteristics of the user can find its cheating, known as primary cheating. This kind of cheating is characterized by not even the user characteristics of the simulation. When you analyze the characteristics of the user, you can see the cheat data at a glance.

For example: The app's data shows a PC's UA; a lot of PV has only a small amount of IP; a cookie keeps brush flow;

For example, the user's address is concentrated in an area, and the user's device is focused on a brand or even a certain model of the phone.

The above mentioned, are static, fixed, user characteristics, or called user information. This information, as long as the data to observe, you can find out whether false.

2, intermediate characteristics cheating

When the user simulates the mobile phone UA, automatically replace the IP and disperse the IP location, or even use the meat machine to do the click; At this time, simple features and information analysis can not recognize cheating. It is necessary to further analyze the user's behavior that is click to identify cheat data.

The user's click. itself into the number of clicks, the number of distribution, click Time, time distribution, click trajectory, trajectory characteristics, and so on, if the click and user characteristics, such as IP, can be more comprehensive analysis.

The amount of data that needs to be analyzed at this time is even bigger and longer. A large number of data, professional analysis tools, professional analysis methods, professional analysts are required;

3. Advanced Behavior Cheating

What if even the user's behavior can be cheated? What if you can't find it through feature analysis and combinatorial analysis?

First of all, the cost of cheating is too high, the general cheaters do not have the strength to implement. Even if we do, we have the last killer!

Check the value of users! That is, each click, each user, and the last click, this user for the site to bring the true gold and silver hook! The industry chain of web revenue, the last is a deal! or E-commerce, or the value-added services of the website. Advertising clicks can be deceptive, but the electricity quotient and value-added are largely impossible to cheat.