Among all the analysis problems faced by data analysts, there is one type of analysis that is very expensive, but it is often encountered in business. This kind of analysis is attribution analysis. Let’s look at the following two cases: Case 1: Xiaoguo saw a Moments advertisement on her mobile phone and released the latest Apple mobile phone. During her lunch break, she swiped Douyin to see an internet celebrity reviewing the latest Apple mobile phone. The little friend received the mobile phone and was posting pictures, so he drank a cup of Jiang Xiaobai boldly and went home to apply for funds with his wife. In the end, his wife approved him to buy it on JD.com, which is guaranteed.
So the question is, how much did the mobile number list advertisements in the circle of friends, Douyin, the circle of friends, and JD.com contribute to this transaction? Case 2: Xiaodan bought a pair of basketball shoes on Taobao, and found AJ through the homepage search. He clicked in to see the style and color. During the May Day, Xiao Ming opened Taobao again, saw the promotions on the homepage, clicked to enter the event sub-venue, saw AJ again, clicked in and had another eye addiction, thinking about the living expenses for the next month, and reluctantly quit. front page. However, fate is so magical, I saw AJ again on the homepage - guess you like page, clicked in to watch the reviews and buyer show, no matter what, I'm going to buy it.
So the question is, how much did the search on the homepage of Taobao, the event venue, and the resources in these sites contribute to this transaction? With the rise of the mobile Internet, business forms are becoming more and more complex, and the demand for attribution analysis is increasing. The above two cases are the attribution analysis in the two classic scenarios of off-site channels and on-site resource locations. The scenarios are different, but the purpose is the same. That is, according to the current scene and goal, how to reasonably allocate the "contribution" to each pit.
In fact, there is no standard answer to this kind of question, because the real business is complex and it is difficult to accurately and reasonably allocate contributions, but the demand for attribution analysis is so high frequency and requires strong timeliness, so it needs some With the support of the methodology, we can quickly try and locate the problem quickly without being confused when facing the problem. B does not know where to start. Here are some common attribution analysis models for reference. 2. Common attribution analysis models 1. Last Attribution Model Also known as the last click model, this attribution model assigns 100% of the credit to the last channel before conversion, i.e. no matter what the user does, only the last one. This is the simplest, straightforward, and most widely used attribution model.