App access path and sales attribution analysis

User browsing path is the key to analyze user behavior, and also the key of page analysis. Although the PV data can be easily obtained, it is not possible to know which pages the user is browsing to, and what other pages the user has browsed to the current page. such as the question is: how much sales of a specific activity, a page for the subsequent page drainage how much PV. Sales attribution is based on access path completion and is used to evaluate the contribution of each page to sales. For example, the user completes 100 pieces of sales, from which entrance which activity brings.   Sales attribution is actually just an application of the access path, so before you do the sales attribution, you need to complete the user's access path first. In the underlying data, 80% per day of user access behavior is a regular path, but the remaining 20% of the behavior will be very complex, there will be a variety of unexpected situations. For routine analysis, 80% of the regular path is sufficient, and when the analysis goes into refinement, that 20% is the key. The key three points in the app path are: The judgment of the homepage, the determination of the shopping behavior, and the decision of push.   Home decision      When selecting the source data of the access path, it is unavoidable to lose the browsing record before the critical point, such as the data before 0 O ' Day. After this part of the data, it may cause the user all the path of the day confusion, such as the user the first page of the day is an active page, all subsequent paths will assume that the active page is the homepage, so that the day all sales have the active page drainage effect, which will exaggerate the effect of the activity.      therefore must determine a home page, the definition is that when the user normally open the app will be seen, the user in the normal browsing process will inevitably go through the page, but the splash screen page and start page should not be the homepage. For simple processing, users can enter the first page of the app. But in the actual application is not so simple, the company to do channel page operation, the app all the main page is a channel page, channel page name can be arbitrarily modified according to operational policies. This leads to the concept that there is no home page in the app, and any pages are likely to be home. The better solution is to replace the home page with the concept of home, any pages placed on the homepage is the homepage.   Shopping behavior Determination      shopping behavior refers to the user in the process of browsing, click Add Shopping Cart or click to buy a similar behavior, the behavior does not represent the sale of goods, but the end of the purchase of a product. The difficulty of this step is to know exactly which page the user is to complete the purchase, to which item finished the purchase, this time the browsing path is how. Shopping behavior belongs to the behavior log, the behavior log and the page log is generally separate storage, to know exactly what pages users are browsing after the completion of the purchase, you must be the behavior and page Accurate AssociationUp to form a correspondence relationship. The accuracy of the associated field directly affects the final accuracy, which generally requires the error rate of the associated field to be less than 1%.     In addition to the usual options, there are some special pages of the purchase, such as the history of shopping cart page re-join, the order page to continue to purchase, the purchase of the collection page and other special methods, these methods need to consider  push's decision       in the app, clicking Push can go directly to a specific page, which has a big impact on the path splitting. If you do not identify, from the path to see a variety of unthinkable page upstream and downstream relationships. The best solution is to mark the push click directly in the PV data, and the subprogramme is to find the landing page of the push click through the push click Log and the page Log Association, the most unlikely scenario is that the push cannot be directly identified from the data, and only the rules can be set to identify possible push clicks.   In addition, there are other difficulties and:1.  in order to increase the user experience, the app will be decorated with a lot of function shortcuts, such as button hover box, navigation bar, etc., these portals will increase the difficulty of the path analysis. 2. Access path analysis is based on PV, the accuracy of PV is very important. Data loss, duplication of data can cause the user to clutter all paths throughout the day

App access path and sales attribution analysis