Buried program data
Buried data program is divided into three portions
- Set trigger conditions
- Mapping relationship management
- Acquisition Rules reported treatment
Shence data Buried learning program
For Shence data, the underlying data model used is "Event + User" event model, so buried in Shence data point called the event here. Buried requirements document called event design.
Three core events (Buried) design
- The event is split into single user clicks and browser actions
- The goal of the action need to be analyzed transformed into an event
- Combined with business objectives analysis, design events
1. The combination of scene design event
Such as airline tickets and submit ticket orders submitted, in the design of whether an event designed to the same scene or dealt with separately?
Two design ideas: both the scene and the little difference when analyzing the scene is usually the overall analysis, can be designed to the same event; each event scenarios vary widely, the analysis of multi-scene analysis, can be designed for different events.
2.Session analysis
For the website, a series of user behavior, is a visit, called a Session.
For user behavior we generally record 4W1H model: Who who is accessing, When when to visit, Where location, What the user what to do, how to How to access
Based on the behavior of such a user point of view of the record, whether it is a mall, or a website, businesses can know what the user did, such as when to enter the site, when to buy something, and so on.
But for some like this demand, for example, the average user will come several times, each time for how long on average, with an average every visit a few pages, and so on, these analyzes require the continuity of the message, the user will need a single point the behavior of the series as a whole.
Key Analysis Session consists of two aspects,
Session events which behaviors should be included
How to cut Session: Set the length of the cut, that is greater than the spacing between adjacent long event is to be cut.
For example, a user first opens the electricity supplier website home page and then search, enter the product page, the final product was added to the shopping cart, generating orders, final payment orders and so on.
Traditional statistical tools will only collect action page views, so the Session component contains only the home page, product page and order pages three events, but God making all event data can be collected, including the home page, search, product page, etc. Add to Cart long time so when calculating the user to stay long in the commodity page, you can ADD tO cART time and commodity page subtract time, accurate stay.
Traditional statistical tool cutting length is fixed, for example, the PC 30 minutes, the APP end of one minute, assuming that the user open the home, to switch back, then back to the APP after a certain time, for example two minutes, at this time, Session was cut into two sections, the first page that opens when switched to 0 long stay, and long depending on the business, industry and the need to adjust cutting, Shence data provide long cutting configuration.
For example, video sites, cutting assume Session duration is 30 minutes, when a user to browse the site for 10 minutes, then open a video, while watching the video phone with his girlfriend one hour, end the call and then continue browsing Web page.
Session user computing time duration is shorter will be recorded deep enough.
And if this time cutting Session length is 40 minutes or 1 hour point it is able to access the situation is more true reflection of the user.
