Statistical data reference value for the current business: index
So what is a valuable reference for business do?
When I went to look at a business, I usually want to know?
Who did what?
What happens after finished result?
Usually it needs to answer user data, behavioral data and business data.
User data indicators:
Nikkatsu (DAU):
Statistical data system definition: event-based reporting
Definition of business: Based on the key critical event reporting (to keep maintenance day live event list)
Understand User:
Parade: every registered user a unique unique ID, but unregistered users will be missed
Identify equipment: a string of unique identifier for each device, not behind the corresponding user equipment
Recognize people or recognize the device?
① Are there account system?
No-> recognition device
② business scenarios depend on whether the landing?
Yes-> recognize people, identify equipment
③ not logged in user traffic is valuable?
No-> recognize people, identify equipment
Yes-> recognition device
New users:
1. Select the appropriate node, the definition of "increasing"
2. distinguish the "new" by appropriate means
Based equipment: ios, android, web
Based on the account associated: Have an account comparison with the background
User Retention:
Why look at it retained?
Understand the quality of a particular channel - Japan retained
In days, a measure of this channel to the user the current and subsequent performance
To [x day retained] as a standard for comparison, to avoid interference from other Japanese data
Japan retained = Specifies the number of daily active users / 1 day active users * 100%
The next day retention = 2 Number of active users every day / 1 day active users * 100%
7 Retained = 7 day active users / 1 day active users * 100%
Observe the entire market - Week Retained / month retention
Weekly / monthly basis, a measure of health products, to observe the user stickiness platform ( be sure to weight )
Week (May) Retained = specified week (month) Week (May) number of active users / Week 1 (January) Week (May) the number of active users * 100%
Second week week week week retained = number of active users / week week number of active users * 100%
Month = month retention times monthly active user / month monthly active users * 100%
Behavioral data indicators:
The PV: the PV (Page View) page views - number
UV: UV (UNIQUE Visitors) number of unique visitors - The number of (de-emphasis)
Possible conversion was calculated PV / PV, UV / UV
Depth of Visit:
Algorithm 1: number of users access to certain key behaviors
Algorithm 2: Web site content / functionality is divided into several levels, this user visited the deepest level computing
Visit Length: Web era - long when the page is opened (inaccurate, do not open operations)
app era - long resident Reception
Note that through the pupil and identification - to see if the camera Note screen pupil
How long visit to statistics:
① by statistical special events, support business needs
Eg video statistics are consumed extent, evaluate the quality of content (after recording pause on / off the page, playing the current position in the video progress bar)
② counting function display
Perform basic analysis Metabase by BI tools through scripts and codes statistics log
Pop-up rate: popular talk refers to the left immediately after a user has come, a conversation based on a single visit
Pop: both in terms of a Web site, the user visits a page left
Pop-pop-rate = the number of users / number of active users
Eg, if a user accesses a pop-up four times five times, the pop-up 80%
Business data indicators: