Video tutorial
Distribution Analysis Explanation Video
Overview of distribution analysis
Version 1.6 upgrades "Return Visit Analysis" to "Distribution Analysis." Distribution Analysis not only tells you how much users rely on your product, but also shows you the user distribution for a specific event metric. For example, you can view the user distribution for orders under 100 yuan, between 100 and 200 yuan, and over 200 yuan.
Specify a user behavior event, and then select the event's metrics. Distribution analysis can help reveal the following questions:
- Did users increase their daily product usage before and after the notification policy adjustment?
- Will users make repeat purchases after their first purchase?
- If we consider users who use a key function more than 3 times a day as core users, what is the trend of changes in the composition of core users?
View application examples of the distribution analysis function
Distribution analysis by time statistics
Select user behavior
Choose a specific event as the behavior to be examined. For example, choose "sharing via mini-program".
Select Indicators
Here you can select the indicator for distribution analysis. You can choose "hours" or "days" and set a custom grouping interval. The following example uses "hours".
Select time unit
Select the statistical time unit for the behavior (select by time period or frequency in Figure 3.3). Options include:
- Within one day: User behavior distribution during the 24 hours from 0:00 to 23:59.
- Within a week: User behavior distribution over the seven days from Monday to Sunday.
- Within one month: User behavior distribution from the first day to the last day of different months.
Set the filter criteria for the selected events
For example, only view the behavior distribution of the iOS version.
Set user filter criteria
Filter the analysis targets based on user attributes. For example, only view the behavior distribution of female users.
Behavior distribution table
The first column is the query time period selected by the user, divided by the time unit of event occurrence (day/week/month). The second column is the total number of users who took action within the corresponding time range. The following columns show the number and proportion of users who took action at the corresponding frequency.
Hovering the mouse over a cell will display the specific meaning of each cell's number.
Clicking on this number will display detailed information about these users, and even more details about the specific behavioral sequences of individual users.
Custom grouping range
If the selected attribute here is a numeric type, you can customize the grouping interval. If not set, the query engine will dynamically calculate the grouping interval. This setting only applies to the current query and will also apply to bookmarks after the query is saved.
When you hover your mouse over a cell in the table, a download link will appear. Click it to download your user details.
Set to display metrics simultaneously
In distribution analysis, you can add metrics to display simultaneously, and simultaneously analyze the performance of the metrics for the corresponding user groups. For example, while viewing the distribution of purchase frequency, you can also view the average spending per user.
Distribution analysis by frequency statistics
Select 'Number of Times'
Selecting the frequency count will determine the actual number of occurrences without deduplication. Alternatively, you can set a custom grouping interval.
Behavior distribution table
The screenshot above shows that 4,542 users shared the mini-program on August 3rd (Wednesday), of which 4,533 people shared it 1 to 3 times (excluding 3 times).
Distribution analysis by event index statistics
The metric selected is "total order amount". On September 13th, 1004 people submitted orders, and 3 users paid between 10 and 30 (excluding 30).
How is distribution analysis calculated?
Distribution analysis offers three statistical methods: statistics by time period, statistics by frequency, and statistical indicators by event attributes. When selecting frequency and event indicators, you can define custom intervals.
Statistics by time period
This task counts how many natural time periods (hours/days) a user performs a certain action within a day/week/month.
Let's take the analysis of the number of hours a user clicks on an ad within a day as an example. If a user clicks on an ad 3 times between 3:00 PM and 4:00 PM, and clicks on an ad once between 5:00 PM and 6:00 PM, this is counted as "clicking on an ad at least in two time periods throughout the day". If a user clicks on an ad once or more within a certain hour, it is counted as one click.
To view the number of days a user triggers a certain event within a week/month, you need to count the number of days a user has triggered a certain event within a week/month.
Statistics by frequency
It counts the number of times a user performs a certain action in a day/week/month, recording each occurrence as it happens.
Statistics based on event attributes
This function calculates the statistical values of a specific attribute for events that occur to users within a day, week, or month. The statistical indicators for this attribute are consistent with those for event analysis, including sum, mean, maximum, minimum, and duplicate count.
FAQ
What are the advantages of distribution analysis over simply looking at daily active users (DAU)?
Daily active users (DAU) only tell you about changes in the number of users, but distribution analysis can reveal changes in the degree to which individual users depend on the product. For example, if a product doesn't see a significant increase in active users during March and April, but the frequency of key user behaviors (such as placing orders or posting content) increases significantly, it indicates that the product's value to users has increased. Conversely, if DAU grows rapidly, but the frequency of these behaviors is lower than before, it's very likely that the newly added active users haven't truly experienced the product's value.