Basic concepts
With the analysis module, you can quickly create specific metrics based on aggregate calculations of events and properties, display the resulting data as charts, observe metric trends, and run multidimensional cross-comparison analysis. This article introduces the core concepts of the AE analysis module.
To learn more about events and properties, see Data management.
Analysis metrics
You can quickly build the analysis metrics you need in the metric configuration area on the left of the analysis model page, and get data results within seconds without writing complex SQL statements.
In the Events Analysis model, analysis metrics are defined very flexibly. You can freely combine events and properties according to your definitions, and you can also use arithmetic operations to build formula-based analysis metrics.
For models such as Retention Analysis, the calculation logic of analysis metrics is predefined, such as retention rate, churn rate, and number of converted users. You only need to select the corresponding events and set the time range of the statistics to get results.
You can also build a project metric system directly and use the created metrics during analysis to keep definitions consistent. For details, see Metrics.
Date range
The date range determines which events, by the time they occurred, are used to calculate analysis metrics. The same metric produces different data results for different date ranges. Except for Composition Analysis and SQL IDE, all models require a date range for analysis.
You can use the built-in options on the left to quickly select a rolling date range relative to Today, such as Last 7D or Recent 30D. You can also use Custom to configure the rolling date range you need, such as the past 4 months, this quarter, or last year.
If the shortcut options don't meet your needs, you can also select the start date and end date separately on the right. Besides rolling dates, you can also select Exact Date, a specific date that doesn't change with Today. When you select Exact Date, you can choose a date later than Today. When the end time is an Exact Date, the start time can only be an Exact Date as well.
The system records the date ranges you have used, and you can add them to favorites for easy reuse next time.
If your project has commonly agreed date ranges, such as the period of a specific campaign, you can save these common ranges in Project Settings > Date shortcut settings so that all members of the project can use them.
Time zone
When the users of your product are distributed across different time zones, you need to understand how time zones are handled in analysis. If you need to collect user data from multiple time zones, record the user's time zone when the event occurred in the time zone field when reporting events, and enable the multiple time zones feature in the project time zone settings.
During analysis, a display time zone must be selected to keep definitions consistent. AE uses Coordinated Universal Time (UTC) as the standard for recording and displaying time zones. The system shifts the event time (#event_time) of events from different time zones around the world to the display time zone based on the time zone field reported with the event, and then calculates, avoiding inconsistencies caused by time zone differences.
If the time zone field reported is invalid, or the selected event doesn't have this property, the time is not shifted.
For example, a user in UTC-8 registers at local time 2022-12-31 20:00:00, and another user in UTC+8 registers at local time 2023-01-01 12:00:00. The two users actually registered at the same moment. If the display time zone is UTC+8, both are counted as new users on January 1. If the display time zone is UTC-8, both are counted as new users on December 31.
- Besides event time, other time-type event properties are also shifted to the display time zone, including custom properties and dimension table properties
- Time-type properties used in custom property rules use their local time values, that is, they are not shifted
- Time-type user properties are not shifted
Entity
Whether you analyze user retention rate, conversion rate, or distribution across ranges, you first need to define what a User is. In AE, a User is also called an Entity.
By default, analysis models use the AE user ID (#user_id) for analysis. You can also create custom entities in Project entities and use them in models, tags, and cohorts.
Suppose one AE user ID corresponds to one character in your game, and one account can create multiple characters. When the Distribution Analysis model analyzes login days by Account, the account is considered to have a login event on a day as long as any of its characters logged in that day. In Retention Analysis, you can also observe from both the character and the account perspectives to get more accurate results.
As with the AE user ID, you can drill down to the behavior sequence page of a custom entity to view the detailed behavior of each entity.
Learn more
After you've mastered the core concepts, read the following sections to learn about the common logic of the models: