# Behavior Analysis

# 1. Event Analysis

# 1.1 Overview of the event analysis

An event is a tracking or recording of a user's behavior or business process. For example, an e-commerce product may include the following events: users register, browse items, add a shopping cart, pay for an order, etc. Event analysis refers to query analysis based on the functions of event indicator statistics, attribute grouping, condition filtering and so on. Event analysis can help answer the following questions:

Which channel received the highest number of user registrations in the last three months? What are the trends?
What is the average amount of deposit per person during each period?
The number of independent users from Beijing who had made a purchase last week was distributed by age group?

# 1.2 Create an event analysis

Select "Behavior Analysis > Event Analysis" in the side feature bar to enter the event analysis list, which displays all newly created event analyses.

Click the "New Analysis" button to go to the new event analysis page. It consists of two parts: the data bar on the left and the condition area on the right.

# 1.1.1 Add event indicators

Start by adding event indicators to the analysis. Select events, indicators. Up to four event indicators are currently supported in the same analysis.

(1) Set indicators: Each event can count the following indicators, which can be directly selected:

Number of people: The number of deduplication users that triggered the event.

People: The number of times the event is triggered.

Number of times per user: The average number of times each user has triggered the event, i.e. the total number of times / deduplication.

(2) Custom metrics: For all types of attributes of an event, you can use the following values as an analytical indicator: Deduplication count: The number of deduplication occurrences of this attribute in the selected time frame.

For numerical attributes, you can use numerical attribute as an analysis indicator: Sum: Sum the values of the attribute over the selected time range. **Mean: ** The arithmetic mean of the values of this property for the selected period of time. **Maximum value: ** The maximum value of this property for the selected period. **Minimum value **: The minimum value of this property for the selected period. **Median **: The minimum value of this property for the selected period.

You can add filtering conditions for event indicators, and you can add conditions for the default public comments and custom properties included in the event. An attribute is a field in the event. You can analyze the additional filtering conditions for attributes. The attribute filtering options are up to 10 and only support "all and" or "all or."

# 1.1.2 Global screening conditions

The filtering conditions that need to be added for all event indicators can be set here. The optional range of properties is the intersection of properties for all events.

# 1.1.3 View in groups

Select the properties in the event group to view for more granular analysis. The optional range of properties is the intersection of properties for all events. Supports the splitting of up to 2 properties.

# 1.3 Data Barnacle

Depending on the analytical conditions, the chart in the data dashboard area is refreshed in real time, showing nearly 14 days of data by default, with a maximum analytical span of 90 days.

# 1.3.1 Trend chart

Show data trend changes, showing up to 8 curves at once. When a group view exists, the group selection range is the specified indicator Top50 group for the selected time range.

# 1.3.2 Data details

Show the event trigger in a table, and if you have Group View, you can filter the specified group in the table.

The table displays up to 1,000 pieces of data, and more data can be downloaded by "downloading," with a maximum download limit of 20,000 pieces. The full data can be obtained by modifying the time frame.

# 1.4 Preservation analysis

The analysis needs to be named before it can be saved. The name must not exceed 20 words and must not be empty and must not repeat.

# 1.5 Analytical applications

# 1.5.1 Push to a sign

Push to a tab allows you to push the current analysis to a custom tab for everyday data observation.

# 1.5.2 Replication analysis

A replicable analysis can be created as a new analysis based on the analytical conditions in the analysis.

# 2. A funnel analysis

# 2.1 Summary of funnel analysis

funnel analysis is a process data analysis model that is mainly used to analyze the transformation and loss of each step in a multistep process.

For example, the complete process for a user to purchase an item may include the following steps:

1. Products Browse
1. Add merchandise to cart
1. Product settlement
1. Payment for goods

The above process can be set up as a funnel to analyze the overall conversion situation, and can also dissect the comparison from multiple perspectives, locate the cause of loss, and improve the conversion performance.

# 2.2 Create funnel

Select "Behavior Analysis > funnel analysis" in the side feature bar to enter the funnel analysis list, which displays all new funnel analysis.

Click the "New Analysis" button to go to the new funnel analysis page. It consists of two parts: the data bar on the left and the condition area on the right.

# 2.2.1 Funnel information

(1) Set the funnel type. The type includes open loop funnels and closed loop funnels.

(a) Open loop funnel: No attention is paid to the order in which a single user and a user trigger events. Only the event corresponding to a funnel step needs to be triggered within a specified time frame to be considered a transformation of that step.

(b) Closed-loop funnel: focus on the individual user and event trigger sequence, if the user specifies the time frame to trigger the funnel, And when the first step is triggered sequentially (i.e. there is a subsequence consistent with the funnel step), the user is considered to have completed a successful funnel transformation.


Example 1: Suppose the steps of a funnel are three events A, B, and C. The user has events B, A, C, B, C in the time frame, then the open-loop funnel and closed-loop funnel calculation logic is as follows:

Open-Loop Funnel: Step 1 (Event A) is triggered once in the time frame, Step 2 (Event B) is triggered twice in the time frame, and Step 3 (Event C) is triggered twice in the time frame.Step I -> Step II The conversion rate is 2 / 1 = 200%, Step II -> Step III The conversion ratio is 2 = 2 = 100%.(Note: The open loop funnel does not focus on the order of occurrence, not on the individual occurrence, only on the indicators of group occurrence)

Close loop funnel: The user has subsequences of events in the order of occurrence of A, B, C (B, A, C, B, and C) in the time frame, so the user is considered to have completed a complete transformation. For the entire group, calculating the number of users who completed the conversion / the number of all users is the conversion rate of the funnel.

Example 2: Suppose the steps of a funnel are A, A, and B in turn.

Open loop funnel: If events A, A, B, B occur within the user's time frame, then step one (event A) is triggered 2 times,Step 2 (the A event) is triggered 2 times, Step 3 (the B event) istriggered 2 times. The conversion rate of Step 1 -> Step 2 is 2 / 2 = 100%, Step 2 -> Step 3 is 2 = 2 / 3 = 100%

Closed-loop funnel: If events A and B occur in sequence within the user's time range, then because the sequence does not exist a subsequence of A, A, and B, it does not constitute a transformation; If the user successively occurs events A, C, A, B, then because of the existence of a subsequence, a conversion is completed; If the user sequentially occurs events A, B, and A, then because the sequence does not exist a subsequence of A, A, and B, it does not constitute a transformation.

(2) funnel indicators

Closed-loop funnel only supports the number of people (uv) display, open-loop funnel supports the number of people (uv), the number of times (pv) display, and can customize the index analysis:

Number of users (uv): The number of deduplication users that triggered the event.

Number (pv): The number of times the event is triggered.

Custom metrics: For all types of attributes of an event, you can use the following values as an analytical indicator:

Deduplication count: The number of deduplication occurrences of this attribute in the selected time frame.

For numerical attributes, you can use numerical attribute as an analysis indicator:

Sum: Sum the values of this attribute over the selected time range. Mean: The arithmetic mean of the values of this property over the selected time period. Maximum value: The maximum value of this property for the selected period. Minimum value : The minimum value of this property for the selected period of time. Median : The minimum value of this property for a selected period of time.

# 2.2.2 The funnel steps

Start by adding funnel steps to the analysis, selecting the event corresponding to each step. Up to 10 funnel actions are supported in the same analysis.

You can add filtering conditions for funnel steps, and you can add conditions for default public comments and custom properties included in events.

An attribute is a field in the event. You can analyze the additional filtering conditions for attributes. The attribute filtering options are up to 10 and only support "all and" or "all or."

# 2.2.3 Global screening conditions

The filter conditions that need to be added for all funnel steps can be set here. The optional range of properties is the intersection of properties for all events.

For example, if you want to see the transformation funnel for registering with Chrome, you can set the filter to: Browser = Chrome.

# 2.2.4 View in groups

# 2.3 Data Barnacle

Show the funnel transformation in the form of a funnel diagram, trend diagram, and table. Depending on changes in analytical conditions, the chart in the data dashboard area is refreshed in real time, showing nearly 14 days of data by default and supporting analysis of up to 14 days of information.

# 2.3.1 Diagram of the funnel

Display the overall conversion rate for each step by filtering criteria.

# 2.3.2 Trend chart

Show the conversion rate of each step on a daily basis within the screening criteria.

When a group view exists, the group selection range is the specified indicator Top50 group for the selected time range.

# 2.3.3 Data details

Show the funnel conversion data in a table, and if you have Group View, you can filter the specified group in the table.

# 2.4 Preservation analysis

The analysis needs to be named before it can be saved. The name must not exceed 20 words and must not be empty and must not repeat.

# 2.5 Analytical applications

# 2.5.1 Push to a sign

Push to a tab allows you to push the current analysis to a custom tab for everyday data observation.

# 2.5.2 Replication analysis

A replicable analysis can be created as a new analysis based on the analytical conditions in the analysis.

# 3. Retention Analysis

# 3.1 Overview of retention analysis

Retention analysis is an analytical model used to measure user engagement and activity. How many of the user groups that can be used to perform the initial behavior will perform the subsequent behavior.

Retention analysis can help answer the following questions:

· Has the new user accomplished the behavior that the user was expected to accomplish over a period of time?If a transaction is completed
· A social product has added a 3D interaction effect to improve the frequency of interaction between users. How to verify it?
· To determine whether a product change works, such as adding a feature to recommend friends you might know, and to see if someone uses the product for a few months more because they quickly find friends and interact with them in the product

# 3.2 Create Retention Analysis

Select "Behavior Analysis > Retention Analysis" in the sidebar function bar to enter the Retention Analysis List, which displays all new retention analyses.

Click the "New Analytics" button to go to the New Retention Analytics page. It consists of two parts: the data bar on the left and the condition area on the right.

# 3.2.1 Set initial and retained events

First, you need to set the initial behavior events and the retained behavior events for the analysis.

Initial behavior event: The event that sets the initial behavior. The first event that is triggered is used to screen the original number of people in the retention analysis.

Follow-up behavior events: Set an event that retains the behavior, triggers the person who initiated the behavior, and repeats the events that are triggered at a later time.

After a target user completes the initial behavior and completes a specific retention behavior on a subsequent date, the number of retentions is + 1; If a target user only completes the initial behavior and does not complete the corresponding retention behavior, the retention number is + 0. Two selection strategies for initial behavioral events and retained behavioral events: 1. The initial behavior event is different from the subsequent behavior event: the initial behavior is selected within the normal usage cycle, Events that are triggered only once, such as "register," "login," etc., and subsequent behaviors select events in the core path that are repeatedly triggered by the user, such as video likes, "upload video," etc. This retention analysis is most commonly used to compare the participation of new users who started using a product at different stages to assess the gains or losses of product iterations or operational strategy adjustments.

2. The initial behavior event is the same as the subsequent behavior event: an event that is expected to be triggered by the user repeatedly. This retention is used to analyze usage patterns of loyal users.

You can add filtering conditions for events, and you can add conditions for default public comments and custom properties included in events. An attribute is a field in the event. You can analyze the additional filtering conditions for attributes. The attribute filtering options are up to 10 and only support "all and" or "all or."

When the user performs multiple subsequent behaviors in the same subsequent cycle, and the property value changes in subsequent behavior, the user is attributed to multiple label values at the same time by default.

# 3.2.2 Global screening conditions

The filtering conditions that need to be added for both initial behavior events and retained behavior events can be set here. The range of attribute options is the intersection of attributes of initial behavior events and retained behavior events.

# 3.2.3 View in groups

Select the properties in the event group to view for more granular analysis. The optional range of properties is the properties of the initial behavior event. Supports the splitting of up to 2 properties. If we select the initial behavior event attribute to group by registration channel, we can see the subsequent retention of the different registration channels.

# 3.3 Data Barnacle

Charts in the data badge area are refreshed in real time based on changes in the analysis conditions.

# 3.3.1 Initial Scope

To select the time frame when the initial behavior occurred, the default displays nearly 14 days of data, with a maximum analysis span of 90 days.

# 3.3.2 Time granularity

Set the time granularity to observe retention. Including daily retention, weekly retention, and monthly retention:

· Daily retention: Count the number of users who complete the initial behavior on a daily basis and observe retention on a day-by-day basis on subsequent days
· Weekly retention: Complete target user numbers for starting behavior on a weekly basis and observe retention on a week-by-week basis
· Monthly retention: Complete target user numbers for starting behavior on a monthly basis and observe monthly retention for subsequent months

# 3.3.3 Indicator trends

Shows trends in retention rates and shows up to 12 curves at once.

· Retention rate: number of target users who complete retention actions / number of targeted users who complete initial actions

When a group view exists, the group selection range is the Top50 group of the number of active users for the initial behavior event that triggered the selected time frame.

# 3.3.4 Data details

Display the event trigger as a retained table, and if you have Group View, you can filter the specified group in the table.

By default, groups are grouped by the date of the initial action. The first column of each line represents the date of the initial action; The second column is the total number of people (number of active users) who triggered the initial action on that date; The following columns are the indicator data that triggers subsequent behaviors after the corresponding time (including the number of users, and as a percentage of the number of initial behaviors).

· Retention Number: The number of users targeted to complete retention activities
· Retention rate: number of target users who complete retention actions / number of targeted users who complete initial actions

# 3.4 Preservation analysis

The analysis needs to be named before it can be saved. The name must not exceed 20 words and must not be empty and must not repeat.

# 3.5 Analytical applications

# 3.5.1 Push to a sign

Push to a tab allows you to push the current analysis to a custom tab for everyday data observation.

# 3.5.2 Replication analysis

A replicable analysis can be created as a new analysis based on the analytical conditions in the analysis.