What are the differences between the new version and the old version, and how to migrate from the old version to the new version-Quick Audience(Quick Audience)-阿里云帮助中心

If you are migrating from the previous version (V3) to the new version (V4), or are considering whether to migrate, this article can help you understand the differences between the new version and the previous version, and provide migration guidelines, hoping to help you quickly migrate to the new version.

Note

The new version and the previous version have some of the same features. When you read and search for documents, you must distinguish the version of the documents. The documents of the new version are stored in the V4 User Guide directory.

Introduction to new features

V4 has made a major functional upgrade on the basis of V3. Before the migration, please take a moment to understand the new features:

Understanding ID Mapping

One of the important improvements in the new version is to add a ID Mapping process when importing data. The original user data from different channels is used to identify and deduplicate users by ID. After calculation and integration, users are assigned a unique identity QAID in Quick Audience. For more computational logic, see ID Mapping and Unique User Identifier QAID.

In subsequent operations such as user analysis and audience filtering, the QAID will be used as the unique identity of the user. Regardless of the table from which users or audiences come, they use their QAIDs to query all ID types, tags, and behavior data to integrate data across channels, and finally build an omni-channel tagging system, as shown in the following figure.

346u

Function advance

Because the new version introduces user identification, the new version has the ability to achieve stronger functions. The following table lists only the major feature differences. New features are being added. For more information, see V4 User Guide.

Feature	Console of the new version:	Console of the previous version:
Data input	Eliminates the concept of data sets and reduces the difficulty of product understanding. Support statistics tables are added to analyze the statistics of operations performed by users. Data tables support more redundant attributes to facilitate subsequent analysis and selection. The more redundant attributes of the user behavior table can avoid the row expansion problem that occurred in the previous version. All types of data support hour-level automatic updates. All data entry configuration is done by the administrator at this stage, isolating IT use from business use.	The data set concept is difficult to understand. Behavior datasets do not support automatic updates. AIPL/RFM models do not support hour-level automatic updates.
User authentication	The ID Mapping process is added. Multi-terminal data is used to identify users based on IDs. This enables cross-channel data pulling and application to analytics and marketing.	User identification is not performed. User information from different sources is fragmented and cannot be collaborated.
RFM model	When you create an RFM model, you can set the data statistics range: time, channel, and category.	You cannot set the data statistics range.
Custom Tags	You can create Preference Tags, Loyalty Tags, Purchasing Power Tags, and User Stage Tags to tag users from multiple dimensions.	The type of the custom tag is simple, which is only equivalent to the user stage tag in the new version.
Filter Audience	Supports flexible filtering across data tables and ID types. The free combination is calculated by the intersection and difference between the different dimensions of the cross-filter. You can make an unlimited number of audiences.	Flexible filtering cannot be performed across datasets or across ID types. You can use only intersection and merge to calculate combinations between different dimensions for cross filtering, and only intersection and merge can be used at the same time. Audience merge is limited to audiences that have been generated. Direct participation in filtering is not supported. A maximum of three layers of audience merge is supported.
Update Audience	After the bottom table data is updated and imported, the automatic update of the audience can be triggered to ensure that the marketing reaches the latest population.	Audiences support only periodic updates. Data-triggered audience updates are not supported.
Automated Marketing	Real-time behavioral event data is interoperable with existing user data. You can use behavioral events as filtering conditions anywhere in the process. A new AB test component is added to compare the effect of the solution by testing. More flexible process canvas operations, execution records management, activity version management, and more.	Behavior events can only be used as the first component to determine whether a user enters a process.

Automatic synchronization of system configurations

The new version of Quick Audience shares the same account system as the previous version. You do not need to manually configure the configured organizations and spaces. The legacy Manager system configuration includes:

level	Parameter	Sub-configuration items	Migration description
Organization Management	Manage organization members	-	Shared configurations. No manual migration is required.
	Workspace management	-
	Data permissions	Report authorization
	Data permissions	Data source table authorization	The new version does not support organization-level data sources.
	API management	Alibaba Cloud SMS and email interfaces	Shared configurations. No manual migration is required.
		Third-party SMS interface
		Quick BI authorization configuration
		PUSH interface
		Reflow data source configuration	The new version does not require settings and will automatically flow back to the only analysis source.
		Data tracking interface	The new version replaces the data collection interface configuration with QT event authorization.
		OSS authorization configuration	The new version does not support the corresponding feature.
	Organization System Configuration	-	Shared configurations. No manual migration is required.
Workspace management	Manage workspace members	-	Shared configurations. No manual migration is required. Note In the new version, the space system configuration does not include the automatic update time of RFM/AIPL models. Instead, it is set separately for each model.
	Manage workspace roles
	Space user group management
	Space system configuration
	Spatial Interface Configuration
	User Access	Access statistics
	User Access	User download details	The new version does not support statistics on user downloads.

Migration procedure

The new version of Quick Audience uses a different data storage system than the old version. If you use the new version for the first time, you must prepare data sources and tables based on the requirements of the new version. Then, you can manually migrate business data, including:

Data Source Migration
Dataset Migration
Audience Migration
Marketing Task Migration
Marketing Campaign Migration
Analysis Dashboard Migration

Note

During the migration, contact Alibaba Cloud service personnel for more technical support.

1. Data source migration

In the new version, data sources are divided into two types: computing sources and analysis sources. Their functions in the system are shown in the following figure: 数据流

Computing Source: Raw data is first stored in the computing source. After Quick Audience is connected to the computing source, whenever raw data is imported, the computing source performs ID Mapping on the raw data and then imports it to the analysis source.
Only MaxCompute is supported. You can add only one computing source for each workspace.
Analysis source: When you import raw data from a computing source, the data mapped by the computing source ID is imported to the analysis source. When Quick Audience uses this data in the future, it will be directly obtained from the analysis source, and the audience data will also be stored in the analysis source.
The type of the analysis source is Hologres or AnalyticDB for MySQL 3.0. You can add only one analysis source per space.
Note
To synchronize data from MaxCompute to AnalyticDB for AnalyticDB for PostgreSQL 3.0 or Hologres, you need only to bind a computing engine to Quick Audience, analyze the computing engine, and then import the data to the data table. You do not need to use professional Data Transmission Service tools.

You can use the previous version of ADB3.0 data sources to the new version, as shown in the following figure.

数据源

Follow these instructions to configure the new data source:

Bind an analysis source

ADB 3.0 types of data sources in the previous version can be used in the new version and bound to the new version of the space as the analysis source in the space.
In addition to ADB 3.0, you can also select Hologres as the analysis source to bind to the space.
For more information about how to bind an analysis source, see Analysis Source.
Note
The new version no longer supports organization-level data sources. You need to bind analysis sources to a workspace. We recommend that you create a database for each workspace in the purchased ADB 3.0 or Hologres instance.
Data sources of the AnalyticDB for PostgreSQL and AnalyticDB for PostgreSQL 2.0 types are no longer supported in the new version.

Bind a computing source

In the new version, in addition to the analysis source, a MaxCompute project must be bound to each workspace as the computing source in the workspace.

Your raw business data must be adjusted based on the new Data Table Format Requirements and then saved to the computing source.

For more information about how to bind a computing engine, see Computing Source.

Note

We recommend that you create a project for each space in the purchased MaxCompute instance for binding.

2. Data set migration

In the new version, you need to manually migrate datasets.

In the new version, the concept of datasets is removed, and data tables are directly imported from analysis sources, and then RFM/AIPL models are generated and audiences are filtered.

The following table lists the mapping between the table types supported by the new version and the table types supported by the previous version. The previous version of the user tag dataset, behavior dataset, and RFM model base table can be used in the new version. However, the previous version of the AIPL model base table cannot be used in the new version. The new version of the AIPL model uses the user behavior table data.

New data table types	The type of the base table in the old dataset.
User tag table	User label dataset bottom table
User behavior table (table structure slightly different)	Behavior Dataset Bottom Table
Order details table	RFM Model Bottom Table-Transaction Data
Order summary table	RFM Model Bottom Table-Customer Data
Statistical Table	-
-	AIPL model bottom table (obtained from behavioral data aggregation)

You need to adjust the base table based on the Data Table Format Requirements of the new version, store the table in the bound MaxCompute computing source, and then operate in the new version. The following table describes the types of datasets and models.

Migrate user label datasets

Prepare a data table. The new version of the User Tag Table Data Requirements as the old version. You only need to transfer the table to the MaxCompute computing source.

Then, perform the migration operation in the new version. The configuration process takes about 2 minutes:

User Tag Table: Declare the table structure of the table to be imported.
Configure Import Scheduling Task: Set the import frequency and time.
Important
Each time you import a scheduling task, you need to re-map the ID of all users involved in all data tables. To avoid frequent consumption of computing resources, we recommend that you create the scheduling task for all data tables as the same task, including the following user behavior table and order details table. If you need to create RFM/AIPL models and audiences from imported data tables, wait until all data tables are imported.

Migrate behavioral datasets

Prepare a data table. The new version of the User Behavior Table Data Requirements slightly different from those of the old version. Adjust the table structure according to the new version and store it in the MaxCompute computing source.

Old version: A line records only one attribute. If a line records a brand, other attributes such as categories and products cannot be recorded at the same time. As a result, multiple lines of user behavior records can be saved to meet business needs, resulting in additional computing resources.
New version: allows you to record multiple attributes in a row. You only need to record one row of user behavior at a time. This is more efficient than the previous version.

Then, perform the migration operation in the new version. The configuration process takes about 2 minutes:

Configure the User Behavior Table structure.
Configure Import Scheduling Task.

Migrate the RFM model

Prepare a data table. The data requirements of the new Order Details Table Data Requirements and Order Dummary Table Data Requirements are the same as the transaction data and customer data requirements of the old RFM model. You only need to migrate them to the MaxCompute computing source.

Then, perform the migration operation in the new version. The configuration process takes about 5 minutes:

Configure the Order Details Table or Order Summary Table structure.
Configure Import Scheduling Task.
Create an RFM Model.

Migrate the AIPL model

Because the old AIPL model base table is not available for the new version, it cannot be migrated from the old version. In the new version, you must re-create an AIPL Model from the imported user behavior table. For more information about how to import user behavior tables, see Migrate Behavioral Fataset.

The AIPL model is created. The configuration process takes about 5 minutes.

3. Audience migration

Similarly, audiences need to be re-filtered or re-uploaded from the migrated data. As business data is refreshed, the lifecycle of audiences in the old version is gradually ended.

For more information about how to create an audience, see Audience Filtering Overview, Upload Audience, and Create Audience from Analysis Source.

The APIs and accounts of Data Bank, Dharma Disk, and Kafka involved in audience push are shared with those of the previous version. You can directly use the push feature in the new version to create a push task.

Note

In the new version, when you use some features of the social interaction, retail CRM, and media delivery modules to use audiences, you need to push the involved audiences to the corresponding modules first. For more information, see Push Internal Modules.

4. User marketing task migration

The new version and the old version share a set of marketing interfaces and account configurations. You do not need to manually configure the marketing interfaces and accounts again.

However, because the target audience of marketing cannot be migrated to the new version, you need to recreate the audience in the new version, and then create the same marketing task for the audience. Gradually switch the marketing task to the new version and gradually end the lifecycle of the marketing task in the old version.

For more information about how to create a marketing task, see User Marketing.

You can contact service personnel to migrate marketing tasks (including touch marketing, advertising marketing, and automated marketing) in the old version to the new version of the task list. Data such as task configuration, execution records of executed tasks, and sending result feedback will be migrated with the tasks. In this way, you can continue to view configurations, execute records, and analyze data in the new version. However, you cannot edit, copy, or execute the migrated tasks again.

Note

Special cases: For In-progress or paused recurring tasks triggered by specific groups of people, you can associate the created automated task with the old task to be migrated in the new version. After the association is successful, the old task is used as the old version of the new task, and the corresponding execution record is also included in the execution record list of the new task.

In addition, content management and event management data that do not involve audience data will be automatically migrated, and you do not need to manually configure them. You also don't need to modify the reporting event code in your party's application or in a third-party application.

5. Calendar Marketing Campaign Migration

Marketing Campaign may be associated with audiences and marketing tasks. You must manually reconfigure and reassociate the audiences and marketing tasks in the new version.

6. Analyze Kanban Migration

The new version of the Analysis Dashboard shares the Quick BI interface and report menu configuration with the old version, and references the dashboard generated from your business data from the Quick BI in the same way. You do not need to manually migrate the dashboard.

If you have configured the feature of selecting audiences from reports in the previous version, you need to select audiences from reports in the new version. The generated audiences are managed by the new version. Make sure that the Analysis Source configured for the workspace with the same name in the new version uses the same Analytic Database 3.0 database as that in the previous version. Audiences that have been selected from reports in the previous version need to be re-selected.