DocsICP FilingConsole
官方文档
首页

Create a new RFM model

更新时间：
复制 MD 格式
产品详情
我的收藏

You can create RFM models for RFM analysis, filtering audiences with an RFM model, and other operations.

An RFM model measures customer value using three metrics: Recency (R) for purchase interval, Frequency (F) for purchase frequency, and Monetary (M) for purchase amount.

The RFM model quantifies and scores the values of the three metrics for each customer. Then, the model compares an individual customer's score to a comparison value, such as the average score of the entire customer group or a specified score. This comparison determines the customer's relative value within the group. By combining the three metrics, you can segment the customer base into various types. This makes it easier to apply targeted operational strategies for different customer types. For more information, see RFM user type segmentation rules.

You can choose from two analysis types when you create an RFM model:

  • Customer data: For large data volumes, you can first aggregate the raw data into tag data at the customer level. The aggregated data volume must be within 100 million rows. In this table, each customer has only one record. This type is suitable for analyzing large-scale RFM models.

  • Transaction data: Use this type to analyze small-scale transaction data, up to 100 million rows. Each row represents a user transaction record and includes a customer identity, transaction date (date type), and transaction amount (numeric type). This type is suitable for analyzing small-scale RFM models.

Prerequisites

Procedure

  1. Go to Workspace > Configuration Management > Data Center > Datasets to open the dataset management page.

  2. In the upper-right corner, click New > User Model > RFM Model. This opens the RFM model configuration page.

  3. Click Customer Data or Transaction Data.

  4. Click Select Data Table. In the dialog box, select a data source and a data table, and then click Confirm.

  5. Click Next and configure the mappings.

    • If you selected Customer Data as the analysis type, set the mapping fields for User Identity, User Identity Type, Last Purchase Time, Cumulative Purchase Count, and Cumulative Purchase Amount.

    • If you selected Transaction Data as the analysis type, set the mapping fields for User Identity, User Identity Type, Transaction Date, and Transaction Amount. The number of transactions is calculated by counting the transaction records.

      Note

      • The user identity is the unique identifier for a user in this dataset.

      • The user identity type corresponds to the ID type of the user identity. When you select an audience, the ID type that you set here is used as the default.

      • Supported ID types include OneID, UnionID, mobile phone number, email address, Taobao ID, Taobao nickname, Taobao OUID, Alipay ID, Weibo ID, mobile phone International Mobile Equipment Identity (IMEI), mobile phone identifier for advertisers (IDFA), mobile phone International Mobile Subscriber Identity (IMSI), mobile phone Open Anonymous Device Identifier (OAID), MAC address, and OpenID.

    • Add other ID types.

      If the data table for the RFM model contains other user ID fields, you can add these IDs to the dataset. When an audience generated from this model is pushed to Data Bank, you can select from multiple ID types.

      Note

      You must set an ID type for any new IDs. This ensures that you can select from multiple ID types when you push audience packages that are generated from this RFM model to Data Bank.

  6. Click Next and set the RFM parameters.

    Set the number of segments for R, F, and M. Then, set the range and score for each segment. The user distribution for each segment is displayed below.

    The scoring rules are as follows:

    • Recency (R): The fewer days since the last purchase, the higher the score.

    • Frequency (F): The more purchases within the last N days, the higher the score.

    • Monetary (M): The higher the purchase amount within the last N days, the higher the score. The default currency is CNY. You can select a different currency on the right side of the page.

    If you selected Transaction Data, you must also set the statistical period on the right side of the page. You can set it to the last N days or All Time. All Time includes data from the beginning to the current day.

  7. After you configure the scoring rules for R, F, and M, click Next. Set the parameter comparison values. You can use the average score of the entire customer group or set custom values.

    Note

    Comparison values are used to compare a user's score against a benchmark. This helps determine the user's relative value and segment them into different types. For more information, see RFM user type segmentation rules. RS, FS, and MS are the scores for the R, F, and M metrics. You must set comparison values for RS, FS, and MS.

    • If you use the average score of the entire user group (the weighted average) as the comparison value, the page displays the average RS, FS, and MS scores for the current customer group based on your scoring rules, as shown in the following figure.

    • If you set custom values, you can adjust them as needed by referencing the average scores shown on the page.

  8. Click Finish. In the dialog box, enter a name and select a save location for the RFM model, and then click Confirm.

    You are redirected to the dataset management page. The new RFM model appears in the dataset list. For more information about how to manage the model, see Manage RFM models.

References