Overview of the RFM model

RFM, short for Recency ®, Frequency (F), and Monetary (M), is a data-driven approach to customer segmentation and analysis. This methodology evaluates three key aspects of customer behavior: how recently a customer made a purchase, how often they engage, and how much they spend. By quantifying these factors, businesses can gain actionable insights into customer segments and develop targeted marketing strategies that better meet individual customer needs.

Understand customer behavior with the RFM model

The RFM model segments customers based on transactional behavior using three key parameters.

  • Recency measures the time since a customer’s last purchase, indicating engagement levels and future buying potential.
  • Frequency tracks how often a customer interacts, serving as a clear indicator of loyalty and sustained engagement.
  • Monetary value assesses the total spending by customers, highlighting their overall value to the business.

By combining these factors, businesses assign numerical scores (typically on a scale from 1 to 4) to each customer. Lower scores indicate more favorable outcomes. For example, a customer scoring 1 in all categories is considered among the best, demonstrating recent activity, high engagement, and significant spending.

Benefits and limitations of the RFM model

Every marketing modeling technique involves trade-offs, offering both benefits and limitations. RFM modeling is a valuable tool for understanding customer behavior and refining marketing strategies. Its advantages include segmenting customers to personalize messaging, optimize revenue, and improve response rates, retention, satisfaction, and Customer Lifetime Value (CLTV).

However, RFM modeling has limitations. It assumes uniformity within segments based on recency, frequency, and monetary value, which may oversimplify customer behavior. The model also assigns equal weight to these factors, potentially misrepresenting customer value. Additionally, it does not account for context, such as product-specific traits or customer preferences, which can lead to misinterpretations of purchasing behavior.

Build a dynamic RFM score-based SQL audience

The following infographic provides a high-level overview of the RFM SQL audience creation workflow described in this tutorial.

An infographic titled "RFM-Score-Based SQL Audience" illustrating four steps: upload CSV, explore data, enrich with RFM scores, and activate the audience.

Before starting the Luma case study, you need to ingest a sample dataset. First, select the link to download the luma_web_data.zip dataset locally. The sample dataset is a csv file in a compressed .zip format to align with the use case. Decompress this ZIP file using Adobe Acrobat or a trusted file extraction tool, such as your operating system’s built-in utility. In practice, you would typically source data from Adobe Analytics, Adobe Commerce, or Adobe Web/Mobile SDK.

Throughout this tutorial, you will use Data Distiller to extract relevant events and fields into a standardized CSV format. The goal is to include only essential fields while maintaining a flat data structure for efficiency and ease of use.

Step 1: Upload the CSV data into Experience Platform

Follow these steps to upload a CSV file to Adobe Experience Platform.

Create a dataset from a CSV file

In the Experience Platform UI, select Datasets in the left navigation rail, followed by Create dataset. Then select Create dataset from CSV file from the available options.

The Configure Dataset panel appears. In the Name field, input the dataset name as “luma_web_data” and select Next.

The Add data panel appears. Drag and drop the CSV file into the Add data box, or select Choose File to browse and upload the file.

To learn more about this process, refer to the batch ingestion tutorial and the dataset creation workflow in the Dataset UI guide.