Every October, my company organizes a well-established “Discount Week” for its in-house brands, offering select products at discounts of up to 50 percent. To streamline logistics and sales execution, this annual promotional event is limited to five designated retail locations. Over time, customers have come to anticipate this campaign, making it a key part of the retail calendar. However, concerns were raised at group headquarters following reports that the Tsing Yi branch may have been using a more aggressive pricing strategy than authorized, including additional discounts aimed at attracting customers from outside its typical catchment area. As a result, Internal Audit was asked to conduct a review, including on-site observations.
As expected, detecting any form of non-compliance during a short visit of only a few minutes proved challenging. However, a closer look at backend data revealed some telling patterns.
- The Tsing Yi store consistently exceeded its sales targets during the October promotion period, yet it regularly underperformed throughout the rest of the year.
- Customer purchasing behavior appeared to be shaped by this promotional model. Over 70 percent of the store's customers had not made any purchases during the six months prior to October, and their spending during the promotion remained below the average level of customers who typically shop once every two months.
- Purchase records showed that customer activity was mostly concentrated in the New Territories West region, with no meaningful indication that the store had successfully attracted customers from other areas.
How Can We Use Data to Segment Customers?
When we want to quickly understand our customer base, one effective approach is to define each customer using measurable indicators, then group customers with similar patterns together. This allows us to understand how different business strategies affect each group. These models are not designed to directly answer strategic questions such as how to increase customer purchase frequency or how to improve retention. However, they do provide a fast and structured way to assess customer behavior and business performance.
In practice, such models are best used as tools to test and validate marketing strategies. For example, based on our understanding of business operations, we can create customized models for different sales channels, product categories, or campaign types. This helps us verify which strategies are more effective, such as which sales approach brings higher retention or which product lines have more loyal customers.
Customer Segmentation Model: RFM
What Is RFM?
The RFM model uses three key metrics to measure customer value:
-
Recency (R): How recently a customer made a purchase
-
Frequency (F): How often the customer makes purchases
-
Monetary (M): How much the customer has spent in total
Each customer receives a score for these three factors. The scores can be summed into a total score or assigned different weights depending on the priorities of the business. The model was introduced by George Cullinan in 1961 and is still widely applicable today across industries including retail, e-commerce, and software.
RFM Analysis Steps
Step 1 – Prepare your dataset:
You will need four fields: order date, order ID, customer ID, and order amount.
Step 2 – Structure the data by customer:
Aggregate the data to show each customer's first purchase date, most recent purchase date, number of orders, and total spend.
Step 3 – Assign scores to each customer for Recency, Frequency, and Monetary value:
Use percentile or quintile distribution to assign scores between 1 and 5. This step helps reveal the overall customer profile and consumption behavior.
Step 4 – Label each customer segment based on their scores:
Group customers into categories such as high-value loyalists, recent buyers, or inactive customers. This labeling helps teams understand the size and nature of each segment.
Step 5 – Use the segments to evaluate business strategies:
Compare how different customer groups respond to marketing campaigns, product launches, or pricing changes. This supports data-driven decisions on where to invest efforts for retention, reactivation, or cross-selling.
Comments
Post a Comment