# RFM, Part 3

This is the final article of a three-part series discussing RFM models. If you don’t know what RFM stands for, I strongly advise reading Part 1 and Part 2 before Part 3. Not that anything presented here requires an advanced degree in statistics (quite the contrary), just that the progression is important to understanding the content of this article.

At the end of Part 2, we finished building an RFM model for an Internet media company selling impression-based advertising. The result left us with the equivalent of numeric soup — 27 customer segments identified by rankings for each of the RFM variables (remember the numbers are always represented in RFM order):

 111 222 333 121 212 323 131 232 313 112 211 332 113 213 331 122 231 321 123 233 322 132 221 312 133 223 311

The cut-off points — the top 20 percent, middle 60 percent, and bottom 20 percent — were analyzed, and each variable was defined as follows:

Recency
3 = Visitors whose most recent Web site visit was in the last 15 days
2 = Visitors whose most recent Web site visit was between 15 and 50 days ago
1 = Visitors whose most recent Web site visit was 50 or more days ago

Frequency
3 = Visitors who made a minimum of 45 Web site visits in the last 90 days
2 = Visitors who made between 10 and 45 Web site visits in the last 90 days
1 = Visitors who made a maximum of 10 Web site visits in the last 90 days

Monetary
3 = Visitors who generated an average ad revenue of at least \$2.00 per site visit
2 = Visitors who generated an average ad revenue between \$0.05 and \$2.00 per site visit
1 = Visitors who generated an average ad revenue of no more than \$0.05 per site visit

Now that the analysis phase is complete, we can develop marketing strategies targeting individual customer segments to maximize the value of each customer relationship. This is where the rubber meets the road.

Assuming that the Web site is in business to make money (if it’s not, then why is it in business?), start by focusing on the customer segments that can make the greatest impact in the shortest period of time, otherwise known as the low-hanging fruit.

Increasing Recency and Frequency

One method is to isolate customer segments that end in 3:

 113 213 313 123 223 323 133 233 333

Visitors within these segments each generate ad revenue of at least \$2.00 per site visit, meaning their site visits are more valuable than site visits from visitors scoring a 1 or 2 on the monetary variable. Generating more site visits from this group will have the greatest positive impact on ad revenue. The marketing campaign should attempt to move visitors from their existing customer segments toward the customer segment in the lower right-hand corner of the above nine-box matrix (333) — increasing the recency and frequency values. The marketing messages employed to reach each customer segment will differ.

Illustratively, the people in the lower left-hand corner (133) made their most recent Web site visit 50 or more days ago, made at least 45 Web site visits in the last 90 days (meaning they visited 45 times between day 50 and day 90), and generate average ad revenue of at least \$2.00 per site visit. This is a group of visitors that is moving in the wrong direction. As recently as two months ago, they were in the 333 segment, the group of most valuable visitors. Develop a marketing strategy to get these visitors to come back before you slip off their radar screen entirely. As someone with a background in email marketing, I’d advise testing multiple email messages to various groups within the segment to determine which message generates the highest probability of moving visitors into the 233 segment (hopefully visitors provided email addresses and permission to send email while they were still loyalists).

The people in the upper right-hand corner (313) made their most recent Web site visit in the last 15 days, made a maximum of 10 Web site visits in the last 90 days, and generate average ad revenue of at least \$2.00 per site visit. This is a group of visitors that is ripe with opportunity. They rank high on the recency and monetary variables but low on the frequency variable. The marketing strategy to target this customer segment should encourage more frequent visits to the Web site. Again, I’d advise testing multiple email messages to test groups within the segment to determine which message generates the highest probability of moving visitors into the 323 segment.

Increasing Monetary

Another method is to isolate customer segments that end in 1:

 111 211 311 121 221 321 131 231 331

Visitors within these segments each generate an average ad revenue of less than \$0.05 per site visit. Why are we targeting this group? We’re not going to target the whole group, just the four boxes in the lower right-hand corner (221, 231, 321, and 331).

The people in the lower right-hand corner group (331) made a Web site visit in the last 15 days, made a minimum of 45 Web site visits in the last 90 days, and generated an average ad revenue of less than \$0.05 per site visit. These visitors rank high on the reach and frequency variables but, unfortunately, are not being exposed to many advertising units. Either these visitors are viewing content that has a low percentage of advertising inventory sold or they are viewing a very low number of pages on each visit to the Web site. The key then is to leverage the affinity these visitors have for the site and migrate them to areas that have a higher percentage of advertising inventory sold. This can turn a visitor from a money-losing proposition into a cash cow very quickly. Note that the same holds true for visitors in the 221, 231, and 321 segments.

Wrap Up

As you can see, each customer segment tells a unique story. Once the story is understood, the marketing strategy for visitors within each segment becomes clear. There’s no magic here. It is all about utilizing customer analysis to identify customers who have the potential to generate the largest return on investment and developing targeted marketing strategies to influence those customers. If you are analyzing customer data for the first time, building a simple RFM model is a great first step to segmenting your customer base.