Model vs RFM
Each of us wants to mail to the right customer at the right time with right product and the right offer. And particularly with circulation (that’s the term used to figure out which customers or prospects receive the campaign) there’s a wonderful combination of historical performance (industry or specific within the company) as well as a constant influx of data to tell us what is working and what is not. That data is response rate, average order value, items per order, channel the order was placed, etc. It’s truly remarkable how insightful the data can be. I think that’s why I love circulation…I love the data. I love how the data reveals the predictability of buying behavior.
In the most recent few months I seem to cross paths with modeling opportunities of the house file (aka current buyer file, your customer database.) I scream from the mountain tops that I have have yet to find a model that beats RFM (RFM is the acronym for Recency Frequency and Monetary–which is a traditional way to segment your customers into the purchasing behaviors of RFM.) I know, I know, I just threw down the gauntlet…which truly is not why I write.
Two recent examples. A cataloger has a small mailing planned–just to best customers and simply a way to stay top of mind during the off-peak season. The mailing is so small that it doesn’t even mail to the entire 12-month file (customers who have purchased in the most recent 12 months.) A vendor convinced them to do a house file model. I was floored. I about fell over! Why would you waste money on a model when your mailing doesn’t even mail to all of the best customers? Fast forward. The model picked a small percentage of names that were not originally chosen for the mailing, and the performance revealed a very, very low response rate but a very high average order value. So the question is, do you want more customers, or do you want a few customers that spend a lot? The other issue which for me was more important, how do roll out this model when it was mailed to fewer than your 12-month file? Rhetorical.
Example Two. Again a house file model on the 12-month file. (Seriously, was there a memo I missed?) The model thankfully was a flag file so we could maintain the RFM segmentation and just have counts of the records that the model identified. Because I’m a huge proponent of RFM, I take RFM one step further and also identify customers with a promotable email, those without, and those unknown. Back to the example, the first step was to select the mailing status quo. Turns out, 95% of what is usually chosen was already flagged by the model. The model found names older than the 12-month file that met the criteria. So we are mailing those straggler records found in older segments (how does a segment of 4 sound?) Definitely not worth the expense or extra time in the process for the model. RFM continues to be validated by models (notice how I use the model to emphasize my point?!)
With all that said, I do believe in the value of models, just not better than RFM. You may be surprised to read that I successfully use models for scoring catalog requests, and other non-buyer groups (such as sweepstakes or trade show leads.) Modeling has been very helpful to identify potential buyers with the ever-increasing volume of one-time buyers (I do have more examples but I wanted you to know, that I know, models are useful tools.)
Data drives decisions…and that’s what I love. I’ve been doing circulation long before e-commerce was a word in the dictionary. RFM remains steadfast in my arsenal. Modeling, proven very helpful, but not a replacement for RFM.
There will be more about RFM…Like RFMP or RFMC, or RFME…lots to talk about!!!
Did you enjoy this post? Sign up for more.
Subscribe to Blog via Email
The Catalog Blog covers opinions and information on all things catalog. Have something to add? Leave a comment below. Catalog University is devoted to helping you get ahead in the fun and fascinating world of catalogs. If you want even more information about cataloging, including FREE Pub Talks, be sure to sign up for the Cat-U mailing list. We will never share your name with 3rd parties.
Dean of Marketing, CatalogUniversity.com
President, Hemisphere Marketing, LLC