|
|||||||||||||||||||||||||||||
|
Behavior Beats Demographics Every Time! Using RFM in your business -- by Chip House My friend, Otis really likes to play guitar. Actually, he really, really likes it and is very good to boot. Why am I sharing? I'm telling you to make a point. The point is you'd never know just by looking at his demographics that Otis is interested in guitar (other than his guitar-guy name, anyway)Ä In fact you might assume that Otis is really into things like golf, home repair, ice fishing, or riding snowmobiles based on his demographics (male, over 30, Minneapolis zip code, etc.). Demographics, as defined by The American Heritage Dictionary, are: "The characteristics of human population segments." Problem is, demographics can tell us only where we are and what interests we might have. In the direct marketing world, demographics are great for directional guesses on how a subscriber might respond. However, a marketer using only demographics would be wasting their money trying to get Otis to subscribe to Golf magazine, or buy fishing lures. Truth is, Otis hates pretty much anything but guitar. Only way you'd know it is by watching his behavior. Marketers focus on behavioral data when they can get it. Why? Because behavior is a far better predictor of future behavior than demographics are. Whereas demographic data is very impersonal and general, behavioral data tells us a lot more about an individual. Behavioral analysis focuses on what people do rather than who they are or where they are. That means keeping track of customer purchase data: when they buy, how often they buy, how much they buy, what they buy - and, on a web site or email campaign, what they click on. Bottom line is, behavioral data will beat demographic data every time. Last week we talked about how using Recency, Frequency, Monetary (RFM) Analysis to help reduce customer attrition. We saw how Recency (how recent a customer's last purchase was) is the most powerful statistic for determining buying patterns. This week we'll focus on Frequency and Monetary data. Though less powerful predictors than Recency, they are also key behavioral statistics you can easily apply to your own customer database. To illustrate the point, I'll use two of my friends, Otis and Craig. Otis is a real musician. He buys frequently from The Guitar Center, in fact he probably just made a purchase this week. He is a steady customer with Guitar Center, a big Guitar Center advocate, and will purchase from them again as long as he is treated well. He spent about $4,500 with them last year. Craig, on the other hand, is more of a wannabe bass player. He bought his bass, his amp, his cords, a microphone, and several sets of strings all in one visit to Guitar Center. He spent about $5,000. He hasn't been there since. Let's use Craig and Otis to illustrate the power and potential pitfalls of Frequency and Monetary analysis. Frequency, for starters, is a more powerful predictor of future response than Monetary. Why? Look at our example; Otis regularly makes purchases. His purchases are predictable, and therefore likely to occur in the future. Craig, however, is not a frequent purchaser, and since he really doesn't play his bass, he isn't likely to visit the Guitar Center anytime soon. This example also illustrates the "trap" that using only Monetary analysis could create. Craig actually spent more than Otis on equipment last year. In some marketers' models, they'd rank Craig as a stronger customer than Otis since he spent more than him. But you and I both know he's not! In fact, Otis' behavior is ideal. That's why Recency is the only statistic you can safely use on its own. Both Frequency and Monetary need to be used together, or in conjunction with Recency. It is easy to apply Frequency and Monetary data in your own business just by sorting your customer database by number of purchases each customer has made over a specified time period. You decide the time period, then divide the database into 5 equally sized segments or "quintiles." So, if you have 5,000 total customers, the 1,000 that bought most frequently will be in the top quintile, the second most frequent 1,000 will be in the second quintile, and so on. Then, to keep track of who is in each segment, code each customer in the top segment with a "5", those in the second most recent segment with a "4," etc. For monetary value, sort your database by amount spent per year, then divide into 5 equal segments. If you combine this with Recency data, your top customers (those most likely to purchase again and tell the world) will be coded "555" and your worst customers (those you'll never see again) will be coded "111." By using this coding when selecting your customer names for mailing, you'll spend less money saying the wrong things to the wrong customers in your database and begin to focus your communications on the customers and messages that can earn you revenue. That's what its all about. If you are not already tracking behavioral data on your customers, now is a great time to start. |
|
|||||||||||||||||||||||||||
|
|
|
