Charles (Dewar) and RFM Analysis

I read your post on the other board about RFM analysis. This is something I learned about in college and tucked WAY back in my mind until about 3 months ago. I started using it on my database of customers in my effort to increase frequency (which I’ve posted for help recently) and advertising ROI.

I don’t want to post a link to the other board because it might not be considered appropriate, but you think you can post your explanation again here? I think a lot of operators would benefit; it’s actually worked very well for me. Even if your POS doesn’t do it for you automatically, it should be pretty easy for anybody to accomplish by exporting their database into Excel (which is what I do.)

I’d like to mention that I added one attribute to the traditional RFM analysis. This attribute attempts to define a customer’s “coupon elasticity”. You may have customers that are a “555” (using your example from the other board) but are coupon junkies. The only way to keep them a “555” is to keep sending coupons.

On the other hand, you may have customers that NEVER use coupons. These people may remain “555” customers without needing a coupon. No point in sending anything to them.

Using your example scales, a person in my database that never uses a coupon would be “xxx5” and a person that always uses a coupon would be “xxx1”.

A “5555” and a “5551” have the same value to me, but I know the latter needs some prodding.


It is a method of quickly classifying customers with a three digit code, RFM. Each digit of the code represents which quintile they are in for Recency of visit, Frequency of visits, and Monetary value.

First digit of code: Organize all customers by how recent they have ordered. The 20% most recently-visited customers get a 5. Next 20% a 4 and so on.

Second of digit: Organize all customers by how frequently they have ordered. The 20% most frequent customers get a 5. Next 20% a 4 and so on.

Third digit: Organize all customers by how much they spend per order (we are actually using contribution margin). The 20% biggest spenders get a 5. Next 20% a 4 and so on.

So a customer with a code of 555 is gold. They recently ordered, they frequently order, and they spend a lot per order. Do you need to spend a lot on coupons or specials for them? NO. Do you need to show extreme appreciation? YES.

How about a 525? They spend a lot when they do come in, and they just recently came in. Perhaps time-sensitive offers to build frequency.

111’s? Dead weight. No promotion.

You can improve upon it by monitoring the migration of groups and individuals. So if a 555 one month becomes a 455, or a 355, then perhaps they have moved, had a bad experience, or been teased away by a competitor. You can also see what the average spending is of each quartile and how they change. Are your top spenders, on average, spending less? Is your average frequency dropping. How recent did the top 20 visit?

However, RFM provides just a snapshot and is not the end-all of marketing. But it is a nice tool. Certainly more meaningful than just noting total purchases, last 30 days, or last order date.

And to add, from what I have read, Recency is the best predictor of future business, whereas Frequency and Monetary determine the value of a customer.

I like your coupon idea. We are trying to avoid them at all costs but if we use them we should know who responds to them.

Thanks for posting. This really is a GREAT way to utilize database information in our business. It’s almost tailormade to the pizza business and I think it could help a lot of operators. I’ve seen the ROI on my database mailers increase significantly with this system.

I had the same plan before I opened, but it didn’t pan out. Found out from other business owners in the area that people here just love coupons, and they became a necessary evil. I make sure to only send out “smart” couopns though; no dollars off type stuff.

My store is right in the middle of two very different demographics; I’m on the street that separates them.

To the north is a city that is very “middle class” - $66,000 median income, median home price $243,000.

To the south is a city that is considered “wealthy” - $110,000 median income, median home price $483,000. Out of 10,000 homes, about 800 are valued at over a $1 million. Several are over $5 million.

Guess where I get the majority of coupons from? The “wealthy” area utilizes coupons at a 3-1 rate to the “middle class” area.

Jury’s still out on whether that’s how they got all of their money, or if they just can’t afford their houses :shock:

I also think RFM is a good method, what I have done in the past so that I do not miss the “525” customers is add the values together so that 15 is my best or my “555” customer. Using this approach you will have a 12 for the “525” customers.

I attempt to focus on the 10-15 value RFM customers.

Michael A. Davis