Category Archives: Driller Q & A

Modeling Customer Behavior with Small Databases

Jim answers questions from fellow Drillers
(More questions with answers here, Work Overview here, Index of concepts here)

Topic Overview

Hi again folks, Jim Novo here.

We’re about to take a trip into the world of small scale databases. In particular, how does a not-for-profit with a small database of donors go about using predictive models? Answer: Keep it simple. Try to avoid using a lot of variables; look for the most powerful and stick with those until you are able to uncover additonal info and grow the database. Ready?


Q: I am new in the NFP (Jim’s note – Not For Profit) sector and would like some advice re:  segmentation models to optimize campaign results – both response and value (Short Term  and Long Term).  Do you know or is there any knowledgebase of how the various techniques – behavioural, RFM, demographic, geographic – generally rate against each other?

A: Not other than my web site / book, which generally covers all the simple models. There is plenty of info around on the web though.

Assuming the end Objective is a donation, the behavioral stuff is going to be much more productive than the geo / demographics are. It’s like a pyramid.  My friend Avinash “stole” (with my permission) a slide from my presentation on this topic and put it on the web, you can see it here.  You’re looking for an “action” (donation), so actions (behavior) will be the most useful segmentation, at least as a primary cut.  Then you can get into geo /  demo stuff if it improves the model.

Q: As my database is small I don’t have the luxury of testing multiple techniques and causal factors.  I will probably run tests in series but would like a general idea of which ones to test first to cut down the time.

A:  Not sure what you mean by “small”, but in general, the more complex a behavioral segmentation approach is the larger the database it needs to be useful.  So for example, with classic RFM (125 segment scores), the bare minimum for it to make any sense is probably 5000 records, and you should really have at least 10,000.

Continue reading Modeling Customer Behavior with Small Databases

How Much is Promotional Proneness Costing You?

Jim answers questions from fellow Drillers
(More questions with answers here, Work Overview here, Index of concepts here)

Topic Overview

Hi again folks, Jim Novo here.

Is your mission to increase Sales or Net Margin dollars? Worth getting some clarity on if you’re not sure, and if it’s Margin dollars you are after, watch out for Promotional Proneness. What’s that? The tendency of customers to learn promotional patterns and “wait for a discount”, which can significantly impact campaign profitability. Got Proneness? Read on Fellow Driller, you will learn how to find out – by measuring it using control groups!


Q:  I really have enjoyed your book.

A:  Thanks for buying it, and for taking the time to tell me you enjoyed it!

Q:  I’ve created a first draft of a customer retention strategy that outlines proposed offers at various trip wire stages, and based on your order frequency.  So, if you are a one time buyer, and you are 8 weeks over your average buying frequency, you get a certain offer, and this would differ if you were a 4-time buyer, and are just one-week over your average buying Frequency.  As you suggested, the offers increase in value the longer it’s been since you’ve purchased.

A:  So you are segmenting by Frequency and Latency and then using Recency as a trigger.  You must have really learned something from the book, I don’t think I ever covered that one specifically!  But it makes a lot of sense to use Latency instead of Recency to segment in a category with a high percentage of consumable products (FYI Dear Reader – office supplies), since there is some expectation for re-supply and the purchase rate should be relatively constant for paper, toner, pens, etc.

Q:  But how do you prevent teaching behavior that causes the customer to wait until the better offers come?  These offers would only be sent to people that hit the trip wire (not individuals buying on their own).  How do we not teach a behavior that encourages the customer to wait for the better offer?

Continue reading How Much is Promotional Proneness Costing You?

Difference between RF(M) Scores & LifeCycle Grids?

Jim answers questions from fellow Drillers
(More questions with answers here, Work Overview here, Index of concepts here)

Topic Overview

Hi again folks, Jim Novo here.

Both RF(M) scoring and Lifecycle Grids use the same key predictive metrics – Recency and Frequency. So what’s the difference? RFM is a predictive “snapshot” at a specific point in time; LifeCycle Grids are more like a “movie” designed to be predictive over different periods of time. Another way to think of this: RFM is tactical, LifeCycle Grids are strategic.

You dig? Let’s Drill …


Q:  We’re a telecom company trying to get a handle on customer churn and defection, so we can come up with some programs that will hopefully extend customer participation.  We live in the no contract space, offering a service that’s an add on to wireless phone service, so we don’t have a good indicator as to when the customer relationship might end.

A:  Ah, yes.  Your business model is “built for churn”, as I said on my blog the other day.  The behavior then is more like retail, where independent decisions are made in an ongoing way, deciding again and again to purchase.

Q:  I think your LifeCycle Grids method will show best what is happening to our customers.  If using this method, there doesn’t seem to be any reason to do the RF scoring as customers are just going into cells based on where they fall in the Recency and Frequency spectrum.  Is that correct?  Is there any real  difference between RF scoring and the LifeCycle Grids approach?

A:  You are partially correct, they are two versions of the same idea – both are scoring using Recency and Frequency. The traditional RF(M) scoring where customers are ranked against each other is a “relative” scoring method used primarily for campaigns – it is tactical, an allocation of resources model. 

Continue reading Difference between RF(M) Scores & LifeCycle Grids?