VARportal™

The two most common mistakes we see marketers make are 1) sending the same campaign to every customer, bombarding them continuously, and 2) putting heavy emphasis on their very best customers, going back to them again and again while ignoring many other potential purchasers. We call the first “spray and pray”. We call the second shortsighted, even though we recognize the temptation of sales people to pitch their regular customers time after time.

Both of these mistakes represent a failure of customer segmentation. In the “spray and pray” case there is no segmentation of any kind. All customers are treated identically, even though we all know there are many differences among them. An inability or unwillingness to differentiate them forces us to treat them all alike with spray and pray campaigns. The second case is a segmentation failure too, because the only segment that has been successfully identified is the top one, the set of best performing customers, which is the easiest segment to identify.
 
Since these two common mistakes are facets of the same problem, it is not surprising that there is a common solution: properly analyzing all customers and marketing to them according to their segment characteristics. Note my emphasis on all customers. Recently I got an invitation to a webinar from a company who promised they would help “identify your best customers.” I trashed the invite immediately. There are some tough tasks in customer segmentation, but finding the best customers is not one of them because, and I’ll say it again, that’s the easiest segment to identify. Heck, if you don’t already know who they are, just sort on recency (days since last purchase) or revenue received.

Doing customer segmentation and predictive analytics right is not that trivial. We’re going to show you how in later chapters. In the meantime, however, let’s look at some quick fixes to jump-start your campaigns. That’s what this chapter is about, Band-Aids for marketing. If you tell some of my colleagues about what I’m about to suggest, and they question me on it, I’m likely to deny I ever said or wrote it because these are not the best, long term ways to fix the common segmentation mistakes I’ve described. But to some extent they work, they are probably better than what you have been doing, and quick and dirty is not always bad.

The quick and dirty way

Start with a sort of your customer list by recency. How long can customers go without buying before you consider them inactive? Let’s assume it is six months for your business.  For most firms this number typically ranges from three months to two years. Start by selecting customers whose recency is between six and nine months.  They are over the assumed six-month edge, inactive, but not terribly so. They are reasonable candidates for a win-back campaign. There are better ways to pick these win-back candidates, but that is more work for later on. Now, come up with a really attractive offer for these customers before they drift even further away, a better offer than you could afford to make to your entire population. It is less expensive to bring these customers back to active status than it is to acquire new customers to take their place.

Next, from the same recency segmentation, carve out the customers whose recency is between three months and six months. If your inactivity threshold is at six months (our assumption), then this group of customers is a fair approximation of your middle or middle-lower tier customers. Now try to get another purchase out of them by making a somewhat customized offer using a product affinity table.

A simple product affinity table

To build a product affinity table, make a list of your top ten or twenty five or so best selling products. Next to each product name, write the name of the next likely product purchase by someone who bought the product in the first column. Entries in the second column do not need to be unique; there could be several of your top sellers that have the same follow-on product purchase. Your entries in the second column are not going to be completely accurate because you are using your experience and intuition to make the choices, rather than using mathematics to determine what the next logical purchase actually is. But if you know your own business, I’ll bet you will be right on for several top sellers and not too far off for the others. Now go back into your customer records and find the last purchase by buyers with three-to-six month recency, and make them a new offer based on your product affinity table. Make a good offer; getting an extra purchase from this group could be a big revenue boost.

So what’s wrong with this quick and dirty approach? It is certainly an economical way to boost revenue by making customized offers to segments of the customer population that are likely to respond better than they would to your usual spray and pray campaigns. There are two problems. First, it is far from the most accurate way to segment and campaign. Second, there are actually faster and easier ways to accomplish the same tasks.

It’s not the best way to segment because it ignores many aspects of customer behavior, including what a customer is buying and whether their rate of buying is constant, slowing, or accelerating, to name just a few ignored factors. Again and again, we are asked to show that our mathematical marketing methodology yields better results than the prospect’s approach. We know that quick-and-dirty Band-Aids work to some extent because we see companies using them. But we also know our more sophisticated methods are better because we’ve been asked to do so many comparisons that we now build free A/B testing (usually our method vs. the one you’ve been using) right into our services. Clients can now see by how much their usual methods fall short.

Secondly, ‘quick and dirty’ is often not the quickest or the easiest. (It is ‘dirty’.) Many companies don’t keep records of what customers have previously purchased. For those that do, pulling that last purchase from the data can be an ornery task. It can be and should be automated. With the right tools it is easy to segment a customer population in under an hour or certainly overnight. We’ll get to those tools in a later chapter of this guide. In the meantime I’ll dodge my colleagues’ accusations about recommending less than optimal solutions. You try out some mildly customized campaigns. Together we’ll move forward and explore how mathematical marketing can transform your business.