Where Are My Predictive Models

What do you think of when you hear the word, "Analytics?"

Again and again your humble host of The BeanCast hears about measurement and results and ROI and other ways to evaluate the effectiveness of a marketing campaign. And then occasionally I'll hear someone talk about using analytics to target a specific demo or zip code. But I rarely hear anyone on my marketing podcast or in my consultantcy or pretty much anywhere talk about how their models are being used to actually predict behavior.

What happened?

A Once Bright Future

I remember when I first entered into the marketing world, I was exposed to an arcane piece of math that seemingly put Einstein's theory of relativity to shame. To my please-no-math brain, I couldn't understand a bit of the equation. But when the benefits it represented were explained to me, my eyes lit up and I became a life-long evangelist right on the spot.

The model shown to me was called the Cultivation Opportunity Index Valuation model, created by Ward Thomas, an executive at the then Response Marketing Group, which after many reorgs and purchases has become EuroRSCG Discovery. This COI Valuation model (as it was called for short) could take a three-dimensional view of a customer based on three axes of data -- demographics (age, income, etc.), psychographics (interests, hobbies, predilections, etc.) and past purchase behavior. And what it would essentially do is tell us with a high-degree of accuracy what the lifetime value of each customer was.

Measuring Value Instead of Just Response

Notice that this wasn't just about response. This wasn't about just meeting immediate sales objectives. This model (and supporting models like it) essentially told us which individuals in the database represented the highest profits over time. So when we went out to market, we not only could reach or exceed our response objectives, we could also fine-tune our targeting to such a degree that we could achieve these goals with sometime 50% or less of the mail volume previously needed.

Unless you are VERY new to marketing, you'll understand that hitting your response numbers at 50% mail volume, can mean at least a 2x increase in revenue and usually much more, considering the reduction in hard costs.

A Neglected Gem

Yet even back then, when I heard a lot more about this type of modeling, it was hard to convince marketing managers that the investment was worth it. There was nothing tangible (or understandable, for that matter) they could take to their superiors for approval. And even if the investment was made, when the results came in it was the creative that took all the glory.

So true predictive modeling remains an arcane art, practiced well by a few and ignored by the majority of marketers. And it completely boggles my mind that a discipline that can show such solid return on investment is still over-shadowed repeatedly by the marketing darling du jour. It's time that we righted this ship.

Send to Kindle

Add to Flipboard Magazine.