Marketing analytics is changing the way businesses identify and communicate with customers


It’s been promised for a long time, and it’s finally true; the computers are taking over the world.  The capacity of computers to process and evaluate data and learn from that data is now enabling marketing organizations small and large to make better decisions about the allocation of resources and communicate with the target audience more efficiently.

Fortune 500 firms have had access to this kind of capacity for a few years, but only recently has the technology become affordable for Main Street businesses.  I sat down with Jeff Cassman, the Creative Director for Invocabo, a Nashville, TN based creative agency, to learn more about how machine learning and predictive behavioral modeling is helping SMBs grow.

Image result for Marketing analytics

“Everyone has heard of Moore’s Law and the theory about the doubling of processor speed every 18 months, and the effect this has on pricing.  But Moore’s Law was really based on a specific application of Wright’s Law.  Dr. Theodore Wright had theorized in 1936 that the cost of a unit decreases as the scale of production increases.  Or put another way, “We learn by doing”.”

Aside from the historical trivia, it’s not a particularly profound observation; most undergraduate business students know that costs per unit fall as the volume of production rises; we know it simply as ‘scaling’.  But Cassman says there is more to it.

“We’re not talking about a manufacturing line and widgets…we’re applying this principle to big data, and the ability of a computer to learn.  We have finally reached the point that the computers can process enormous amounts of information quickly and inexpensively.  What was impossible 10 years ago is now a reality, and what was cost-prohibitive for all but the largest organizations just 5 years ago is now affordable for many SMBs.”

Cassman is referring to predictive behavioral models, which involve the assimilation of large amounts of data, say, a list of potential customers and the comparison of that list to a second list, for example, a firm’s existing client list.  

“Rarely does a SMB have enough data on their clients in order to make these observations themselves.  Even if they did have the data, they don’t have the internal resources to observe, analyze and compare that data.  So what we can do is append their customer list with hundreds of different data points for each contact, and then compare that customer list with an audience that was exposed to their marketing message and did not convert, and let Mycroft-that’s what we call our supercomputer-chew on that data for a few days.  Mycroft can compare hundreds of different data points for thousands of different records simultaneously, and actually learn.  The result of that learning is correlation;  we can predict who future buyers are based on those data points”.

Predicting the future is something most everyone has dreamed about at one time or another, but Cassman says that it’s a reality today-at least in marketing analytics.

“We have access to so much information on consumer behavior now that as long as the client has a list size large enough to be statistically meaningful, we can actually predict who is most likely to engage and convert.   This leads us to another famous law…Pareto’s Law.  We’ve all referred to the so-called ‘80/20 Rule’ at one time or another.  With predictive behavioral modeling we can identify that 20% of your audience who will represent 80% of your revenue, so you can spend your resources communicating with that group alone.  The outcomes are profound.”

The results speak for themselves; case studies prepared by Invocado demonstrate dramatic improvements in return on ad spend, costs per click and costs of customer acquisition.  More information on machine learning, marketing analytics and predictive behavioral models can be found at

Leave a Reply

Your email address will not be published. Required fields are marked *