New Tools of Mathematical Marketing

by Mark Klein

Originally published in 1 to 1 Media, on 17 Mar 08

Search Engine Marketing (SEM) is so pervasive and powerful, and so obviously the tool of choice for new customer acquisition, that marketers sometimes put less emphasis on the parallel task of existing customer marketing. They overlook the equally powerful collection of tools for that task which are coming to be known as Mathematical Marketing (MM).

Mathematical Marketing is the process of marketing to existing customers based on a scientific understanding of how past customer behavior predicts future purchases. Key elements of the process include behavioral tracking, predictive analytics, behavioral targeting, testing, what-if analysis, and results measurement. For comparison, the typical elements of SEM are keywords, ad impressions, click-through, search engine optimization, and web analytics.

Software as a Service

Now some larger enterprises with large computing environments, skilled statisticians, and database managers have been using a few of the MM techniques for several years, but new developments are enabling a wider set of marketers to use these techniques too. One such development is the trend to delivering software as a service (SaaS). Since all work is done through a web-based browser, getting started is simplified by not needing to install software. Users need not worry about hosting or managing a database. The considerable computing infrastructure is managed in the Internet cloud, dramatically reducing the cost to users. Further, SaaS applications are usually pay-as-you-go services. Not only does this remove the need for up front investment, but users can opt out if the service is not delivering on its promises.

Visual Math

A second enabling technology is visual math. Before the advent of new MM user interfaces, a company needed a strong statistician to do the analyses and a database person to write SQL queries to extract the answers. Today, the math is done for users inside the SaaS application. Selecting customers for campaigns (targeting) is done by moving sliders in a graphical user interface. SQL queries are generated behind the scenes according to slider positions. Marketers need to know what kind of campaign they want to run, not math or SQL grammar. Removing the requirement for statistical or database skills greatly increases the set of marketers who can deploy MM.

New Algorithms

The third important development is the steady improvement in the underlying algorithms used for predictive algorithms used for predictive analytics, both in accuracy and speed. What used to take weeks is now done in hours. Overnight analysis is the new norm, enabling users to react faster and be more responsive to their customers. Predictive accuracy has moved well beyond what was possible with cruder RFM techniques. As a consequence of these ‘under the hood’ improvements, MM practitioners are showing greater confidence in their tools.

Testing

Testing, one of the two best ways to improve campaign results, is another area where new developments are pushing the adoption of MM techniques. Most marketers don’t test because they are stuck in the old world of split run, A/B testing, where only one variable can be changed at a time. That methodology is been too expensive and too time consuming since it requires multiple tests and many subjects to test multiple variables. The newer approach, called Factorial Design, enables a company to test ten, twenty, or even forty or so variable simultaneously. If the testing is done with email campaigns that usually complete within two days, the whole testing cycle is incredibly compressed, saving both time and money, and the resulting main campaign is much more effective. Consequently Factorial Design testing will be a big driver of MM in the near future.

What-if Analysis

A fifth development enabling MM is the availability, finally, of real time what-if analysis when building marketing campaigns. Spreadsheet jockeys have used what-if analysis for a few decades, but marketers have not been able to apply the same concept to predict the outcome of a marketing campaign because the underlying data is not sitting in some limited number of spreadsheet cells but rather in gigabyte-sized tables of a database. Knowing how a campaign will turn out before the collateral is printed or the mail sent reduces risk enormously, sets expectations realistically, and ultimately improves overall performance.

Some of the same factors contributing to the popularity of SEM apply to the newer discipline of MM. MM works, powerfully, and delivers improved sales results. Users don’t need an advanced degree in statistics to use MM. But to realize this promise, several enabling technologies needed to be put in place. Affordability comes with the SaaS delivery model. Visual math brings ease of use. Speed and responsiveness come from better algorithms. Testing improves campaign results. What-if analysis reduces risk. With these technologies now in place, MM is accessible to a much wider audience of marketers, and should rival SEM as a fundamental marketing tool.

About the author

Dr. Mark Klein is the CEO and founder of Loyalty Builders LLC, a leader in the new science of mathematical marketing, plus three earlier companies. Prior to his career in business, he earned a Ph.D. in Theoretical Physics from Indiana University, and taught and did research at several other colleges. He blogs frequently on mathematical marketing, and his first novel has just been published.