THE RIGHT TOOL FOR THE JOB: MACHINE LEARNING FOR ADVANCED MARKETING ANALYTICS

Have you ever tried using your shoe to drive a nail into wood? Or a broom to swat a fly? (...we’ve all done it.) It would have been much more efficient, and less exhausting, to use a hammer or bug spray. Sometimes we choose the wrong tools for the job at hand. So, here’s another scenario: if your business needs to make critical decisions to keep up with rapidly changing factors, would you rather get answers in 3 months or 3 hours?  

If you get where I am going here, then you understand the benefits of leveraging machine learning (ML), which is a subset of artificial intelligence (AI), to get certain jobs done right, and quick.

EARLY RESULTS

Machine learning has been a hot topic for several years now, and most businesses have some notion of what it is, but when applied to real-life decision making, most of them have been slow to widely adopt this powerful technology. This is changing, and Sentier has been helping commercial biotech and pharmaceutical companies transform the way they are making marketing and sales decisions with this technology. 

When combined with the ability to economically store large, disparate volumes of data, advances in computational analytics allow businesses to model their data and run thousands, if not millions, of scenarios to determine best outcomes. This is done using hundreds of input variables. Prior to the arrival of ML, we would try to answer the same questions using a spreadsheet and a handful of variables. With this tool, an analyst might be able to play out 10-12 what-if scenarios, but certainly not the breadth and depth of what can be done today with random forest and neural network algorithms.

BIG POSSIBILITIES

Three primary concerns for any business are: increasing sales, reducing expenses, and mitigating risk. Each of these concerns raises distinct business questions, and machine learning is not appropriate for answering them all. Yet, if the goal is to produce more revenue, then there is a place for ML to improve customer engagement by better segmenting and targeting customers. And if the emphasis is on lowering spend, ensemble models can be used to identify the marketing efforts that are effective, while eliminating marketing tactics that yield comparatively low ROIs. Finally, machine learning is being used to identify anomalies—blips in the matrix—where unusual behavior gets flagged, questioned, and investigated. The use of cluster analysis for fraud detection by insurance companies is the classic risk mitigation use case. 

Machine learning is transforming how decisions are made. Its limits are mostly bound by the amount of creativity that analytics teams have, and the culture that enables the acquisition of the right technology stack, people, and data assets. So, in returning to where we began, there is an inherent competitive advantage to having a better analytics tool—let machine learning be the hammer to your nail, the spray to your bug.