Marketing and the Data Science Dance

by Kirk Borne

One of the biggest challenges encountered when using big data in marketing is knowing where to start, what to explore first, and what are the most informative initial queries to pose. If data science were a dance, then we would know what to do — follow the choreographer’s lead and take our cues from them. Well, this is critical advice in dancing, and great advice for data scientists too. Taking cues from data science is as easy as one, two, three.

1. Know the number one objective of your exploratory data inquiries.

This should always be your first clue (…cue): What insights do you need to derive from your data? New leads? Better leads? New classes of behavior of your objects of study (e.g., customers or events)? New features to provide deeper insights? Better characterization of known objects or events? Therefore, examining your business conscience and examining what types of data you have are crucial first steps.

Knowing your business question before you begin the journey is fundamental hypothesis-driven science, especially essential in data science.

2. Know which of the two basic categories of Machine Learning you are applying.

  • Is it supervised learning: training on your historical data, exposing its predictive powers, and then deploying classification, predictive analytics, or regression models?
  • Is it unsupervised learning: discovering the natural clusters (segments), associations, links, outliers, trend lines, principal components, or hyper-dimensional data surfaces, consequently exposing the patterns in the data?

Knowing which of these two approaches is driving your marketing use case and delivering actionable insights is your second cue from data science.

3. Know which of the three classes of discovery is the most golden “needle” (i.e., the most exploitable object of interest — customer, object, or event) in your big data haystack.

  • Are you searching for enigmatic known unknowns: aiming to discover the behavioral analytics rules or explanatory causal factors for known objects of interest that you have not previously been able to explain (i.e., unknown) and thus far you have been unable to exploit?
  • Are you searching for exploitable unknown knowns: aiming to discover new previously unseen (unknown) examples of known objects of interest — e.g., customers who are ready to engage, to be convinced, to convert, to come back, and/or to become your loyal fans in the marketplace?
  • Are you searching for elusive unknown unknowns: aiming to discover and exploit new categories of customers, behaviors, objects, or events that may open new markets, reveal new opportunities, or inspire new innovations?

Knowing which class of discovery defines success for you (whether through business metrics, KPIs, targeted marketing campaigns, personalized offers, or lead scoring models) is your third effective cue from data science.

Following these three data science “dance steps” is a good start. A greater outcome is achieved if you are also able to understand and explain what you have accomplished. That is why the journey to “understanding through data science” is peppered with these four data exploration imperatives: examine, expose, explain, and exploit. Taking these cues from data science, as well as adopting a data-driven business culture, can kickstart your next marketing analytics project into high gear (I mean, high-stepping performance).

Taking these cues from data science, and adopting a data-driven business culture, can kickstart your next marketing analytics project into high gear.

For more from Kirk Borne, follow him on Twitter: @kirkdborne

Blog post originally published: May 19, 2015

Share this:
Tags:   Data Science

Kirk Borne

Advisor to Syntasa | Principal Data Scientist, Executive Advisor, First Data Science Fellow at Booz Allen Hamilton

Dr. Kirk Borne is the Principal Data Scientist, an Executive Advisor, and the first Data Science Fellow at the global technology and consulting firm Booz Allen Hamilton. Before that, Kirk was professor of astrophysics and computational science at George Mason University for 12 years, specifically in the data science program. Prior to that, he spent nearly 20 years supporting data systems activities for NASA space science missions. Kirk has a B.S. degree in Physics from LSU, and a Ph.D. in Astronomy from Caltech. He is an elected Fellow of the International Astrostatistics Association for his lifelong contributions to big data research in astronomy. Since 2013, he has been listed consistently each year as a top worldwide influencer in Big Data and Data Science on social media.