By Kevin Gray and Koen Pauwels
Why marketing science?
- You’re a brand manager and looking at piles of data that are each telling you a different story about the health of your brand.
- You’re a research director at a marketing research agency struggling with a proposal for a major client. This is a competitive bid and you know you will need something extra to have a chance to win the work.
- You’re a data scientist at a manufacturer and out of the blue have been asked by marketing to dig up insights from existing consumer survey data. You don’t know much about marketing or survey research and you don’t even know where to begin.
- You’re a research executive at a marketing research agency, and will need statistical support to conduct a sophisticated key driver analysis and segmentation for a project you’re writing a proposal for.
- You’re a CMO and have been challenged to increase revenues by 20%. Beyond your extraordinary creativity, you’ll need more than a repeat of last year’s plan and want to know what has worked in previous years and what hasn’t.
- You’re an IT director and have been asked to set up a data mart for marketing analytics. You know a lot about data and data bases but next to nothing about marketing or analytics.
But, what exactly is it?
These were just a few examples of where marketing science can come to the rescue and add significant value. But first, what exactly is marketing science? Here is a fairly bland definition from Wikipedia:
“Marketing science is a field that approaches marketing – the understanding of customer needs, and the development of approaches by which they might be fulfilled – predominantly through scientific methods, rather than through tools and techniques common with research in the arts or in humanities.”
It’s actually a lot more interesting than this! Besides project work, many marketing scientists are involved in R&D, quality assurance, training and business development. They typically wear many hats.
What skills do marketing scientists need?
Marketing science, as its name implies, requires a solid grasp of the scientific method which, contrary to the impression some may have, is not just math and programming. Besides marketing, some background in the social and behavioral sciences is also very helpful. In our view, a marketing scientist is essentially a quantitative social and behavioral scientist specializing in marketing and not merely a number-cruncher. In other words, context matters. Business matters. A lot.
Contrary to what some software vendors may claim, advanced data analysis can seldom be done proficiently with automated software. One reason is because the analyst often must choose among competing statistical models that are very similar according to model assessment criteria such as the BIC or AIC, yet suggest different courses of action to decision makers. The meaning of the various models to decision makers is the decisive factor, and judgment and experience on the part of the analyst are essential. The exception to this rule is when all that is really required are predictions that beat chance expectation and there is no need to understand the mechanisms underlying the data. In other words, all we need to know is the “What” and we don’t care about the “Why”.
Even in these cases, though, costly errors can be made. For instance, Google Trends was heralded as a wonderful predictor of flu epidemics, but then faltered because Google did not examine the complex drivers of searches for flu symptoms (especially with Google Autocomplete). Likewise, the predictions may be correct, but the execution requires human sensitivity, as in Target’s sending pregnancy-related coupons to a family who did not yet know their teenage daughter was pregnant. Relying on automated software has been likened to drinking and driving…
Being able to understand the qualitative aspects of quantitative research, therefore, is an aptitude good marketing scientists must have. Good marketing scientists are tenacious but, at the same time, creative problem-solvers who can feel what the data are saying. There are analogies with music – a technically gifted classical musician is unlikely to go very far if he or she is unable to play with feeling.
Interpersonal and communication skills are important for many jobs and marketing science is no exception; an analyst who is clever technically will be underutilized if he or she can’t get along with people or communicate with them in language they can understand!
Unexploited opportunities in old and new data
“Old” research techniques continue to show key unexploited opportunities. Take surveys, which so often are short one-off affairs or, conversely, monstrously long and contaminated by respondent fatigue. In either case, the analysis often consists of simple cross tabulations and perhaps a correspondence map or two… Piles of crosstabs can be both hard to interpret and misleading. Typically, cross tabs only examine two variables at a time and running lots of them increases the risk of capitalizing on chance, even when adjustments such as the FDR are made to significance levels. Piecemeal approaches to data analysis thus give us an incomplete or misleading picture of the puzzle. “Bad surveys are a booming business,” in the words of one veteran marketing researcher.
However, this need not be the case. A well-designed survey can gather a great deal of useful data that can be leveraged with more sophisticated analysis such as Structural Equation Modeling, Vector Autoregression or Hierarchical Linear Models. We’ll show you some examples in a bit. Though advanced analytics are frequently conducted “after the fact” they generally work best when “designed into” the research, which usually requires the involvement of a marketing scientist early in the process. This person will assist in designing the research and, ideally, will be the one who carries out the modeling. Combining and analyzing data from other sources (e.g., customer records) with survey data is becoming increasingly popular. Again, marketing science comes into play here.
As for the “new” research techniques, much of what is now called “Data Science” tends to be focused on data management, which is understandable because the requisite computer science skills can be quite specialized. The analytics, however, are usually fairly basic and often consist of predicting whether a customer falls into one bucket or another, for instance. Most data scientists do not have strong statistical skills – knowledge of causal modeling1 in particular is rare – and have little knowledge of marketing.
This is not a criticism – theirs is a different role requiring a different skills set. Few data scientists are marketing scientists and few have been trained or hired to do marketing science. Looked at from the opposite angle, few marketing scientists would be competent in an IT-centered role, either. Agendas from a major Data Science conference a major conference on Statistics point to the many differences between these two fields.
Marketing science in action
Marketers, marketing scientists and data scientists can, however, combine and synergize their skills and experience in an ad hoc fashion that does not require substantial investment or organizational restructuring. Here are a few simple examples of marketing science in action:
- A software company conducted a UX and satisfaction study among a sample of its customers. Key driver analysis was performed with an advanced variation of structural equation modeling (SEM) and the results used for new product development as well as revision of user manuals and customer online help.
- An online retailer wanted to reallocate marketing across tens of options. Suspicious of last-click attribution, they commissioned Vector Autoregressive (VAR) modeling that considered long-term effects and interactions of not just bringing prospects to the retailer, but also of increasing check-out and revenue. They found a much higher revenue impact of content-integrated marketing actions (e.g. affiliates and price comparison sites) versus content-separated actions (e.g. emails, retargeting) and increased revenues 17% with the same overall marketing budget.
- A credit card company wanted to ascertain what certain consumer segments are seeking most, and a conjoint and segmentation study among general consumers was conducted. The results challenged some important assumptions long taken for granted, while also providing context for some earlier marketing research the company had commissioned.
- A fast moving consumer goods company wanted to adapt its global dashboard to reflect different performance effects across mature vs. emerging countries. A Hierarchical Linear Model showed that advertising awareness and brand love are key in the former, but consideration and word-of-mouth are key in the latter.
- A financial services company undertook a segmentation study among general consumers and a sample of its own customers. Internal customer data were leveraged to enrich the survey data and flesh out the segments, and to help management better understand the Why underlying the What for their own (and possibly competitor) customers. For example, one key finding was that risk acceptance/aversion, general financial sophistication and other attitudes predicted interest in new investment vehicle concepts beyond that explained by life stage, demographics and historical behavior.
Like marketing itself, marketing science is continuously evolving and in 20 years time will probably look quite different from the way it does now. Though there has been tons of hype about Big Data, the Internet of Things and Artificial Intelligence, they are likely to have a real impact on our lives as citizens and as marketers at some point in the future.
Marketing science increasingly requires high levels of technical sophistication (e.g., Bayesian statistics, Support Vector Machines) and advanced computer science skills. While this may seem impressive, a potential drawback is that fundamental skills will be watered down because there is too much to learn in too short a time. Moreover, there are now many analytics software products on the market that are very easy to operate but require almost no knowledge of research or statistics, and untutored users can easily make costly mistakes. Consequently, though marketing scientists can accomplish things they could only have dreamed of a few years ago, the risk of shoddy analytics has also risen. There are always downsides to progress, though, and we don’t wish to single out marketing science.
In some ways, marketing science has long been a kind of secret weapon for savvy marketers. Though it has not received the sensationalized press coverage many other professions have and, in our view, has been underutilized, it is a vital tool for marketing research now and will be even more so in the years ahead.
1 “Causal modeling” involves the analysis of potential causal associations but the term does not necessarily imply that the researcher is attempting to prove causation. This is extremely difficult – some would say impossible – even in experimental research.
Co-Author Koen Pauwels is Professor of Marketing at Ozyegin University, Istanbul and Honorary Professor at the University of Groningen. He received his Ph.D. from UCLA, where he was chosen “Top 100 Inspirational Alumnus” out of 37,000 UCLA graduates. Next he joined the Tuck School of Business at Dartmouth, where he became Associate Professor in 4 years and received tenure in 6. Prof Pauwels is Associate Editor at the International Journal of Research in Marketing and has received the most prestigious awards for more than 30 top publications. He consulted large and small companies across 3 continents, including Amazon, Credit Europe, Inofec, Heinz, Kayak, Knewton, Kraft, Marks & Spencer, Nissan, Sony, Tetrapak and Unilever.