Last week I presented at the XXVIII INFORMS Marketing Science Conference held in Pittsburgh, Pennsylvania. Along with Jan Hofmeyr from Synovate, and Gaurav Bhalla from TNS, I was there to discuss the issues facing Marketing Science in the next five years, from the practitioner’s point-of-view. Perhaps naively, I expected that academics would be interested in real-world applications of their work, but only 30 of several hundred conference attendees chose to attend our session. Apparently the pursuit of tenure, a key objective in academic life, makes publishing more important than seeing the application of one’s work, and contributes to the enormous gap which exists between practitioners and academics.
It was apparent from the few academic researcher presentations I attended that model sophistication, not practical application, is important. Here are a couple of examples:
• Analysis of 239 new product launches in France using a polynomial dynamic linear diffusion model resulted in two conclusions that practitioners might find difficult to swallow: promotion is far more important to new product success than advertising, and uniqueness has a negative relationship with new brand success.
• Application of a heterogeneous hidden Markov model to physician prescriptions for a new drug produced interesting results. The model, however, totally ignored the fact that the prescriptions would not just be affected by detailing, sampling and promotion for that drug, but also by the incidence of people visiting the doctor with the relevant disease, and the marketing for competitive drugs.
So is it any wonder that three practitioners, though speaking to very different subjects, had a similar theme running through their presentations? That theme can be summed up as “Please apply your statistics in the context of a general paradigm of buyer behavior that practitioners might recognize.” As a qualitative researcher in the audience put it, no matter how good the statistics, “many models fail because they assume something unreal.”
Guarav led off the practitioner presentations, focusing his attention on key driver analysis. His premise was that clients are “hungry for understanding, knowledge and deeper insights” and need explanation and prediction to help them make important decisions. But he described the practice of key driver analysis as “an exercise in brute empiricism,” often divorced from any set of expectations of how the variables included in the analysis should fit together. I would agree that this is an ongoing problem. How many times have we heard something described as important, when in fact it is simply correlated? Too often, too little thought is put into whether the correlation, important or not, is cause or effect.
Jan followed, suggesting that the application of many different analytic techniques to the same issue implies a lack of a unifying theory. Stating simply “It’s not good enough,” he took aim at work that reported high R2 values but only explained 10% of the data, and research that used attitudes to explain intentions rather than real behavior. He then reviewed the workings of the Conversion Model as an example of one that has good explanation of behavior - although, as usual, his chart purporting to be a two-dimensional representation of five-dimensional space left me somewhat at a loss, even when the basic principles did not.
My ten-minute critique focused on the shortcomings of market mix modeling, as currently practiced by many in industry or academia. In the presentation I highlighted three key issues:
Most models still seek to explain sales of one brand in isolation, even though survey research tells us that few people think of themselves as completely brand loyal in any category. I shared some of the work done at Millward Brown which models the share fight within a category, allowing clients to understand the relative strength of their marketing activities much more clearly.
Even the best models rarely explain more than 50 percent of the variation in volume sales. This is the equivalent of describing the waves on the ocean but ignoring the depths and the currents underneath. A few academics and practitioners are beginning to integrate continuous survey data into their models to help represent the longer-term influence of brand equity. This could also help solve the problem that many influences on purchasing are simply not measured, e.g., advocacy, event marketing, and viral marketing.
Finally, much of the thinking behind marketing mix modeling is rooted in a consumer-packaged-goods mindset, with a short-term, dollars in/dollars out expectation that is inappropriate to considered goods and business-to-business categories. We need to do a better job of matching model structures to the realities of a particular business. For example, in many categories the job of advertising may be to drive traffic, while other factors, such as salespeople, the strength of the product or service offering, or convenience, may determine the final sale.
The reaction of our audience suggested that our ideas won’t be heeded any time soon. Questions focused on what new issues clients would be facing in the next five years – implying that the questioners think the the current questions have been answered. Sorry folks, I just don’t think that is the case. Until new techniques can sensibly be applied to practical issues, business decision makers will be left wondering about their relevance, or, worse still, making the wrong decisions on the basis of misleading results.
(24 votes, average: 3.29 out of 5)
Loading ...
June 14th, 2006 at 1:57 pm
As a practitioner wearing a semi-academic hat (adjunct professor), I couldn’t agree more that the state of academic research in marketing is frighteningly removed from real problems. I can’t think of other disciplines that suffer this foolishness — not medicine, not engineering. I recall when I studied applied social psychology in my post-grad courses — an introduction to the text stated (as much as I can remember) that if we were there to change the world, then we should leave. But if we had in mind some neat regression model or correlational analysis, then by all means feel welcome. Thankfully I escaped, almost unscathed.
June 17th, 2006 at 3:02 am
Practioners and academics have different objectives and also measurements - generally, academics are rarely required to be evaluated by the real market share or benefits. They are expected to provide some analysis perspective and fresh ideas (though maybe useless or incorrect)
June 17th, 2006 at 11:09 am
I guess these comments confirm why I did not contemplate staying on at university. As my colleagues know, I love to try and figure out the general principles behind why people think and behave as they do, but theory without any application to the real world seems such a waste of time and effort.
June 23rd, 2006 at 3:39 pm
Hi all,
I am a new academic (one year graudated with a Ph.D), aka “junior faculty”, housed in a Department of Advertising and Public Relations in a mass communication college (you would think we’d be in the Business School, but apparently the inclusion of Media Planning and any creativity related courses precludes that). I too firmly believe and wish that academia and industry could work more closely together. But I’m in somewhat of a ornery mood today so:
Where were “you” (practitioners”) at the last American Academy of Advertising conference? Or the American Marketing Association’s last two Educators Conferences? Didn’t see many there, as you might expect.
There is a whole universe of comments, from both practitoners and academics, about how we academics do not try to make our work relevant to industry or try to disseminate findings in a readable fashion.
This implied notion that somehow marketing/advertising academics have some sort of “obligation” to the industry is very interesting, because the obligation doesn’t seem to work both ways. To horribly mangle a now-cliched Cuba Gooding quote - “how have you helped us help you?” Phillip Herr mentioned the classic enginnering and medicine line. The difference is that enginnering and medicine departments/schools actually receive quite a bit of industry funding, data/equipment access, and just plain moral support.
What about marketing? Funding? Very rare…when was the last time an ad agency or marketing research company funded a professor’s academic research (and I don’t mean consulting work, which is separate from an academic’s day job)? How about access to some decent company/indsutry data? Ah, well that’s proprietary…again, very rare for an academic to be given access, even to rather old data (this, despite the fact that identity of the company and specific details can be withheld when publishing in most academic journals).
And what about the tenure thing, and trying to publish rather than see our work being applied? Most marketing academics would LOVE to see our work being applied. But rightly or wrongly, tenure is a key means by which academics are rewarded (it’s certainly not with monetary riches), and tenure is only gotten through publication of either academic journal articles or academically oriented books. So it’s no wonder us non-tenured junior faculty are busily pursuing it.
So: academics have very little support from the industry, receive almost no “credit” or incentive within academia for building closer ties with practitoners, and somehow have an unspoken “obligation” to help practitioners with no reciprocal obligation? Perhaps that’s there aren’t more of us academics at these conferences….
June 28th, 2006 at 9:41 am
Nigel,
You might want to check this post on MMM out:http://www.bivingsreport.com/2006/pr-and-roi-a-critical-crossroads/
The PR industry magazines are suddenly touting Marketing Mix Modeling as the silver bullet.
Best,
Rita
June 29th, 2006 at 11:32 am
Federico, thank you for a comment from the other side of the divide. I understand your frustration and you raise some good points. I will say in our defence that Millward Brown has partnered with academics on a couple of occasions but the experiences have been mixed, good and bad. I know for a fact that many advertisers, like P&G and Coca Cola, fund academic work, so the money is there, but there are a lot of people who want to get there share.
June 29th, 2006 at 11:43 am
Rita, thanks for the comment. Sir Martin Sorrell, CEO of WPP, our parent company, has also been applauding the benefits of Market Mix Modeling. Given the intense pressure on marketers to prove ROI it is perhaps not suprising that he would do so, but I believe that it is ultimately a risky strategy. Every purchase is the last step on a path that starts with the consumer recognizing a need and ends with them making a decision to purchase a specific brand to satisfy that need. There are many things which may intervene along the way to effect that decision, but things that have their influence nearer to the point of purchase are likely to have a more visible impact than things that come earlier on. It does not mean either one is more or less important than the other. Advertising and PR typically have their influence before people get to the point of purchase, and so are disadvantaged in the modeling analysis.