By continuing your visit to this site, you accept the use of cookies. These ensure the smooth running of our services. Learn more.

  • Can science still contribute to marketing progress?

    Pin it!


    Marketing research has advanced by incorporating contributions from a variety of scientific fields, mainly in psychology and sociology, but also from mathematics (specifically, from statistics).

    Does marketing have much to gain from current scientific advances?

    In what fields?

    For what applications, and with what kind of benefits for the corporate world?

    These are the questions underlying the topic of the month dealt with in our article in Market Research News.

    I was interviewed on this occasion by Thierry Semblat (*)

    Thierry Semblat: Is it really a given that marketing will be able to move ahead significantly thanks to recent scientific developments?

    François Abiven: Yes. Scientists have progressed in areas of knowledge that should have a far-reaching impact on marketing. One of the most important of these fields, to me, concerns our knowledge of human behaviour, which is linked to neuroscience and a better understanding of how the brain works. The findings of Damasio, whom we've been talking about for some years, have shed important light on the automatic behavioural system, the subconscious, and the role our emotions play in the choices we make and in the decision-making process. That certainly makes us take another look at how we conduct marketing campaigns and surveys.

    What's so extraordinarily new about this research? The discovery of the subconscious is not actually a new development....

    What's new is precisely what we've learned about how we make decisions. Before these discoveries, we thought of the human being as being able to dissect and analyze his or her available options in a thoughtful, rational way, and to make decisions on that basis. But now we realize that in most cases the decision is much more "automatic", unconscious: the brain searches our memory to help us make decisions, going through circuits very rapidly, and the conscious mind comes in later to justify and rationalise the decision we've already made. These mechanisms are widely accepted today, no one can seriously deny them.

    What does that imply for marketing surveys?

    It makes us think twice before we base our approach exclusively on what consumers say. When we look at survey protocols that give the consumer two options for deciding which packaging design is the most effective, it's obvious now that we get answers that are completely rationalised, therefore skewed.

    But these are two different things, one being placing the consumer in the position of an expert (for example, to decide which package is most effective), and on the other hand, recording the consumer's perceptions…

    That's true. But what's clearly discredited now are these overly long questionnaires that ask consumers to express themselves on how they perceive various items, to finally ask which product they'd buy. That's doesn't really fit in with the consumer's reality. Most often, the consumer is not in control of his or her choice. Of course, we've known about the limits of declarative knowledge for a long time: that's why we use projection techniques in qualitative studies. It's a well-integrated first level that is still wholly relevant. But we have to go farther, and rethink the protocols for quantitative studies to take account of the spontaneous, emotional factor.

    What sorts of techniques are taking us in that direction? 

    Carrying it out to the cutting edge, we can talk about MRI techniques that can show us what portions of consumers' brains are activated by different stimuli they are presented with. We know, for instance, that Coca Cola and Pepsi Cola don't stimulate the same parts of the brain! It's an interesting approach, but we still don't know if we're pushing the right buttons. A simpler angle could be electro-encephalograms, which are being more widely used, in the UK in particular. They also give interesting results, but here too we're getting data that we can't exploit directly. And fairly recently I discovered a panel, in the USA, made up of consumers who accepted to be  hooked up to electro-dermal intensity measurement devices, which record the conductivity of the skin. This measures emotional impact. All these techniques are intended to measure that impact, going well beyond the method based on declarative statements. But there's still the question of how to interpret it.

    It's clear what the keystone is: the limits of rationality. We use techniques to measure what is not rational, what stems from the emotions. But what you're saying is that the difficulty common to all these techniques is that we don't know yet how to use them effectively.

    We need to progress in two areas here. The first is industrialising the processes for these types of studies, and we're moving ahead quickly in this field, making much less expensive and intrusive tools and equipment. The other key is indeed interpretation, defining the standards, and framing the knowledge acquired. But we're moving ahead!

    Are there any other techniques?

    Of course. Particularly eye-tracking systems, which we use at Repères to measure emotional impact. More specifically, we measure the degree of emotional excitement by looking at pupil dilation, seeing how intently the respondent eyes the stimulus, the blink rate, and so on. We are now also able to structure a scorecard for the respondent's reaction.

    This works well, and it's easy to set up. But the limit here is that we know how to measure a certain degree of emotional excitement, but we don't know if it's positive or negative...this means that we have to combine these methods with open questions. But the technique we're developing today, that we're very interested in, is facial recognition.

    So let's talk about that!

    The idea is to incorporate non-verbal and emotional responses by recording and analysing consumers' facial expressions and gestures. And there, we can clearly determine the valence, whether the emotion is positive or negative. We know how to determine whether the person is expressing joy, fear, surprise, or something else. The basic emotions are well apprehended, codified and validated universally. Naturally, we go back to the work of Paul Ekman, the American researcher who's been studying facial expressions since the 1970s, revealing an entire language of expression. And this is innate to all humans; even persons born blind use the same facial expressions as those who don't have that disability to deal with. We've done several experiments on those bases with foods, and more recently with a sniff test in cosmetics, or in concepts.  The results are really very interesting. We've expanded Paul Ekman's classification of basic emotions to include non verbal descriptive language specifically adapted to our test protocols. We've been able to verify significantly finer discriminations between objects tested. Now we are in the phase of standardising the measurements and codifying these expressions to fit them into survey processes.

    What are the main advantages of this type of study?

    One of marketing's big challenges is to reduce the number of failed new product launches, of which there are many today--some say as high as 90%. Of course it's very difficult to draw the line between the quality of the concept and the execution and implementation aspects. But the fact is that the predictive capacity of "traditional" approaches to marketing studies is insufficient. These new approaches should be better at predicting consumer behaviours.

    Another interesting aspect of these techniques is that they enable us to better select concepts and products, an area where the results based on declarative knowledge are most often fairly "flat". Whereas the advertiser's question is of course: which one of these products or concepts creates the most enthusiasm, which one has greater impact? 

    Does this completely undermine the usefulness of declarative knowledge? 

    Oh no! For me, they should be combined. In the trials we've conducted, matching declarative with non-declarative findings has revealed a great consistency. But the non verbal adds a certain depth, a contrast, which is indeed superior. Keeping the declarative approach is a supportive first step! But above all, the idea is the measurement, the real score, will come from the non verbal, whereas the explanation will come from declarative knowledge--particularly from open questions that enable us to pinpoint the consumer's spontaneous associations.

    Are there any other scientific advances that seem important to you?

    There has been some progress in economics, and in statistics.

    We are witnessing the development of a new branch of economics called "behavioural" economics, which is coming on strong and is now widely recognized since the Nobel Prize was awarded to Daniel Kahneman in 2002. This branch invalidates classic economic theory which sees man as "homo economicus", a rational being who always makes the most useful choices for himself. Behavioural economics demonstrates and explains that on the contrary, human beings often behave in a way that seems paradoxical and irrational. For example, the fact that for human beings, the aversion to loss will most often win out over the desire to win, which produces inertia. There's also another example, very useful from a marketing viewpoint, that we find in a very well done book by Dan Ariely: when you ask people to choose between product A and product B, then you add a third product which is a degraded version of product A, it makes the latter much more attractive.

    We haven't got very far yet in determining how to use that in marketing studies, but I'm convinced it will have impact. What is clear, however, is the usefulness of taking into account the effects of the context in which the decision is made (for instance, different offering scenarios). It makes you want to learn how to manage complex experimental schema.

    All this is closely related to the development of neuroscience…

    Yes, the bottom line is the idea that human beings are not simply rational beings. We have to take into account parameters from the subconscious, the emotional realms. At any rate, it signals the end of purely utilitarian visions of things.

    You spoke also about important progress in the field of statistics.

    Right. There have been some discoveries in the field of what we call power laws, and fractal laws. Benoit Mandelbrot is one of the foremost theoreticians, with very interesting applications in the financial field which are in turn expanded on in an exciting book by Nassim Nicholas Taleb: The Black Swan. The questioning of "normal distribution" is particularly interesting from a theoretical point of view. In other words, there are certain phenomena for which it is perfectly valid to think in terms of averages and typical differences, especially with regard to physical sizes (people's sizes, for example--where the notion of average size is meaningful). However, there is an entire area of existence, often linked to human activities, where those laws do not function (the idea of "average" doesn't have any particular meaning, for example, when speaking about individuals' incomes). 

    What this is basically saying is that reasoning theoretically on the basis of normal distribution us to underestimate the probability of extreme events coming into play.

    The dramatic consequences of this skewed reasoning were flagrant in the financial crisis of 2008. The great majority of financial risk projection models are unfortunately built around the assumption of normal distribution, which considerably underestimate the possibility of variations.

    What kind of application could this have in the field of marketing?

    It's still applied very little, but power laws are undoubtedly interesting in estimating future sales, for instance. They would also make it easier to deal with phenomena such as the "long tail", which is widely used in numerical economics (explaining that in final cumulative calculation, total demand for things in low demand is greater than demand for products in high demand). Here again, we are practically just beginning to incorporate these advances into marketing and into surveys, but they open many roads for our future progress.


    (*) This article is reproduced with the permission of MarketResearchNews. You can find by clicking on this site all contributions (in French) to this topic: Yves Krief (Sorgem), Jean Paul Frappa (Expert in data analysis), François Laurent (Adetem and ConsumerInsight), Bruno Poyet (IM! impact mémoire--"Memory Impact") and Eric Janvier (Numsight).