Apodemo conference – Lisbon – nov. 96
rev. Stockholm – may 97

Relier la sémiologie, les études de marché et la théorie du chaos, dans les années 90, c’était risquer de passer pour un hurluberlu frappé de démence. Or, la théorie du chaos apporte des clés d’interprétation, de diagnostic et de décision stupéfiant. Malgré sa complexité, le caractère abstrait du propos, la présentation de cette théorie a attiré de nombreux responsables de marketing en France et en Europe. Elle est toujours valide plus de vingt ans plus tard.


CHAOS is a theory that developed from the late sixties to the eighties. Like many new theories, it appeared in many areas at the same time (meteorology, physics, demographics, biology, mathematics and even in economy). 

Each time it appeared (without being called the chaos theory, it was related to the same kind of event: an apparent disorder growing in an impredictable way within a supposedly well ordered system.

As many new theories, it led to three odd developments:

  1. a pure refusal of crazy and uncredible elucubrations (we never saw that before, thus it must be wrong)
  2. an over expoitation for any kind of subject, even subjects with no possible chaos exploitation (gambling)
  3. a recuperation by alternative people (New Age, pseudo metaphysics, market research)

Behind a fascinating notion that suggests the abandon of rational thinking, there is a very serious and innovative scientific background that can be very difficult to understand (for instance non-linear equations). This very difficult and technical background can be at the origin of a pure refusal or of an over exploitation of a theory. A so complicated background allows almost any transposition in a private delirium.

Our purpose in this paper, will be to avoid both derivations:

  1. Not to be so technical that nobody can use it
  2. Not to derive silly and odd consequences from a very important and useful knowledge.

One of the most interesting aspects of CHAOS theory is its capacity to provide a better rendering of innovation market:

  1. by giving new explanation on how innovation occurs
  2. by giving a new explanation on how consumers may adopt it or not.

The following paper will develop and illustrate the major keywords of chaos theory by:

  1. illustrating them as the source of reflexion for theoricians
  2. explaining the most understandable theorical interpretation to such phenomenons
  3. giving some keys for using and taking it into account in market research. 

The major issue of this paper is not to simply propose a more or less new toolbox for analysts, based upon old knowledge, in fact it is the reverse. We can still only suggest new possible tools, but the overall principle is to introduce a completly new view on marketing, what we could name «non linear marketing».


The most well-known notion about chaos is the story of the butterfly wing fluttering in Peking that will provoke a hurricane in Miami. 

The usual interpretation of that story is that small and remote events may have considerable consequences. But, above this anecdotical idea, a very important concept is operating.  This concept can be described as « the sensitivity to initial causes ».

We generally agree on the fact that if all the conditons for a test are « globally » gathered, this test will reproduce almost the same results. The remaining differences can be considered as « a small noise in the system » and therefore be neglected.

Chaos theory just says the opposite: The small noise, the micro-events and differences in the initial conditions will produce growing differences in the following events development.

This phenomenon was first explored by meteorologists who discovered that « very similar » initial conditions lead to progressively more and more different developments. Therefore, they invented the butterfly wing fluttering story to illustrate it in a very evocative way.

In a more theoretical way they created the theory of unpredictable results do to minor initial different conditions:

  • at the begining, nothing seems to happen,
  • then some differences appear that were in general described as neglectable phenomenons
  • but after a shorter or longer while, the two phenomenons develop as if they had nothing in common.

The cause for such phenomenons has been explained via the mathematical equation that must be used to describe them. The classical models used  « linear equations ». Such equations obey to the principle that small causes have small consequences. By contrast, chaotic phenomenons can only be described via « non linear equations » in which any event has growing consequences along with iterations.
Hence, a linear equation will not reflect the influence of minor differences, noises, initial events, but a non-linear one will make such elements be more and more important.

The following table shows such a development from « very similar initial condition » towards a totally impredictable pattern:

By working on extremely context dependant matters, marketing research is necessarilly very sensitive to initial causes:

  1. respondents personal context
  2. interviewing structure
  3. stimulus material organisation

If we apply the same kind of analysis to the process of innovation adoption and integration, we can notice that it may rather well correspond to the big question marks about the integration of the innovation by mass market consumers:

At the begining, most innovative products or brands have a similar growing curve in being adopted by a few opinion leaders or trend setters,

But the integration in mass market implies many iterations of the same attraction/interest/need event. Then, minor details that did not have yet an influence begin to play a major role and determine the slope of the curve.

The most important consequence of this primary element of chaos theory is that the destiny of an innovation is not only depending on major aspects, obvious responses to innovation, but also on minor and simple items that will grow in importance along with the basic fact of being many times confronted to the purchase/not purchase situation.

Non linear phenomenons can also be related to the notion of POSITIVE FEED BACK that were studied by the Palo Alto communication theoricians.

Positive feed back occurs when a cause will reinject  its influence at each cycle of a communication or situational pattern.

Those researchers discovered that most of human psychological and communication problems are linked to the famous system of « always more of the same thing ». They noticed that most conflicts are born in very little and minor events getting amplified through times.

Therefore, human thought is a non linear and virtually chaotic system giving unpredictable development to normal situations alterated by minor events.

Such a derivation from chaos theory is a clue to the interpretation of group (or crowd) strange and even odd thinking, attitude or behaviour.

It is also a clue for analysing in detail what small cause could produce such an effect in individual interviews, above all when the responses become extremely divided.


Chaos is disorder, yes but not any disorder. It is disorder in a system that was foreseen as being in order.

The usual way of thinking relies on progressiveness and linear evolutions. In other words, a system is supposed to evolve towards a new state by slow and continuous alterations. A Gauss curve is meant to represent such phenomenons by showing a majority of « normal » response to a stimulus and minor alternative responses, with a direct correlation between frequency and difference from the mainstream:

Chaos theory just says the opposite:

  1. Chaotic developments appear all of  a sudden and then they follow a process that is closer to catastrophy theory than probability
  2. Chaos doesn’t obey statistical rules by appearing at an unpredicatable rate in an unpredctable repartition

To illustrate such a phenomenon, let’s mention presidential elections.
All the indicators gave Mr X to be the easy winner, up to the time when, suddenly Mr Y who was the « looser » grew fastly in the polls and was elected by a majority of people, even by those who were supposed to be his worst ennemies.
This event cannot be explained by usual theories because it contradicts them or at least makes them very uneasy to apply. It is typically a chaotic event.

The same situation often occurs in market research as well. In particular, it can explain how an innovation will:

  1. slowly spread from trend setters to mass market
  2. brutally invade all the market place
  3. die for unclear reasons.

Even if chaos theorie doesn’t give an instant key to the above problem, it reflects it better than the usual disciplines and techniques.

There are two illustrations of chaotic events:

  1. The first one is the turbulence system. If one puts water in a recipient and heats it from the bottom, we can notice three stage:

The evolution from stage one to stage two is progressive and seems to say that the evolution from cold to boiling water is linear, but, suddenly, with no apparent cause and much before water is boiling, the systems becomes completly disordored and can only be described as a chaos, even in the most stable (noise free) conditions. The turbulences are chaotic and can only be described by chaos theory.

The second example is the folded pastry or the Smale horseshoe. If we have a pastry with two small points or objects in it, just next one to the other at the begining, the process of flattening and folding again and again the pastry will lead to the situation that those two points or objects are at a strictly unpredictable distance at the end of the process.

This illustration shows that the elements constituting a chaos have a random distribution within the whole system. One cannot foresee where elements will be located, due to the dynamic of the system itself.

Even if this system appears to be simple (flattening and folding again) it creates complexity. In other words, chaos is not due to sophisticated organisations, only to the iteration of very symple factors.

This surprisingly leads to the famous second law of thermo-dynamics: every ordered system tends towards disorder.  Water coloured in yellow and water coloured in blue will necessarilly mix into green water which is disorder.
The same happens in human brain and thinking. Ideas, opinions and even creativity are the result of unpredictable combinations due to turbulences (accumulation of informations) and to concept manipulations (lateral thinking is nothing more than the accidental encounter of information units during simple associative operations).


We usually describe the process of innovation as a system acting on two parameters:

  1. it is the introdution of a new knowledge that breaks current norms. For instance, electricity changed all the set up notions about the way to produce and get light, in particular it broke the traditional opposition between day and night by providing easy and strong light during the night.
  2. it creates an unstability because people have to learn this new knowledge, adopt those new norms, and live in a different way. Therefore, during a shorter or longer time, there is an active opposition between the old and new system. At the end of this period, the innovation is adopted or not, in other terms it becomes a stable norm or not.

This situation of generating new norms and unstability is solved when the innovation is integrated and DEFINITLY  replaces the old system. It comes again when a new innovation reproduces such a conflct and makes this status quo obsolete also.

All the innovations cannot be ranked at the same level. In fact, marketing tends to call innovation products and features which are almost non innovative at all (a new decoration on a pack, a variant that already exists in the competition). Most so called innovations are just bare me too activities. Hence, we determined three levels in innovation:

  1. CONTINUITY :  the simple me too activity proposed as innovation by advertising but provoke neither new knowledge nor unstability;
  2. EVOLUTION :  this is the domain of improvements and better performing products that simplify and facilitate life. This the domain of problem solvers, of product responding to needs and expectations. Most innovations belong to this category (instant sauces and soups, new computer or camera or TV performances…). They provoke low unstability and only improve knowledge and norms.
  3. REVOLUTION :  this is the domain of great discoveries (electricity, numeric, antibiotics, telephon, television, airplane…). They generate new ways of considering time and space, of considering life. They occur at a much slower rate than any of the above categories.

It is usually considered that the higher an innovation is ranked on the scales of norm and stabilty, the longer is the time cycle for its adoption, integration and obsolescence. For instance, it took many years for integrating electricity and air travels. By contrast, the so called innovative yoghurt with fresh strawberries will have a much shorter life cycle since it brings in no new knowledge and have no influence on our food habits. 

But this system doesn’t apply all the time. For instance, the adoption of antibiotics which was a real medical revolution, was almost immediate; many evolution products (CDs, APS photos, dish washer tablets) were integrated in a very short time. By contrast, continuity and evolution proposals, that take a much longer time than expected to be adopted, are countless.

Sometime, the integration time rate can fluctuate by growing faster or slower for more or less unknown reasons.

The explanation, if not the solution, is simply that CHAOS theory plays a considerable role in innovation. The unstability produced by an innovation corresponds to the turbulences in a system that will take a new structure in an unpredictable time scale.

Normally, the higher an innovation is ranked, the longer it should take to be adopted, but it can also create such conditions that the market takes instantly a new shape.

This aspect of the problem should therefore be in the focus of all innovation research.

Instead of investigating the liking or disliking (which actually are the signature of no innovation), we must examine the kind of change the innovation brings in and the reeadyness of consumers to accept and integrate such change.


The initial definition of chaos is disorder. When a serie of events seem to be completly disordonated and have a completly impredictable repartition, we are used, in our common language, calling it a chaos.

In order to illustrate such a situation, we must start with order. For instance, an oscillating pendulum swings in an ordonated way with an easily calculable period. If no reinforcement is given, this pendulum is attracted by one point where it remains steady:

But, just imagine that the movement of the pendulum is maintained and that another movement is introduced, for instance an up and down oscillation. Then, the movement and the positions of the pendulum become impredictable and clearly represents a chaos:

In fact, even if those positions and movement cannot be precisely determined One can also notice that all the points where the pendulum will only be located in a given surface that will be more and more precisely drawn by iterating the movement many times.

The actual route of the pendulum is unpredictable. One can even say that the pendulum will never go twice on the same route. From the starting point, it will draw an infinite line.

But this line will be localised in a precisely defined surface, never out of it.

This surface, that can also be a volume is called « strange attractor » by the inventors of chaos theory. Why is it « strange »? 

Because a finite space is representing an infinite situation: 

If we try to apply the « strange attractor » phenomenon to market research, we can easily notice that all the occurrence that can be produced around one given subject, whatever it can be, will generate the space of responses. Even if the responses seem to be at random at first sight,  the accumulation of them are organised around a « strange attractor » that will give a sense to the chaos.

In the innovation theory, we also find strange attractors: they may correspond to the establishment of a knowledge and behaviour pattern. As long as the innovation is not integrated, it will not appear in the space of the current strange attractor designed by the consumers knowledge and behavioural patterns:

For instance, in the late seventies, in teh medical industry, the beta blockers (complex anti-hypertensive substances) were described by doctors asvery intersting and new substances, but they never appeared in their spontaneous knowledge and behaviour patterns. In fact, it took quite a number of years for doctors to integrate beta-blockers. When it happenned, the market start to really develop till the time new substances made beta-blockers more or less obsolete under the same conditions.

The fact of considering something as new is not sufficient, it must also correspond to an obvious integration in the strange attractor of our knowledge and behaviour patterns.


Chaos obeys to a sort of determinism. All the events that belong to the chaotic system will display in a given space.

Such a space cannot be defined until enough iterations of events are provided.
At the begining, points (or responses) seem to arrive anywhere in a complete and unpredictable disorder.

Then, after enough information is collected, the points (or responses) are still in a complete disorder, but they strictly stay within the space of the attractor. 

The accumulation does not increase the predictability of responses, but leads to a clearer and clearer definition of the framework, of the space that is organising the responses.

The accumulation of very simple items gives birth to a complex and detail object:

In other words, chaos may be a determinist system  when it is analysed at its structural level rather than at each event level.  That means that a systemic analysis is more relevant than micro analysis of categories of details.

This does not exclude differencial analysis since a chaotic system may be divided in sub-systems that are by  themselves.  In other words, chaos theory doesn’t forbid differencial analysis:

The configuration of an attractor system may have considerable consequences in terms of market research analysis since it can provide decisive informations on the consumers decision making.

For instance one can localise in the consumers discourse all the items that correspnd to a buy/non buy attractor system. This may be rather basic, but doing so, it is also possible to identify the change factor: the moment when the consumer change his attitude that correspond to a «phase change».

The possibility to identify the change of phasis can operate within a single consumer discourse, but it can also be identified on the basis of a great number of consumers allowing many iterations of the same questioning/topic. This is the field of lexicometric analysis in which word and expressions are litterally collected and clearly related to their context.

The change of phase, in other words, the change of attitude can therefore be linked to an objective  and concrete expression and lead to the identification of the decisive criteria for modifying consumer’s attitude.

We may now consider a change of phase as the key phenomenon in the integration of an innovation. Till the time innovation is not integrated, the discourse is organised around one attractor. As soon the integration occurs, the discourse goes around another attractor.

This phenomenon is sudden, not progressive. Since the innovation implies different views and norms, it brutally reorganise the whole consumer discourse:


The markhovian theory tells us that each element of a discourse has an influence on the following ones and that each new information item increases the predictablibility (reduces the choice) of the next one.

The first letters of a word are giving many indications about the final letters of that….
This is the reason why, after correction, so many mistakes remain at the end of words or sentences and not at the beginning: people just don’t read the end of words or the end of sentences.

This principle is true within one discourse (for instance in one sentence or in the plot of a novel), but it is also true within the sequence of different discourses occuring through time. 

Therefore, according to this theory, we can conclude that a brand very much depends on its initial discourse (its first positioning or, at least, the first contact consumers had with it):

The Shannon theory is relying on the same principle, but it enrichs it and applies on chaos.
Shannon who was a researcher working on telecommunications also noticed that the begining of a message was giving more information than the end.
But he also noticed that this principle had to be modulated:

The basic conclusion of that was that well ordered systems are less informative than chaotic ones because the repetition of items gives quickly an indication about what will be the following items in the sequence. In chaotic system, the information remains always unpredictable, then it stays all the time information full.
In market research we can conclude that too basically predictable response systems are poorly informative and therefore should require very few iterations in order to determine their attractors. This should have an influence on the way of organising research in order to avoid repetitive responses, therefore very poorly informative.

Another major conclusion comes out of the innovation theory. By creating unstability, an innovation necessarilly produce discourse diversity.  One of the most interesting keys for appreciating the rank of innovative power of a concept or a product is to determine in what extend it produces language diversity. In other words, the consensus that is often seeked as a sign of safety could easily mean that the so-called innovation is not really a big event.

When a system is stable, the discourse remains well ordered and predictable, when it becomes instable, the discourse become much richer whatever the kind of judgements it conveys. We know that in our everyday life: we use to speak a lot about something that fascinated us, to link it to many memories and ideas. By contrast, when we have to speak about a known and usual thing, we just make banal and boring phrases.

Brands are much less able to innovate than products. A brand is a notion linked to its past. Even by proposing new and innovative products, the brand is dependant on its background personality that leaves unerasable traces in its today’s discourse.
This is so true that, often, even consumrs who could not have been confronted to the initial positioning of that brand, have it in mind when speaking of that brand.

The personality of a brand is of great help to allow it to propose innovation. Just consider the personality of Philips and Sony. Sony has an innovative background, not Philips. When they proposed the compact disk, anyone aknowledged both brands of inventing it, but the real benefit went to Sony.

Products are concrete objects and thus are not that depending on a pre-established personality and set up discourse. This is why innovation is much more visible when considering products. To evaluate a branded innovation is therefore dangerous because it can neutralise a big amount of discourse richness. The brand influence should be disconnected from the product innovation evaluation, even if this relation is of major interest in a second time.

Anyway, the richness of discourse is a signature for innovation. Simple consensual adhesion shows that we remain in the continuity.


When systems correspond to one or two variations patterns, they are not chaotic. As soon a third level of variation is involved, it seems that chaos is almost unavoidable.
In order to illustrate that point, let’s refer to the pendulum. Its movements are clearly predictable all the time it is:

  1. steady (no movement means a constant response)
  2. in a single oscillation (all its positions are calculable)

But, as soon as a third level is introduced (a second oscillation), its movement becomes chaotic. This situation seems to be the basis of most chaotic systems. Above two levels of interrelated variations the systems becomes more and more unpredictable.

Another phenomenon also happens: periods of order emerge within the chaos as sorts of reorganisations. Then the system splits again and chaos reappears:

An innovation provokes chaos since it creates a bifurcation between two norms systems and therefore a dual attractor pattern. We can represent that in the following way: 

  1. before the innovation the system remained linear and stable as a one point attractor (no movement)
  2. the innovation introduces new knowledge and new norms, then it creates a two points attractor
  1. The innovation generates a behaviour and attitude unstability, generating a third point attractor

The three possible situations are all possible when considering an innovation, they imply three different time scales for integration, from never  to immediate integration.

If we consider now the level of innovation, we can notice that it has a strong influence on the chaos it can produce: 

  1. at the level of continuity, there is no change of norms and almost no unstability, therefore the chaos generation is neglectable («I prefer the pink one or the blue one»)
  2. at the level of evolution, the bifurcations are more visible and generate a chaotic situation within a narrow range
  3. at the level of revolution, the bifuractions are well marked and generate strong contradictory positions.

However, we can oppose continuity with both the two other situations because the later ones are the only to generate a real chaotic situation and therefore provoke an actual integration process. Continuity just maintain a status quo even if it pretends to innovate, which is most irritating for modern consumers.

The marketing illustration of the attractor concept can be illustrated by three well-known examples:

  1. Nutella, on many markets is the historical leader in chocolate spreads. Most launches of competitors products, even with better organoleptic performance or better marketing preparation, have more or less failed. Nutella is Nutella, it is the word for such a product and the situation remains stable, despite of the attempts to make the situation move.
  2. When two major brands lead the market, they create almost the same stable situation with no real need for a third competitor. They share the market and create a strong and well defined oscillation in which one can recognise his own positioning and target group.
  3. The situation becomes much more unclear in atomised markets (dairy, drugs, drinks) in which many brands and products are competing in offering similar or hardly noticeable benefits. Then chaos becomes the rule of the game. In the absence of real innovative event, only short terms event can occur.

We can then discover that a chaotic system may alternate periods of order and periods of complete apparent disorder. Each new period is a sort of reproduction of the previous one, but at a smaller scale.

The market place is a very good candidate to chaos. The development of brands, ranges of products, and subvariants within products ranges and varieties are setting up many variators that create chaos at several levels. This system has led to an almost non legible universe in which no one (not even the producers) are able to define any relevant codings. 

Confronted to such a disorganised universe in which too many decision criteria are involved to make the offer match the real needs, the consumers react at random with a decreasing loyalty and involvement.

Since the system is chaotic and provides no reading keys, the consumers have created their own keys for decoding it.

But it would be an illusion to consider that consumers put order in the market place chaos. Consumers are also driven by a diverging plural system and, therefore, are producing their own chaos in the market decoding.

The aim of market research is not to presume that consumers could provide an ordered vision of the market place, it is just and barely a way to decode how their chaos work and respond to it the best manner.

Classical marketing and psychological procedures tend to construct consumer attitudes as a more or less linear phenomenon:

A – Mrs X has never been exposed to the Brand Y and therefore has no attitude towards it;

B – Mrs X is exposed to the brand Y and is not convinced to buy it;

C – Mrs X is more exposed to the brand Y and happens to be convinced.

This representation of consumer evolution does not reflect the reality. We can give two simple and usual examples to illustrate it:

  1. Someone decides to buy the brand, but has not really changed his attitude towards the brand, he hardly can explain why he bought it
  2. Someone has an excellent attitude towards the brand but still doesn’t buy it, he can explain it by giving side reasons («I did not think, saw, find, had time for it»)
  3. In other words, the change of attitude or phase, doesn’t obey to a pure linear phenomenon:

The above pattern corresponds to the most basic situation, with only few criteria of choice and a binary decision. As soon as the situation is getting more complex, we can assist to a real chaotic development (as soon as three factors intervene in the selection matrix).

Hence, simple purchase intentions as they are usually proposed are in pure contradiction with the chaos theory and make very unpredictable the results of such procedures (something like ±50%)…


The notion of fractal equations was invented by mathematicians like Cantor and more recently Mandelbrot. The « Cantor dust » principle just shows how an ensemble can be divided at the infinite in order to produce an infinity of segments with no length:

Later, this principle was applied to chaos by Mandelbrot who demonstrated that the same process could apply the same way at different scales.

If the Mandelbrot models had not been invented, nothing in the virtual reality or in synthesis images could exist.

At a deeper level, we can assume that a big part of the most recent knowledge about how reality is built would not exist.

The basic principle of fractals is that whatever the scale, you look at same realities, the same patterns are reproducing.

In other terms, the structure of a phenomenon is independent of the scale one looks at.

In addition, one could not interprete a reality by mixing the scales. The length of Britanny coast can be measured in kilometers. This is very useful for a geographist or a driver. But if one wanted to measure its real length in centimeters, this length could be multiplied by ten thousand. And again if it was in millimeters…

Mandelbrot demonstrated that whatever the observation scale, a chaos is structured the same way. Biologists and anatomists discovered that it was true for living beings. Sociologists are not far from considering that it is also true for social structures.
In fact order and linear phenomenons are simply in contradiction with the norion of life because they mean static systems and no development. Life needs chaos.
So does market research.

One of our most permanent concerns is to delimit reality in order to determine clear YES/NO situations and feel safe in our decisions. The fractal dimension of the world just says the contrary.  Depending on the point of view and the measuring system one can find that the Britanny coast measures 1000 km or 100 000 km or even an infinity of kilometers.

Such a notion is known since Antiquity, but its mathematical representations only exists since twenty years.

Are we that far from market research and human sciences? No.

Marketing people are trained to delimit their market and to decide whether their product will belong to this or that field or niche. The chaotic response can be very disconforting.
For instance, the frontier between bread and pastry is perfectly fractal. Whatever the level of observation, we can discover that the frontier remains unclear, even if we have a perfectly clear notion of what is bread and what is pastry

The unclarity of the frontiers grows up with complexity. One can easily draw the limits between a CD player and a loaf of bread, but such a question is rarely asked.
The frontier between a loaf of bread and brioche, the frontier between a shampoo and and a shower gel, between two competitive brands is much more difficult to establish.

In fact this situation is much closer to what happens in nature. 

For example, it is almost impossible to precisely delimit the waters attraction territories of two rivers. Depending of  the observation scale those two territories will grow in complexity, even if they remain located in the same overall pattern.
In other words, the frontier will depend as much on the observer as on the global structure.
We live in a fractal world.

This situation is also of great interest for marketing research and innovation.
In former times, it was very easy to delimitate markets and product type areas. A shampoo was a shampoo, a detergent a detergent, a chicken a chicken. The diversification of the offer, the multiplication of variants, the never ending game of 2 in 1, 3 in 1, products with no this and more that, ombrella brands with product brands and variant brands, products one can find in different versions in various departments of the supermarket,  and bare me too competition, have created a beautiful fractal consumer universe. The limits between one single offer and another are so unclear that a normally aware consumer  has great difficulty to identify it.

The marketplace has shifted from a reduced and well defined system to a pure fractal universe that had at least two main perverse effects:

  1. everything being in everything, the coding system turned towards a bland repetition of the same repertoire (everything is light, healthy and natural, even bleach…). Therefore, diversity creates a paradoxical poverty in the offer;
  2. the consumer become more and more bored by the offer. Most respondents feel irritated by not being able to differenciate between brand X super yz and brand Z ultra xy. They even cannot read a range that lost any visible structure. And when a liquid fabric washing brand starts telling him that nothing is better than its new powder version…

Anyway, the consumer assumes that any premium brand is able to offer everything, considering the repertoire of all the advantages offered by all brands as granted.

Hence, when a brand proposes, as an innovation a new quality that has already proposed by competitors, or in similar markets, or even globally on the market place, the consumer don’t consider this as an innovation because it is already a constituent of the fractal coding of the whole system.

The final consequence of this problematics is to consider that innovation cannot only rely on already existing codes within a wider market place that are read as a cloudy set of expected features, it has to be made of:

  1. the minimum basics that say at least that a soap is made to wash
  2. the new basics that say that the same soap is at least adapted to the time being standards
  3. the real new items that say that this soap is really a different and new one.


For a better understanding of the fractal concept, let’s consider attractors from each dimension:

  1. a single dimension attractor is one point (no dimension in geometrical terms)
  2. a two dimensional attractor is an infinite line in finished surface
  3. a three dimensional attractor is an object with no volume and an infinite suface:

The notion of fractals relies on the idea that chaotic reality is represented by fractionary dimensions. 

This very complex notion seems to have  very little use for market research as such. But it gives also birth to the notion that a reality is not dependant of the observation scale since:

  1. there is a correlation between the scale and the type of measure,
  2. the information given at one scale will remain true at another scale, above all in its complexity.

Chaotic systems develop when a certain number (three at least) of variable (degrees of liberty) are operating. Each degree of liberty may be considered as a dimension, therefore one can imagine universes with many more dimensions than the usual four dimensions of our perception. Universe of five, ten, a thousand dimensions…
But the fractal principle also shows that the perception depends on the observer who apprehends reality as a chaos that can only be described via fractional dimensions, sorts of in between dimensions (eg: 2.43) for describing unpredictable variations, bifurcations, changes.

The fact that the equation x3-1= 0 has only one solution is not true anymore, it gets three solutions:

  1. 1 as one possibility
  2. -1/2+ an imaginary root
  3. -1/2- an imaginary root

This may seem very abstract, but it has a lot to do with what actually happens with consumers responding to a stimulus. The single one solution is in fact the most irrelevant one, the two others are dividing respondents between acceptation and refusal.

But what chaos theory tells us is that the frontiers between yes, perhaps and no are not clear, they overlap and are finally fractal.

In other words the reasons to say  yes may be very similar to the ones to say no…

The above theoretical explanation is the root of the most disturbing aspect of marketing practice since it clearly denies the notion of safe prediction and installs the notion of risk even in the best prepared campaigns.

The situation is the following one

in a linear context, one could easily say that if all the initial conditions for sucess are gathered, there is no chance that the project will fail;

in a non linear context, linked to multiple solutions to a simple initial equation, there are always at least three options for the project development:

  1. it can be successful
  2. it can fail
  3. it can remain steady and neutral

Very often, one try to explain a failure via errors in the forecast and preparation and a success via a perfect preparation and forecast. This may be true in a number of cases, but only the chaos theory can describe (if not explain or justify) the other situations:

  1. successful launches despite of most pessimistic prognosis
  2. complete failures after perfect preparations

From  pure philosophical point of view, the chaos theory is a barrier against too much power devoted to marketing. If the development of markets and trade was purely linear, we would run the risk to be 100% dependant on the good will of major and well equiped companies. There would be no room for new and unexpected creations. In other words, marketing  has to cope with life, and life is chaotic.

Fragmentary dimensions also have the most perverse effect on the image of marketing research itself, just because of the evolution of marketing effect. For many years, gurus and very well paid experts could with no danger make solid recommendations starting with «by contrast with what the respondent said…».

A well established experience, a solid knowledge in psycho-sociology and a good nose guaranteed the relevance of such recommendations. Up to the end of teh eighties, most launches, if they corresponded  more or less closely to fashion and psychological models, were granted with success. We could eve perversly say that no one could verify if the gurus were right to advise not to launch a product or a 

message since it was rarely verified.

The growing complexity of the markets and the paradoxical loss of diversity within fractal markets has made much more hasardous the advices from gurus or careless institutes. Research has lost much of its credibility  not by being actually of lower quality, but by not coping with the new reality.

When the market gave more chance to new launches than the research probability to be right, research could be trusted.

When the market gives less chances to success, research easily risks to be 50% right, 50% wrong and therefore not being more reliable than playing with a coin.

Nowadays, many companies just mistrust research and just want to listen by themselves their consumers. It has the advantage of cutting prices and to give direct access to what consumers said.

This is definitly not the solution for getting in touch with the new markets because, in addition with the loss of depth of the investigation, we also loose the expertise for analysing the complex growing structure.

Research has first to go back to its primary and most reliable anchors:

  1. actually listen to consumers in depth and taking the time for doing so; quick and dirty interwiews and focus groups CANNOT render the market complexity;
  2. not interpreting consumer responses according presumption grids and pseudo-psychological grids, in other words, what consumers said is not a pretext for interpretation, it is the real subject we must understand;
  3. even if the consumer is not a marketing expert, he is our customer and we have to take into account his advises and to consider the diversity of his problems.

Research should no more be a black box, it should be involved and involve the whole marketing process in order to put into perspective how the whole situation is evolving.
Chaos is made of bifurcations, small changes can have considerable effects, innovation is in NPD, not in advertising.

Research is an interface between producers and consumers, this interface has to develop into transparency and in reliability for helping the producer to adress the necessary innvation market.


The Palo Alto researchers, in the sixties, noticed a very interesting phenomenon in behavioural analysis: When an information item occurs at one level of language, it almost automatically appears at another level.

For instance, if a respondent produces an important information by using language, he also  almost automatically produces a sign with gesture language:

The principle of universality is one of the most interesting features of chaos theory:

  1. the same phenomenon can be verified in all the expressions of chaos, for instance language, images and gestures,
  2. the same phenomenon can be verified at all the scale levels of he chaos, since it is a fractal universe, the structure remains constant at all levels in physics, biology and of course psycho-sociology, the most interesting feature is that the same pattern can be verified, independantly of the measuring instrument or method.

If we consider market research from the chaos theory we may derive important conclusions from the above properties:

  1. the exploration should be done by using various expression systems (verbal and non verbal, rational and irrational) in order to discover the transcodal patterns.

The exploration should be carried on different scale levels:

of the problem (in general and in details) of the target group (focused or extended)

The exploration should rely on different measuring instruments (methodological diversity).

This may seem rather trivial since it reproduces the classical catalog of means used for doing research.

But the major factor relies in the cross analysis of all the levels, objects and means of investigation. Monadic split analysis is unable to render the chaotic structure of a market research subject.

In other words, a multiple interactive set of simple observations is much more appropriated to problem solving than any big sophisticated single minded exercise.


Like the irisations on the surface of a lake, chaotic systems seem to have a strange interrelation.
Normally any chaotic information should develop according to different and unpredictable orders. In fact, the same chaos patterns seem to reproduce themselves as if they were in resonance. 

This principle has been shown by putting several pendulums oscillating next one to the other. After a while they all oscillate together…

The principle of resonance can explain the strange situation in which the same idea may appear simultaneously in different locations. The whole market place could be structured like a fractal ensemble. This is not proven, but there are few hints to imagine that, at least in a certain extend, it can be true.

We also may derive the conclusion that market, opinions, fashions don’t develop according linear processes. The growth of an idea is not a constant slowly evolving from zero to the whole market. It rather develops by « catastrophies » in an impredictable (sorry for that) and fractal process. One day it doesn’t exist, the day after it is everywhere, the next one, it doesn’t exist anymore.

In other terms, if we consider innovation, the process of its development is likely to be non linear, the cycle being rather sudden modifications than progressive mutations.
The objective of research should not be to detect the conditions for a slight evolution, but the decisive set up that will make most people shift towards adoption and integration.

This decisive mutation conditions can be diluted in many interrelated micro items that will lead to an attitude and behaviour more or less sudden modification.
In sum, innovation obeys chaotic developments and not linear evolution.

Resonance can also be noticed in market research. It is the well known phenomenon that happens in consumer workshops when respondents progressively shift towards the same attitude. 

It can easily be proven by doing individual written evaluation of a stimulus material at the begining of the sicussion and, then, compare those evaluation to the final ratings done orally on the same topics.

It is often related to «leadership», but in fact it is pure resonance.

This same phenomenon can even happen with individual interviews or questionnaires just because the interviewer or even the question sequence induce a resonance phenomenon in an apparently non inducing system.


Jupiter is an enormous planet only made of liquid gas. The movements of the gas on the surface of the planet is chaotic to an extreme point so one can only see the long horizontal layers of its turbulent flow linked to the planet rotation.
One can also notice a strange spot on the surface of the planet, this spot is a mass of gas, exactly the same gas as on the rest of the surface and also in a chaotic structure. But this chaos seems completly independant of the rest of the flow.

It is a sort of ordered chaos within a non ordered one.

Such a phenomenon also happens near us, for instance turnmoils in the chaotic flow of a river. Such events just show us that chaotic systems can be more than random phenomenons, they are non linear structures developing according to disordered rules within conventional perception.

In other words, what we call chaos is not a pure random disorder. That are only conventional analysis tools that make them be called chaotic. In fact they reflect a more elaborated reality that requires non linear analysis to be described.

Such specific events, like an ordered chaos within a more disordered chaos, can be transcribed as the specific patterns appearing within a-typical target groups.
The major conclusion we can derive from the Jupiter example is that we must be careful in making conclusions from specific target groups to mass market consumers.

Like a chaos that develop independantly of another, within another, the rules that operate in that specific target group may not be active beyond those given limits.
In other words, we should rather be aware of the structure of the mainstream attractors than of a-typical so called trend-setters who are just reflecting their own reality.


It is also possible to widden the chaos theory applications to a more psycho-sociological and historical framework, in order to develp a clearer view on cultural trends. Lets observe how it can work both on the evolution sequence of occidental societies and on the integration patterns of innovation.

The theory of semiotics describes three major factors and processes in making a conviction:

PRAGMATICS which is related to the usefulness, convenience and practical interest of a product or concept.

HEDONISM which is related to the emotion and in particular to the pleasure that is provided by a product or a brand or a concept.

ETHICS  whic is related to the cultural anchorage, the seriousness and the moral values conveyed by a product, a brand or a concept.

In general, we can consider that the sequence is generally proceeding from pragmatics, then hedonism to ethical values. This is related to the fact that needs start by being concrete, then require an pleasure added value and then are questionned in terms of moral values.

The theory of children psychological development also propose three steps:

SYMBOLIC which is the way a child develop his own status before the others (his mother, his family, other kids, the whole society).

IMAGINARY which is the way the child builds up fictional set up in order to solve the conflicts he is confronted to.

REAL which is the incorporation of his own person into actual situation via knowledge and experience.

This process that belongs to children development can also be applied to any human development since it corresponds to the normal sequence of learning and adaption to developping realities. Therefore it applies to any marketing and cultural evolution.

It may appear quite strange to make a correspondence between pragmatics and symbolics, but there is actually a correspondence  of the three steps of the two theories for explaining how concepts develop through times:

The primary level of integration majorily consists in acquiring usage patterns, so pragmatics come first. But it also comes with a symbolic loading since it is still not stable and integrated in current knowledge.  Hence, the first step of integration relies on analogy and metaphors: to behave at a table, to understand what is a CD requires a metaphorical anchorage to be justified in the current set of knowledge and understanding. For example, aviation entirely developped itself on the basis of a comparison with marine. 

The second step corresponds to the adoption of a knowledge and requires the addition of distinctive values, in general linked to the added pleasure of using a product or behaving a new way. Hence, we can notice a correspondence between hedonistic and imagination values: kids as well as consumers of an innovative device invent stories where they take place as the hero and solve their problems. A this stage, aviation generated a big amount of fictions where pilots and passengers were seen as distinctive people.

The third stage corresponds to the integration of the new set up and relies on a rational analysis of the situation. The new situation or product are tangible realities belonging to existing norms and a stable usage pattern. Hence they can be analysed as realities and quetionned on an ethical basis. Nowadays, aviation belongs to our norms, therefore, we can question its impact on our real life, flight safety, pollution, prices… All that on the basis of actual information and with the possibility of producing a nuanced judgement. This is the stage where discourse takes place (discourse means rational and sound analysis)

This evolution system obeys to non linear structures, therefore, no one should imagine that we just pass from one stage to another via well defined steps. For instance clonage, which is the innovation of the nineties, provokes a lot of ethical behaviours («you should not do it!» or «it will change life perception!»). But the difference between the three steps are obvious:

step 1: the ethical discourse is strongly loaded with symbolic issues (religion, fear…)

step 2: the ethicall discourse is loaded with narrative elements speaking about life styles and fiction

step 3: the ethical discourse is strictly loaded with the idea of experience and sound information.

This system has to be considered on two levels:

the level of the the object itself

the level of the context.

There is a cultural evolution that can easily be describes as a cycle of predominant reference system (we all agree on the fact that the nineties are ethical). In that framework, innovations occur and are considered with the meta-language of the cultural status quo.

This is the starting point for a pure chaotic system. The discrepancy between the code (meta-language, context) generates a bifurcation level over the innovation status itself (the object, the concept).

This evolution is not exclusive, it only add a new set of values at each step and gives more weight to each depending on time. In other words, we cannot consider a period that would be purely pragmatic, then hedonistitic and then ethical. There are only:

the predominance of one entry over the others;

the progressive enrichment of the repertoire of values.

There is an historical cyclic system operating with this evolution. This cyclic principle can rather easily be identified through decades since the end of WWII.
This cyclic effect corresponds to a chaos development with its basic attractors construction in three phases:

The introduction of ethical predominance should not be considered as generating chaos even if we may admit that ethical questionning (as during revolutions, in 68 and now in the considerable interrogation about our civilisation future).
We should rather considered the presence of three main attractors in the perception of the world as a source of unpredictable patterns.

Those unpredictable patterns end in general in new clearly defined pragmatic proposal that fit well with simple symbolic issues. Exactly as dictators often come to simplify the richness of revolutions by instauring one law, one thinking, one solution to every problem.


Life and marketing have more in common than we could expect, the last extensions of the chaos theory may explain it. The first aspect of this is a paradox linked to life:
By contrast with all the rules of game of the world, the more life evolves, the more it enriches its own structure and become deeper in information.

In any normal situation, the second law of thermodynamics says that all str’uctured systems tend towards disorder. For instance, if you consider a well built wall, well structured, after a few hundred years, it becomes a ruin with no more apparent order in it.

If we consider life, it is the contrary, it starts with a disordered soup and become a well structured system.

The theoretical explanation of such a phenomenon starts with the type of information needed to describe the two states:

The evolution, from a cluster of elements to a structured system corresponds to a decreasing number of complexity and an increasing depth in the structure that makes it shorter in descrition, more organised in its levels of description.
Acccording to the principle of thermodynamics, the loss of structure and oraganisation makes the information more complex and less deep.
But, how can it be the opposite way as in life and marketing?

The first way is to consider that by producing at random a great number of organisation, the probability is never excluded that it happen to fall into the right order. The only problem is that it will take an infinite time.

In the infinite number of possibilities, there is only one that create order, life or success. Hence, an other factor is requested. For a wall, it is the manpwer of builders who put every brick at the right place; for life it is an unknown energy that forces the molecules to work together, in market life, it corresponds to a well balanced investment.

Life may have happenned by an unexpected chance or as the result of a great amount of energy.

We know now that the first signs of life was mineral and that it appeared in orgenic cristals. Cristals only appear when energy is applied to a soup of molecules:

The consequence of that is that life, as well as new marketing success stories may only happen under the condition of an equation betweenn time and energy;

The chaos can get organised into a life system, a noticeable living structure under three conditions:

  • having the right ingredients
  • stabilising the equation between time and energy
  • reaching the critical mass that escape the pure probabilistic framework

If not, life can only appear by an unpredictable chance, it may just not appear. On the marketing side, this put under theory the fact that unless developing the right amount of energy (investment, share of voice), a brand can only:

  • never reach the level of real life
  • return to the normal rules of the universe: loss of structure, loss of significance, death.

Death is nothing more than the return to the basic application of the second rule of thermodynamics:

no energy means no tendency to organisation

time means a tendancy to loose depth and increase complexity

this is true for living beings as well as for brands.


The research approach to innovation has two major facets:

to render the usual well known criteria that allow to evaluate an innovation onto marketing active criteria

to integrate the chaos theory that is obviously in action when considering innovation.

On the bifurcation system, we can see that the evaluation of an innovation follows a strict and necessary agenda:

In terms of marketing research, this model corresponds to a list of decisive aspects that are listed in the following table:

This structure corresponds to

  • the active patterns of innovation as they have been described in innovation research
  • the fractal system that is active in the chaos theory applied to innovation

An innovation implies three fundamental parameters:

  • the object itself as it speaks and appears in the reality
  • the consumer as he feels and appears when confronted to the innovation
  • the relation that is established between the consumer and the innovation.

An innovation has to be significantly relevant on three essential factors:

  • its novelty value: it must be new, unique, different of whatever existed before and generating a surprise effect; if not, it is in the best situation an improved me too
  • its acceptance value: one must see it as superior to the current offer, to be ready to pay for it and be mentally acceptable; if not it is a gimmick for exceptional situations
  • its involvement value: one must find it convenient, like it and see it as a transformation in his life and aspirations; if not, it will not reach the consumers..

An innovation must be evaluated according the following criteria:

  • how it is understood and corresponds to the actual proposal
  • how it actually performs on each criterion and what elements contribute to that performance

The innovation proposal can be localised on the following matrix that determines wether it is perceived as new or not  and wether it is suceptible to be easily integrated or not.

novelty with no acceptance/involvement has little chance to get in a fast integration circuit

low acceptance or involvement make the unstability non existing  or unnecessary. Therefore the integration is simply unpredictable and not measurable according to innovation criteria.

The innovation matrix corresponds to the various possibilities of integration:

  • no integration
  • slow and / or unpredictable integration
  • fast and easy integration

The following chart shows the correspondence between the innovation matrix and the non linear integration system.  It makes as clear as possible that an innovation that doesn’t break positively the current norms and beliefs or doesn’t reach the critical mass of unstability is unlikely to provoke the necessary process of integration:

  • many marketing failures just correspond to a non noticeable change
  • many marketing failures correspond to a negative norm breaking
  • the fractal markets have generated a growing number of failures due to the lack of critical specificity.


  • Initial causes may provoke considerable divergences in the end findings. They can be reduced by checking for the relevance and neutrality of the stimulus material and by taking into account the influence of the context on responses.
  • Fractal markets have generated a considerable gap between real innovation and expectation. Many innovative features of a product are not perceived as innovative since the consumers assume that the product should already propose them. Therefore, one has to detect the level on which the critical mass is achieved for perceiving innovation.
  • Innovation is creating a new order, thus it requires a lot of energy to condense time and give structure to diluted perception. This energy has to be identified and measured, especially in terms of consumers involvement.
  • Resonance, universality and sub-order are considerable biases for judging an innovation, above all in groupwork, but also in individual interviews. Monadic exercises, diversified approaches and angles and pure mainstream recruitment are necessary for reducing those biases.
  • Liking is not linked to innovation (it is even the contrary), the process of integration is a much more complex process that has to be measured to identify the actual possibility of success of an innovation. Unstability and new norms and knowledge  that are generated should correspond to a real anchorage in motivational trends.
  • Adopting and integrating an innovation correspond to a change of phase in consumer discourse. This change should be definitive, therefore, this is the central key issue for predicting the success of an innovation.
  • Fractal dimensions in the consumer discourse have diluted the adhesion factor into contradictory positions. A positive response may lead or not to actual purchase behaviours. Therefore, one should carefully evaluate how the repartition of liking gives more or less chance to a real innovation integration.
  • Only iterations may give shape to an attractor. Iterations go along with universality which means that the same phenomenon can appear on multiple layers of expression. Therefore, the innovation check should not only multiply the number of cases, but also the number of items and angle of vision on the same topic. This means that qualitative and quantitative approaches should shift towards each other for better performance.
  • An innovation generates turbulences and unstability. This has considerable consequences on the consumer discourse. They make the responses unclear and less predictable. Therefore, the research should focus on the attractors (i.e. the emerging structures) rather than on a mere split of details into single units and details.

Une réflexion au sujet de « Chaos »

Votre commentaire

Entrez vos coordonnées ci-dessous ou cliquez sur une icône pour vous connecter:


Vous commentez à l’aide de votre compte Déconnexion /  Changer )

Photo Facebook

Vous commentez à l’aide de votre compte Facebook. Déconnexion /  Changer )

Connexion à %s