Simulate, to create a winner

Simulate, to create A winner (optimising the design, positioning and launch strategies)

by Dirk Huisman, chairman SKIM Group

Part 1
At various levels in the organisation and with different time horizons in mind new technological options are tuned with changing customers needs. For the innovation and product development process it is essential that the customer needs be taken into consideration from the earliest phases on. The less impact the "voice of the customer" has in the early phases of the development process the more technology driven the product will be. Market research can play different important roles in the development process.
In this paper I will introduce a model to measure the impact of the value creating elements. The roles of the market researcher can play are based on this model. The researcher is able to identify which customer value is created and what the impact of the elements adding value to the customer is for the company.
Underlying the model are a number of advanced research techniques to measure sensitivities to concepts, features, price and brand and to identify the meaning of these value creating elements to the customer. The model provides the information based on which choices of individuals can be simulated. The simulation tool enables the researcher to manoeuvre the new products to pole position and create the best starting position for product launch.
Innovation and the new product development process

Innovation is a combination of systematic and continuous processes aimed at creating a competitive advantage. The competitive advantage is either realised by creating more value to the customer or due to lower cost in providing the product or service.

The innovation process is both technology driven and marketing driven and more specific it is often technology push and demand pull. In the systematic innovation processes customer needs are linked with technological options.

Products are a combination of different technological solutions. For instance a mobile phone includes energy supply solutions; communication technology solutions; information technology or chip technology solutions; liquid display technology solutions; etceteras. But also products that do not seem to be technological like ice cream or nappies include many technological solutions.

Technologies are continuously improved resulting in potentially better solutions. Consequently there are constantly new opportunities to improve the products. This creates the technological push of the innovation process.

The demand-pull in the innovation process is driven by the unfulfilled needs and the changing expectations of the customers resulting in marketing opportunities. Like the technologies, the customer needs are also in development. This is due to changing social and economic conditions, due to marketing activities and due to new products introduced in the past, etceteras. Ultimately new products create new needs. New products first substitute existing products, which might result in adapted behaviour and changing life styles and ultimately might result in changing social structures 1. Examples of this development are cars, mobile phones, and "micro waves".

The flow of new opportunities and the evolution of needs require a company to tune and evaluate the technological developments and customer needs systematically. Which does not mean that all innovations are the result of the systematic link. Many successful innovations were the result of a coincidence. However, the systematic tuning and the systematic interaction between marketing and technology create the environment to identify "the coincidental opportunity" 2. A recent and well-known example of a successful innovation that was the result of a coincidence is "Viagra". During the development process of a cardiac drug the effect of "Viagra" was found as "a frequent side effect" of that cardiac drug in development. The opportunity identified was in a short period of time transformed into a successful product introduction.


Figure 1: The effects of technology-push and demand-pull

Evaluating the opportunities and tuning the technological options with customer needs is a strategic decision process that is executed at various levels, with different time horizons in mind. At the top there are the strategic decisions regarding the direction the organisation has to go. The input regards information on trends and changing life styles and information on basic technological developments and possible innovations. At a lower level the decisions regard questions like how should the product portfolio be structured and which markets or market segments should the company focus upon. But also questions like "which features and solutions should be developed to meet (future) customer needs" should be answered. The level below regards the development and introduction of new or improved products.

At each level the decisions are part of a process in which a number of phases can be distinguished. Figure 2 reflects for instance various phases of a new product development process.


Figure 2: An example of a new product development process

In the early phases of the product development process most options are still open and there are many degrees of freedom. At each decision and after each phase the degrees of freedom are reduced and the characteristics of the product are defined more in detail. The degrees of freedom do not only differ for the various phases of the new product development process, but they also differ at the various levels on which innovation decisions are taken. The higher the decision level, the more degrees of freedom and each decision at the top will in the end reduce the degrees of freedom at lower levels.
It is therefore of essential importance that the needs and expectations of the customer are taken into consideration from the earliest phases on. The less impact "the customer" or the voice of the customer has in the early phases of the development process the more technology driven the product will be.

It is the duty of market research to provide insight into the market and in the needs and expectations of the customer. Market research can play a role at each level and in most phases of the development process. The better the researcher understands the needs, habits and dynamics at each level or at each phase, the more important his/her role will be. In this article I want to limit myself to two different roles. The "test driver role" and the "race driver, manoeuvring to pole position, role". Both roles highlight the importance and the impact of the involvement of market research in an early phase.

The test driver role regards the co-operation with the technicians at the level above the new product development process or in the earliest phase of the new product development process. According the QFD (Quality Function Deployment) paradigm 3 the quality of a product is created in the early design phases, in the beginning of the innovation process. At this stage the role of marketing is subordinate to the technicians or technical marketing. Many product options are open; still it has to be decided in an early phase which options should be pushed, which ones held back and how the development process should be directed. The impact of those decisions is enormous, as at this stage the consequences and benefits offered by future products are defined.

The race driver, manoeuvring the car for pole position, role regards the co- operation with the strategic marketers and the product team responsible for the development and marketing of the new product. This role has to be played later in the product development process when the product must be defined in detail, when the targeting and positioning of the product has to be determined.

Market research can play many other roles. For instance at the top level in elaborating and defining future scenarios or in the other phases of the new product development process. However in this article those roles will not be discussed.

To understand and describe the role the market researcher plays, we first have to create a frame of reference and explain some of the techniques used by the researcher.

Creating value
The innovation processes do not stand on their own, but are part of the value creation process. The specific definition of value depends on the person or organisation whom it concerns: customer value; shareholder value; employee value; etceteras. As the different persons and organisations are part of the same economic system it is always possible to relate the different value definitions to each other. Innovation is aimed at creating a competitive advantage either realised by creating more customer value or by lower cost.

Customer value can be defined as: the sum of the utilities for the customer of all tangible and intangible product characteristics related to the price the customer has to pay. A customer will choose among the available alternatives the combination with the highest value/price ratio.


Figure 3: The definition of Value

To create customer value a company has to:

1. Create the right and differentiating product resulting in the wanted experience or benefit looked for in a specific situation by a specific type of customer;
2. Organise that the product is there when needed or asked for;
3. Activate demand by tuning the flexible characteristics (like price, packaging) to the specific customer in a specific situation;

As company, marketer or market researcher one can be innovative in creating value at each of these three basic functions.

The model we want to discuss and which is used in this case to measure the impact of the innovation process regards the first function. How and in how far can we create the right and differentiating product and what will be the result of the introduction?

The interrelated and partly overlapping strategic marketing decisions to be made for the new product are positioning, targeting, segmentation, product profiling, product differentiation and pricing. These marketing decisions have to be made in conjunction with each other. But the product and the decisions taken in an earlier phase of the product development process condition those strategic marketing decisions to be made.
If the marketer would be able to simulate the impact of his/her decisions he/she would know the impact of each decision in advance and consequently he/she would be able to take the right decision. No doubt this would effect the role of the marketer in the innovation process as well. If simulation would indicate the bottom line results of the product development decisions the marketer could drive the innovation process.
The simulation tool is particularly of relevance when the risks are high, as is the case with innovations and new products, for which no historic data or time series are available.

To build a simulation model one has to understand, measure and model the consumer choice or decision making process, in such a way that the choices the customer will make can be forecasted. The core of our model (figure 4) regards the sensitivities of the decision-maker/buyer to the key elements in the choice situation. The choice situation might be a shop; behind a screen linked to the Internet; a boardroom; a desk when writing a prescription for a drug, etceteras. In this choice situation the decision-maker decides holistically on the image of the total product, often implicitly reacting to visual stimuli, symbols and cues. To be able to simulate the behaviour of the customer we have to measure the impact of each possible specification of each key element on his or her choice. Knowing the sensitivity to each specification of each key element, it is possible to forecast the choice, from a predefined set of alternatives (including our innovation), the customer will make.

At the choice situation the customer may often choose at a glance or on the automatic pilot (particularly with fast moving consumer goods). He/she is sensitised and reacts accordingly. Regarding the product he/she rather takes the concept into consideration than the individual product features. A concept can be defined as a combination of features offering a specific benefit or a cluster of benefits.

Knowing the concept sensitivities is sufficient to simulate the choice process and to forecast customer behaviour. But to be able to change or adapt the products in the right direction or to understand the sensitivities for new concepts, we have to know the underlying structure of the concept sensitivity. Concepts are a combination of features and the concept sensitivity is based on the sensitivity for the underlying features and it is based on the perception and connotation of the features in terms of benefits and end values.

So in addition to the concept sensitivity we have to know for each customer the underlying feature sensitivities and we have to know which features are loading which concepts. Further we have to know how the features and concepts are perceived and interpreted. Because the product choice is made in the mind of the customer, it is essential that we learn what the features and concepts really mean to the customer. In fact, it is not really the feature or the concept that is driving the choice but the interpretation or the benefit of the feature and ultimately the meaning of that benefit for the customer.

Behind the feature sensitivity and concept sensitivity and behind the interpretation there are a number of factors that could influence the choice process, the sensitivities and the interpretation. Therefor these factors have to be taken into consideration as well.


Figure 4: The model used to measure the impact of the value creating elements

The model specifies the "choice behaviour or choice function" (the core of the model) and the factors which influence that choice function. If we know the choice function and if we know how this choice function is influenced we are able to test changes (changing feature specifications, or changing perceived benefits, adding new concepts). Consequently we can test how to increase the value of a (new) product and how to create value.
Methods and techniques

Measuring concept sensitivity
The core of our model, measuring concept sensitivity, is based on Choice- Based Conjoint analysis or discrete choice modelling. During an interview the customer has to react to a number of choice tasks. In each choice task a number of products, defined as a combination of product concepts, brand and price, are shown. By choosing the customer reveals his/her sensitivities to the products and to the underlying characteristics shown.
This technique is used extensively by market researchers. One of the major advantages of conjoint analysis is the possibility to simulate the effect of a change in the definition of the product. The major drawback of the Choice- Based Conjoint technique is that the sensitivities (utilities of the attribute levels) are measured on an aggregated level (for the group of customers) assuming homogeneity. This drawback has been subject of the debate at the 1996

Advanced Research Techniques forum 4.
The second drawback of traditional Choice-Based Conjoint is that the products in the choice tasks are "phrased". In reality a customer does not read and often decides more holistically and chooses at a glance. Therefore it is essential that in the measurement process the following premises, specified by Zaltman 5, be taken into consideration:

Thought is image-based, not word based; Most communication is Non-verbal; Metaphor is central to thought; Emotion and reason are equally important and commingle in decision making; Most thought, emotion and learning occur without awareness; Mental models guide the selection and processing of stimuli.

Using new and advanced techniques both drawbacks can be overcome. To start with the second drawback, using the opportunities offered by the computer technology, market researchers are able to place virtual real products in virtual real environments in which they let the customer choose. Those visualised products are designed based on the choice task specification, but the elements are integrated into one picture.
Dealing with new products this is extremely important and one has to create in the mind of the customer, a realistic situation and a virtual real environment in which he/she has to choose between new and existing products. Darren W. Dahl et al recently described the importance of visualisation in case of new product design 6.

The first drawback (measuring sensitivities at an aggregated level) has been discussed for years. Forecasting in marketing mostly reflect aggregated data. Thus the fact that utilities are calculated for the group in total does not have to be a problem. However, then the heterogeneity in the choice data is ignored. If the heterogeneity could be recognised and detected different segments showing different sensitivities could be found, leading to more valid predictions. Latent class methods as employed by DeSarbo et al. 7 accommodate individual differences by recognising multiple segments. Applied to the choice data the model estimates utilities for each class and it estimates the probability that each individual belongs to that class. Latent class provides more insight and better predictions than traditional Choice-Based Conjoint. But the method still assumes that within the class the individuals are homogenous. And the only information provided about individuals is the group to which they belong as well as the utilities for that group. In reality even within the classes there will be a lot of heterogeneity. In particular if we go beyond the concept sensitivity measurement and try to explain the sensitivity (the link with the feature sensitivity and the interpretation of features and concepts) there are relevant and significant differences within the groups or classes.

In extension of Latent Class analysis Johnson 8 developed a method for Individual Choice Estimation (ICE). The individual utilities are estimated as linear combinations of basis vectors (for example the Latent Class group utilities). The predictive value of ICE is slightly higher than the predictive value of Latent Class because ICE is better in capturing the heterogeneity. However, for us the greatest advantage is the fact that we have utility values for each individual, so it is possible to analyse what determines those utilities or the related sensitivity.

When ICE was developed Hierarchical Bayes methods were studied with Choice-Based Conjoint data as well 9. The strength of Hierarchical Bayes is that it is able to provide estimates of individual utilities given a few choices by each individual. Like ICE it borrows information from other individuals. Hierarchical Bayes seems to be more stable than ICE and it needs less choice data. The downside is the computation time. So with large data sets one needs strong computers and time. Because the examples in the next paragraphs are all based on studies completed in the past 5 years in none of them Hierarchical Bayes has been used, but with stronger computers it is likely that Hierarchical Bayes will substitute ICE. The type of results, individual utilities, is the same.

Measuring the factors explaining concept sensitivity
The advantage of individual utilities as basis of the market simulator is that the simulation model can be extended with data from the same individual collected in the other modules. To explain the sensitivities these data regard feature sensitivity and the meaning and interpretation of the features and concepts. To explain the sensitivities but also to adapt and or correct the simulator the data regard awareness, user habits, motives, attitudes and situational and demographic data. Because for each individual we have a unique data set including all variables regarding that individual the sensitivities and simulation results can be linked to and explained by all the other variables in the model. And all the other variables can be used to optimise the overlapping strategic marketing decisions.

To explain the concept sensitivity we limit ourselves in this article to measuring feature sensitivity and measuring the meaning of the features and of the concepts.

To measure feature sensitivity, in particularly for product development reasons, adaptive conjoint analysis is a common and widely accepted technique. The advantage of this technique is that the sensitivities are measured in such a way that the sensitivity can be related to the specification of the feature. For each specification it is known how sensitive the customer is and comparing the various specifications we know where or between which specification the product improvement is most effective to trigger the customer. The second advantage of the technique is the fact that the sensitivity to the features is measured in conjunction. Otherwise every feature would have been important. The disadvantage of the technique is the fact that it is time consuming and a computer is required during the interview.

An alternative technique is the "chips method". A customer receives a number of chips and must distribute them over the features, either to express his needs for product improvement or to express how important the different features are in the choice process. The advantage of this technique is, like with conjoint analysis, that the importance of the features for the customer is measured in conjunction. The other advantage is the simplicity of the task. The disadvantage is that it is not known for which specifications of the features the customer is most sensitive. Both techniques indicate sensitivity and importance but don't provide interpretation (why is the customer sensitive).

To explain concept sensitivity the insight in the meaning and interpretation of the feature or the concept to the customer is most important. Because the customer chooses in his/her mind we have to know how a specific feature or a specific concept is linked to a perceived benefit and the end value the customer want to achieve. To link a specific feature or concept to the benefits which are important to the customer when selecting that feature or concept the Laddering 10 method is used. The theory behind Laddering or means-end analysis is that individuals are primary driven by a number of basic values, which are important to that individual. If a product is associated with that value the individual is inclined more to react (positive). Product concepts or features are associated by the individual with benefits, which in turn are associated with other benefits or with end values the individual tries to achieve. In a Laddering survey the means-end chains detected are presented in a hierarchical value map.

Laddering interviews are difficult and require specially trained interviewers. Besides analysing a Laddering survey and identifying common structures in the complex qualitative data is time consuming. In practice Laddering surveys are expensive and qualitative in nature. However, to draw quantitatively and statistically grounded conclusions, which are required to make the type of marketing decisions described, qualitative Laddering surveys are insufficient.

The solution found is to create a structured Laddering interview using a series of predefined benefits and end values. This Association Pattern Technique 11 (APT) is a quantitative approach to measure means-end chains. The academia, convinced that a quantitative and a more economic approach are needed, focus on APT.

The interpretation or perception of predefined benefits might differ between customers. In the Structured Open Association Pattern (SOAP) technique each customer is invited to add benefits or to specify his/her interpretation of the pre-defined benefit selected. Comparing the results of a Laddering survey with the results of SOAP the structure of the value maps is identical, but now more respondents are able to reach an end value and statistically the results were more valid. Besides the analysis time was reduced by 60%. Using the SOAP technique to explain the concept sensitivity is more than promising, in particular in case of the positioning of new products. The concept or feature sensitivities of customers following the same means-end chain can be analysed, but it is also possible to use the value map of customers showing the same sensitivity to explain their sensitivity. This is of relevance for positioning and for directing product development.

Summing up: new methods and research technique to measure and explain the concept sensitivities for each individual enable market research to identify how to create customer value. Consequently, innovations in the computer technology as well as innovations in research methods and techniques enable the researcher to play a more dominant role in the race for innovation and market leadership. Two roles the researcher can play will be illustrated: the test driver and manoeuvring for pole position.

Market research: the test driver
Market research can play a crucial role in defining the requirements, for the features and concepts to be developed, to meet customer needs in the future. This role is played before or in phase one of the new product development process. The objective is to define which features create customer value and should be developed or should be included in future products. At this stage the role of the researcher is comparable to the role of a test driver in the racing industry. The test driver feeds the technicians or the product team with information to tune the development process to the customer needs. To feed the technicians the researcher can provide insight into: the feature sensitivity pattern; the feature benefits; the benefit segments.

The feature sensitivity pattern
The sensitivity to product features and product functions changes over time. Customers are sensitive to a feature within a specific range. And that range might change over time. Measuring feature sensitivity every half-year or every two years one can detect specific patterns.

The sensitivity to a specific feature (i.e. battery life 12 hours) may at the beginning be of medium importance. Over time the feature might become more important. In particular this will be the case when a technological break through results in a new standard. The customers are sensitised and the sensitivity to the feature growth (figure 5, pattern a). After a period of time all products comply with that specific feature specification and it starts to become standard. Consequently, the sensitivity has increased but the impact on the choice process drops. However, the increased sensitivity is now a negative sensitivity: if the product does not comply with the new specification it will be a reason not to choose.


Figure 5: Feature sensitivity patterns

The pattern described does not typify all features. Some feature improvements do not result in a greater sensitivity. For instance, if the number of phone numbers to be stored in the memory of the mobile phone goes beyond the level of practical use (e.g. a memory size of 500 instead of 100), the sensitivity will follow a flat pattern (b) or a slowly declining pattern (c).

Another pattern reflects the impact of changing user habits or of an innovation regarding another feature. The sensitivity for a specific feature can be flat and rather low for a longer period of time, but at a certain moment in time the sensitivity increases (d). For instance, the data-transmission option in a mobile phone only became important after the penetration and increased use of Internet.

Systematic measurement of feature sensitivity provides a framework to identify the impact of a feature improvement before the definition of the product. Using the adaptive conjoint analysis technique and measuring the sensitivity for various levels or specifications of the feature enables the researcher to detect the feature sensitivity pattern.

Feature benefit
A first sensitivity measurement indicates the importance but not the pattern. Therefore it is important to "dig deeper". In combination with the feature sensitivity measurement one can systematically measure the perceived feature benefit and the benefit importance. Customers perceive features and don't have problems linking specific benefits to features, even not to completely new features. For instance in the car industry, features can be classified as offering active safety; passive safety; comfort; luxury; prestige; etc. Of a new feature it is not known what pattern it will follow but right from the beginning one can establish what benefits are linked to that feature.
Based on the systematic measurement of feature sensitivity for a number of product we can conclude that features tend to follow a pattern of features which are linked to the same benefit. Further, we can conclude that the sensitivity of features linked to an important benefit tend to increase faster. And lastly, we can conclude that the sensitivity to benefits is much more stable over time than the sensitivity to features. Consequently in an early phase, in addition to the initial sensitivity, the sensitivity pattern and future impact of a feature might be identified.

Benefit segments
The sensitivities to features offering the same benefit correlate. It can be specific customers who are looking for that benefit and for the features offering that benefit. In other words, customers might belong to a specific benefit segment. To tune the product development process to the needs of the market it is important to identify who are sensitive to the new feature (in development) and with which other features the new feature correlates. This indicates which segments should be addressed with the feature or in what part of the product range the feature could be included. At the same time the management has to answer the question whether that segment is reached or can be reached by the company.

Tuning the development of features to specific segments is important because it results in a better fit to the customer needs. But it is also important because knowing the size and strategic importance of that segment indicates the potential revenue of the development process.

The test driver dilemma
Measuring the sensitivity for totally new features the researcher has to deal with two problems:

How to test the possible impact of new techniques, new technologies or new features which are like "algebra" to the consumer, without leading him/her to an answer?
Will the sensitivity to and the impact of the new technique or new feature change when the awareness grows?

With regard to the first problem "how to measure the sensitivity to something which sounds like algebra" there is no simple solution. But it is essential that the feature is visualised or that the customer gets an "mental image" 12. Further it is essential to bear the premises specified by Zaltman 13 in mind. And lastly, the features or techniques should always be phrased in common language, like the marketers would do in case the feature was a part of the product. However for methodological reasons one should prevent specifying the benefits of the new feature. Comparing the perceived importance for the same feature phrased by the technicians and by a marketer we found a significant difference. Customers were more sensitive to the feature as phrased by the marketer than phrased by the technician. This was primarily due to the fact that more people understood the feature and consequently more people were sensitive. In addition the sensitivity of that feature, as phrased by the marketer, correlated better with the sensitivity for other features offering the same benefit.

With regard to the "awareness dilemma", based on the monitoring of the feature sensitivity for eight years, we can conclude that a higher awareness does not by definition lead to a higher sensitivity. However for those features that are linked with an important benefit there was a correlation between awareness and sensitivity. So if in the beginning an unknown feature is linked with an important benefit it is likely that the increased awareness will in the future result in a higher sensitivity.

The test driver role of the market researcher is different from the role the market researcher plays normally. The role differs because the audience is not dominated by marketers, but by technicians and general managers who are allergic for marketing (research) jargon. Besides the time horizon is different. To play the role successfully the researcher does not only have to be completely in control over the techniques used, but he has to articulate the results in a way the technicians and management understand the message.

Market research: the race driver, manoeuvring to pole position
The race driver role, manoeuvring the new product to pole position, differs from the test driver role. Often the product is defined and at this stage the impact of the customer needs on the development process is limited. But knowing the sensitivity of each individual to product concepts, price and brand and knowing the underlying feature sensitivity and interpretation the researcher has the tools to test how to reach pole position. The market researcher can drive the racing circuits a thousand times. Each time he is testing what is the impact of a change in the product profile, combined with price and combined with the brand, for a specific target group in a specific market (competitive field).

To test the impact the researcher adapts the concept by changing one element in the product concept. In marketing terms this would mean that the perception of product by the customer is adapted. For instance, the efficacy of a drug for elderly people could be defined as the perceived extension of the period ("x" months) before admitting an elderly person to a nursing home. By defining and changing the definition for "x" the researcher can test the impact of "x". If "x" would be 3 months the prescribing physician would not react, but if the perceived extension of the period ("x") would be 6 months he might choose the drug. Aggregated over all physicians we learn how many physicians will react and what the impact will be from the change in perception from 3 to 6 months.

Circling the racing circuit the researcher proceeds as follows:

1. Define market (=specify all elements of all product on the market) and test, based on the measured sensitivities for all customers in the simulator, the number of customers who would choose a specific product (=share of choice). To test the researcher uses a logit model.
2. Compare shares of choice with real market shares and calibrate model.
3. Introduce new product and test share of choice.
4. Adapt one element of the new product or of the product concept and test the impact on share of choice.
5. Proceed adapting the product profile, within the constraints of the physical product definition, till a profile (perceived benefits) is defined which results in the largest share of choice. Apart from the physical product constraints one could impose cost constraints as well and optimise within a specific price range or even simulate with an increased price to test how to maximise revenue.

Simulation does not have to be limited to the own product but can reflect the assumed reaction of competition as well. So manoeuvring to pole position includes anticipating competitive reactions as well.

To reach pole position it is not enough to have the best car (the optimised product profile). One should aim at the right target as well. This is where the importance of measuring and simulating at the level of the individual becomes evident. The simulation process, circling the circuit, could be executed for each type of customer or each segment. Instead of optimising the product profile for the total market it is possible to test what the optimal profile and key benefits would be for a specific target group. Being selective and targeting a specific group, which is willing to pay much more, might lead to a higher profit overall. As the computer does the actual driving and circling the circuit it is possible to test for 48 hours or more without a minute rest. As such one could test for all combinations of customers and define the optimal product-market combination.

Finally, there is no reason to limit the race to one car only. Instead of defining one new product one could define two different versions. Now manoeuvring to pole position becomes really interesting as each version of the product can be tuned to a different target group or segment and could be profiled to maximise revenue or market share in that segment and increase total revenue.

The results of the simulations have been tested and proven in the real market place. In one case (soft drink) the simulations per type of outlet have after two years been verified with Nielsen data and were proven to be correct. But also in the tobacco industry, other fast moving goods industries, medical equipment industry and the pharmaceutical industry the simulation results reflected what happened afterwards in reality. A few times the simulation showed to be not precisely correct, mostly because of an insufficient market definition or due to a skewed sample.

How to manoeuvre to pole position differs per industry. Simulating in the fast moving consumer goods markets one will learn that targeting (simulating the optimal target group and optimal message or product claims for the target group), pricing and simulating per type of outlet is critical. Simulating in the automotive industry, the pharmaceutical industry, IT industry or in general the technology driven industries product profiling and benefit segmentation is more critical.
Although there are differences between the industries, like there are differences in manoeuvring on a wet circuit and a dry circuit, there is not one approach per industry. Each race is different and each time the researcher will have to check before simulating how the future of the market "he is driving on" will look like.

The market researcher has the tools, the techniques and the capabilities to manoeuvre to pole position. The strategic importance of reaching pole position for a company launching an innovation is very high and justifies race drivers' salaries.

Crashed
The success scores of new products do not justify the expectation that a car starting from pole position does not crash. We have been involved in pole position research of which the car crashed during the race. In both cases the crash had been forecasted.

The first crash regarded a product from the consumer electronics industry. From each of the different innovations (including competing innovations) which could be introduced the most likely share of choice had been estimated. The forecasted share of choice of the client's innovation was about 20%, but if perception of the product could be changed dramatically the share of choice could increase to 44%. This would require a huge investment in advertising and even then it was doubtful if the image would change as the product had some features which related to the past. Despite the forecast, the product was introduced and it was only backed by advertising expenditures that are average for the industry. After three years the product was taken off the market and it never reached the forecasted share of choice.

The second crash regarded a new drug, which was identified by the financial press as a blockbuster. As the clinical trial data were not available to the market researchers the forecast was specified for a range of efficacy levels and a range of patient types. If the product would reach an efficacy level of X the share of choice would be Y % and to reach Y in the positioning one should specify patient type 123 and one should stress benefits ABC. Already identified as a blockbuster the positioning and the claims were selected without the clinical proof of the product. When the clinical trials didn't prove the expected efficacy the researchers identified the need to adapt the marketing approach and to adapt the expectations. Neither was done, the researcher was sidetracked and the product manager replaced. After one year the drug performed according to the simulation (based on the realistic product profile) but far below expectations from the financial press.

Analysing these two crashes one can see that they stem from a failing corporate culture. With regard to the second crash the cause is based on the fact that the technicians (clinicians) didn't want to co-operate with the marketers in an early stage. In both cases there was an underestimation of the power of information (if the massage is not wanted shoot the messenger).

The racing industry
As the product development time is shortening and the scale on which new products are introduced is enlarging the risks to crash have to be reduced and controlled. This stresses the fact that to successfully launch a new "VW Golf", "Toyota Yaris" or "Nissan Micra" as marketer and market researcher one has to prepare for the race like McLaren.

The techniques used to test new features and to manoeuvre the new products to pole position are as complex and advanced as the techniques used in a Ferrari or a McLaren. In the beginning it took "ages" to complete the studies. But, like in the racing industry, due to the learning curve and proven to be successful these techniques become standard equipment on our own cars (in the standard surveys).

Literature
1 Duta Soumittra, The spider in the web, EphMRA/Esomar pharmaceutical marketing research conference, June 1999, Marne la Vallee.

2 Lieshout, K. & Huisman, D., QFD, integrating technique and customer orientation, Esomar Industrial Marketing Symposium, London, 1993.

3 Akao Yoji (editor), QFD Integrating customer requirements into product design

4 A.R.T Forum 1996, Beaver Creek Colorado, American Marketing Association.

5 Zaltman, G., Rethinking market research, Journal of Marketing Research XXXIV, November 1997.

6 Dahl Darren W., Chattopaghaya Amitave, Gorn Gerald J., Use of visual mental imagery in new product design, Journal of Marketing Research XXXVI, February 1999.

7 DeSarbo W.S., Ramaswamy, V., Cohen, S.H., Market Segmentation with Choice-Based Conjoint Analysis, Marketing Letters, 1995, pp. 137-148.

8 Johnson, R., ICE: Individual Choice Estimation, Sawtooth Software Inc., Sequim, November 1997.

9 Allenby, G.M., Ginter J.L., Using extremes to design products and segment markets, Journal of Marketing Research XXXII, November 1995.

Lenk, P.J, DeSarbo, W.S., Green, P.E., and Young, M.R., Hierarchical Bayes Conjoint Analysis: Recovery of partworth heterogeneity from reduced experimental designs, Marketing Science 15, pp. 173-191.

10 Reynolds, T.J., Gengler, C.E., Howard, D.J., A means-end Analysis of brand persuasion through advertising, International Journal of Research in Marketing 12 (October 1995), pp. 257-266.

Reynolds, T.J., and Gutman, J., Laddering theory, method, analysis and interpretation, Journal of Advertising Research, February/March 1988, pp. 11-13.

11 Hofstede, F. ter, Audenaert, A., Steenkamp, J-B. E.M, Wedel, M., An investigation into the association pattern technique as quantitative approach to measuring means-end chain, 1998, International Journal of Research in Marketing.

12 Dahl, Darren. W., op cit.

13 Zaltman, G., op cit.