Using the Inarticulate Consumer in Product Development: A Holistic Approach

David W. Ingersoll and Maryana Kaplan

Introduction

“Who better than the end-user to help you design a better product?” (see Ingersoll 2010). The main objective of product development is to design products that delight the consumer and sell at the market targeted levels. Hence, you need to incorporate the voice of the consumer (Akao 1990) into the product development process to get feedback on the acceptability of your prototypes. The naïve consumer has no difficulty indicating what they like and dislike to such a degree that many action standards are based on the hedonic measure to very precise statistical levels (e.g., 95% confidence interval). Unfortunately once we go beyond hedonics, other overall measures (e.g. purchase intent, appropriateness, etc.) and basic diagnostics, we find that the naïve consumer has limitations in providing guidance to product development:

• The naïve consumer speaks in a repertoire of impressions and emotive imagery. The product developer speaks in a technical language with many more elements and precise meanings. Another way of stating this issue is that the naïve consumer lacks the technical language to provide meaningful direction to the food technologist. Likewise, the product developer has understandable difficulties using consumer input for meaningful product development guidance. Hence, a translator is required.
• To capture more information from the naïve consumer, the product developer needs to tap the hidden emotional reactions to test products. Hence, a mechanism is needed to reveal useful insights trapped in the hidden mind of the consumer.

The term ‘inarticulate’ refers to the two issues: 1) most consumers have language limitations in conversing with the technical product developer, and 2) most consumers have difficulties in expressing their subconscious reactions to test products. Since subconscious feelings can greatly influence attitudes and behavior, what is very much needed is a bridge to translate “consumer” into “developer” for optimum understanding. No pejorative judgments are implied (see Figure 1).

In addition to this conundrum between consumer and product developer, there is an issue of obtaining flat hedonic ratings among test products, e.g. a lack of product differentiation based on overall liking. Moreover, the situation is acerbated in that all/most of the test products are found to be highly acceptable among other overall measures (e.g. purchase intent, interest, etc) as well.

Is it reasonable to keep repeating the same approach while expecting differentiating outcomes from the inarticulate consumer? Shift your thinking, go beyond hedonics (Beckley, Moskowitz and Paredes 2008) and capture the full consumers’ minds to develop that Holy Grail, the consumer loved product (Roberts 2005, Pawle and Cooper 2006). Product developers need to realise that today’s hyper-competitive market is based more on emotional buying (Danziger 2004) than the old model of only rational decision making.

Today’s consumers are more knowledgeable and aware of the various marketplace options available to them. This situation creates a hyper-competitive environment that is more driven by consumer emotions, lifestyles, consumer-product experiences, and expectations than the old economic model of rational decision making. Emotional effects on buying decisions are no longer just branding and marketing issues. Hence, a holistic approach (hidden and conscious) is not only needed but required in developing the best products possible, i.e. the role of product development.

The following is a quick review of various approaches to (a) circumvent the language/consciousness issue, (b) measure emotions, (c) profile the consumer’s mind and (d) its utility for research guidance. This is meant to be an introduction rather than an exhaustive review of the literature which is growing rapidly (see Marcus et al. 2003).

Measuring Emotions

Non-verbal

Non-verbal approaches acknowledge the language issue of the inarticulate consumer. The commercial Product Emotion Measurement Instrument (PrEmo) (Desmet and Schifferstein 2008) is more direct in that it measures a specific set of emotions. It uses animated ‘puppets’ that are displayed on an electronic device (e.g. PC, laptop, etc). For a stand alone study, 14 emotions are displayed. For on-line evaluations, 12 emotions are displayed. Another commercial method is AdSam™ which uses an indirect approach (language and subconscious) that it is based on a theoretical model and is database dependent. There are only three scales of ‘manikins’ based on Pleasure, Arousal and Dominance. For each of these two methods, (a) hedonics is not measured, (b) measures of emotions are scaled and (c) the mechanism for interpreting consumer reactions is Multiple Correspondence Analysis (MCA).

Score Card:

• circumventing language issue: high
• measuring emotions: high
• profiling the consumer’s mind: low
• utility for research guidance: low

Verbal

Recently, Kling and Meiselman (2009) have developed an inventory of 39 emotions for evaluating food. Hence, 39 scales are used in obtaining an emotional profile of each test product. The Geneva Emotion Odor Scale (Chrea et al. 2009), with 6 dimensions, developed 36 ‘everyday odor’ emotions. This was later reduced to six 3-term items (Porcherot et al. 2010) with fragranced and flavored products. The Geneva Emotion Odor Scale is based on scaled data.

Score Card:

• circumventing language issue: low
• measuring emotions: high
• profiling the consumer’s mind: low
• utility for research guidance: high

Autonomic Measurements

To really circumvent the issues of the inarticulate consumer and non-differentiating hedonics, there are various non-verbal physiological approaches. These approaches typically include devices that detect stimulation and/or relaxation in the central nervous system. They include galvanic skin response (GSR), electroencephalograph (EEG), brain waves, neuro-feedback, eye movement/dilation and functional MRI (fMRI). Most of these approaches are not specifically designed for the food technologist in product development and have their own translation issues. Moreover, it is not clear how well these approaches would adapt to the iterative product development process. Issues surrounding the language issue of the inarticulate consumer are obviously circumvented. Capturing the consumer’s mind is not addressed.

Score Card:

• circumventing language issue: high
• measuring emotions: varies
• profiling the consumer’s mind: low
• utility for research guidance: low

Experimental Design

Rather than probing the consumer for specific reactions to test products, the product developer can rely on the experimental design of the study for detailed guidance. Design of Experiments (DOE) are implemented by systematically varying the physio-chemical (e.g. ingredients) properties of the product. All that needs to be measured is overall liking. Analyses of the DOE design become the mechanism of interpretation. Hence, issues surrounding the naïve consumer language are circumvented. Although capturing the consumer’s mind is not addressed, research guidance is high. The DOE approach, however, is used more for product optimisation than profiling and exploration.

Score Card:

• circumventing language issue: high
• measuring emotions: low
• profiling the consumer’s mind: low
• utility for research guidance: high

Sensory Descriptive Analysis and Data Integration

Many food companies have established descriptive panels (see Stone 2010) to assist the product developer. Descriptive analysis (e.g. Flavor Profile®, Texture Profile®, Spectrum Analysis®, and QDA®) is used in its own right and to supplement the consumer data to the same test products (Stone and Sidel 2004). The latter approach is a long standing acknowledgement that consumer information by itself is insufficient for product development guidance and more detailed perceptual information is needed for the product developer to make a better product.

In actuality, descriptive analysis assists in the translation of consumer reactions to more technical language. Multivariate analysis is needed to integrate the consumer and sensory data sets. Popular approaches are Partial Least Square Analysis (PLS), Multiple Factor Analysis (MFA) and External Preference Mapping. But what do you do when the hedonic measure is flat across the test products?

Score Card:

• circumventing language issue: potentially high
• measuring emotions: has potential
• profiling the consumer’s mind: has potential
• utility for research guidance: high

Sensory Projective Techniques

Sorting, Napping and Free Choice Profiling (Abdi 2007) are qualitative/quantitative sensory techniques that can be used with naïve consumers (see Lawless, Sheng, and Knoops 1995, Dreyfuss, Danilo and Garrel 2009). There is no pre-determined questionnaire and little training is needed. Hence, they can be considered ‘free association’ approaches that can be quantified with various multivariate analyses, e.g. Multiple Factor Analysis, Multidimensional Scaling, Generalised Procrustes Analysis, etc. Their role in the product development process has potential.

Score Card:

• circumventing language issue: medium
• measuring emotions: high
• profiling the consumer’s mind: high
• utility for research guidance: TBD

The approaches reviewed attempt to measure emotions of the naïve consumers. To be considered, however, is how these approaches can best be applied to the iterative process of product development, e.g. exploration/learning, timing and expense. The sensory-consumer approach is specifically designed for product development and can be expanded upon by a different mechanism of translation.

A Holistic Integrative Consumer-Sensory Approach

An integrative approach (see Ingersoll 1996) is to:

• focus on the consumer questionnaire
• use Descriptive Analysis
• use Multiple Correspondence Analysis (MCA) as the mechanism of translation.

The questionnaire is basically your essential instrument in communicating with the consumer. And, as a communication device, it is the main source of feedback and guidance from the consumer. MCA is a non-parametric (i.e. categorical data) multivariate technique analogous to Principle Component Analysis (Greenacre and Blasius 2006) but in the more geometric sense. MCA is used to reveal underlying patterns in a set consisting of categorical data. It is based on Χ2 distances, e.g. associations between attributes and products are uncovered.

The questionnaire must be holistic in capturing as much information as possible. The areas covered need to be:

• overall measures
• diagnostics
• product characteristics
• end benefits/added value
• emotions/imagery.

All sections use scales except for emotions and imagery which use Check-All-That-Apply (CATA) (Cowden, Moore and Vanluer 2009). With the use of CATA, we get quick consumer emotional reactions. CATA is categorical data which is appropriate for MCA, but not with the traditional methods of Principle Component Analysis, Factor Analysis, Preference Mapping, etc.

With this holistic consumer questionnaire approach, a mechanism of translation, and sensory descriptive data, we can provide actionable research guidance to the product developer. Moreover, we can go beyond hedonics and provide a story and/or a personality profile of each test product, e.g. benchmarks, market products and prototypes.

An Example

A potato chip study is used to demonstrate the utility of the integrative consumer-sensory approach. There were ten test products, five of which are marketed products, four are prototypes and there is one current product. Because there were too many samples to evaluate in a single test session, two test sessions were implemented over two days. A Balanced Incomplete Block design (Cochran and Cox 1957) was implemented such that each respondent evaluated only five test samples in a given test session. Since all respondents (N=100 women adults) evaluated all products, the resulting study was a Complete Block Design. As a side note, a similar study was performed among men.

Figure 2: A perceptual map of the consumer data. The number in front of each test product is the percent top 3 box of the hedonic scale. Consumer attributes are in black. Attitudinal and emotive measures are in blue. A specific standard set of emotions are in red.

Figure 2 shows an MCA perceptual map of only the consumer data. Basically axes need not be displayed, but are included for visual orientation and can be informally considered north, east, south and west. In reading the perceptual map, the product developer initially reviews the consumer acceptance of each test product and the ‘neighborhood’ of each test product. The personality profile of each test product is then considered.

There are two products with similar hedonic scores but different personality profiles, i.e. Current Product and Prototype 1. The Current Product (north-east on map) scored a 53% top 3 box on overall liking and is in a neighborhood that can be considered traditional, familiar, same old thing and a simple flavor. The product developer now has an idea as to why the hedonic score is relatively low relative to other market products. In contrast, Prototype 1 (north-center) has a hedonic rating of 55% top 3 box but a different personality profile – happy, comforting, an everyday snack, but mouth drying and salty.

This finding is clear demonstration of two test products with non-differentiating hedonics but different profiles. A similar situation is found for Competitive Products A, B and E; Competitive Products C and D; and Prototypes 2, 3 and 4. Hence, with an appropriate questionnaire and the use of MCA, we are able to differentiate products/prototypes with similar hedonic scores.

Figure 3 shows that the MCA perceptual map has a structure which indicates that the consumer’s mind is organised. Revealing the ‘structure’ of the consumer’s mind, however, is only possible with an appropriate questionnaire and sufficient test products. This map does show two desirable, and quite different, neighbors (i.e. the west side, top and bottom) to target for success. Once a target is identified (e.g. marketing and/or business unit), the product developer can initiate systematic prototype development, but needs more technical guidance.

Figure 3: A perceptual map of only the consumer data. The number in front of each test product represents the percent top 3 box of the hedonic scale.

For demonstration purposes, Figure 4 shows the MCA perceptual map with sensory ratings integrated as supplemental data. Although these sensory attributes are basic, it should be readily intuitive how the product developer can be guided by a more complex set of sensory descriptors.

Figure 4: A perceptual consumer data with sensory ratings integrated as supplemental data. The number in front of each test product is the percent top 3 box of the hedonic scale. Consumer attributes are in black. Attitudinal and emotive measures are in blue. A specific standard set of emotions are in red. Supplemental sensory data are in white.

The neighborhood of Competitor Product A was identified as the target by Marketing and/or Business Unit. To initiate product development, Prototypes 1and 4 (quite different from each other) were considered as appropriate starting points. Two additional studies were performed with four prototypes and Competitive Product A. The latter was not used not so much as a benchmark than as a statistical ‘anchor.’ Figure 5 shows the results of these studies. It needs to be noted that products in the same neighborhood need not be similar in their ingredients, taste and texture. In actuality, products in the same neighborhood can be quite different from each other. This is because the MCA perceptual map is based more on moods, attitudes, emotions and imagery than product sensory attributes. Hence, it is less probable that this approach does not facilitate the development of ‘me too’ products.

Figure 5: Shows the progression of two subsequent studies by initially combining aspects of Prototypes 1 and 4.

Overall Summary

There are many challenges in using the inarticulate consumer in the product development process. More and more hedonic scores fail to differentiate test products. We need to go beyond hedonics. There is a plethora of approaches that attempt to measure emotions either directly or indirectly, verbal or non-verbal. Moreover, there are many statistical methods for translating the emotive consumer reactions into meaningful research guidance. Many of these statistical approaches integrate sensory and consumer information.

The Holistic Integrative Consumer-Sensory Approach was herein shown to not only translate consumer reactions to test products, but also able to capture subconscious reactions to test products, e.g. attitudes and emotions. The three critical aspects of this approach are (1) a relevant questionnaire as the communication device, (2) integration of sensory data, and (3) MCA as the mechanism of translation. With this holistic approach, we reveal the hidden mind of the consumer and provide actionable research guidance to the product developer.

References

Abdi, H (2007) Some new and easy ways to describe, compare and evaluate products and assessors. SPISE, Ho Chi Minh City, Vietnam, 26-27 July. http://www.utdallas.edu/~herve/HerveAbdi-SPISE07-DISTATIS.pdf.

Akao, Y (1990) Quality function deployment: integrating customer requirements into product design. Boston, MA: Productivity Press.

Beckley, J, Moskowitz, H.R. and Paredes, D. (2008). Beyond hedonics: Looking at purchase intent and emotions as criterion variables. Available at: http://ssrn.com/abstract=1439733.

Chrea, C, Greanjean, D, Delplanque, S, Cayeux, I, Le Calvé, B, Aymard, L, Velazco, MI, Sander, D, and Sherer, KR (2009) Mapping the semantic space for the subjective experience of emotional responses to odours. Chem. Senses, 34: 49-62.

Cochran, WG and Cox, GM (1957) Experimental Designs, 2nd edn. Wiley Classics Library Edition published 1992.

Cowden, J, Moore, K and Vanluer, K (2009) Application of check-all-that apply response to identify and optimize attributes important to consumer’s ideal product. 8th Pangborn Symposium, 26-30 July, Florence, Italy.

Danziger, PN (2004) Why people buy things they don’t need: Understanding and predicting consumer behavior. Chicago, IL, USA: Dearborn Trade Publishing.

Desmet, PMA and Schifferstein, HNJ (2008) Sources of positive and negative emotions in food experience. Appetite 50: 290-301.

Dreyfuss, L, Danilo, S and Garrel, C (2009) Fragrance innovation: Understanding consumer perception of fragrances. ESOMAR World Fragrance Conference, Cannes, France, 22-24 June.

Greenacre, M and Blasius, J (eds) (2006) Multiple Correspondence Analysis and Related Methods. Boca Raton, FL: Chapman and Hall/CRC.

Ingersoll, DW (1996) Perceptual maps of sensory analysis and consumer reactions. ESOMAR World Fragrance Conference, Amsterdam, The Netherlands.

Ingersoll, DW (2010) Affective testing: The consumer as your partner in the product development process. World of Food Science Vol.9. Available at: www.worldfoodscience.org

King, SC and Meiselman, H. (2010) Development of a method to measure consumer emotions associated with foods. Food Qual. Pref. 21: 168-177.

Lawless, HT, Sheng, N and Knoops SSCP (1995) Multidimensional scaling of sorting data applied to cheese perception. Food Qual. Pref. 6(2): 91-98.

Marcus, GE, MacKuen, M, Wolak, J and Keele, L (2003) The measure and mismeasure of emotion. Paper presented at the Shambaugh Conference on Affect and Cognition in Political Action, University of Iowa, 6-9 March.

Pawle, J and Cooper, P (2006) Measuring emotion – Lovemarks, the future beyond brands. J. Advertis. Res. 1 March.

Porcherot, C, Delplangue, S, Raviot-Derrien, S, Calvé, DL, Chrea, C, Gaudreau, N and Cayeux, I (2010) How do you feel when you smell this? Optimization of a verbal measurement of odor-elicited emotions. Food Qual. Pref. (In Press). doi:10.1016/j.foodqual.2010.03.012.

Roberts, K (2005) Lovemarks: The future beyond brands. New York, NY, USA: PowerHouse Books.

Stone, H and Sidel, JL (2004) Sensory Evaluation Practices, 3rd edn. London/New York: Academic Press/Elsevier.

Dr David Ingersoll is a Senior Research Analyst (e-mail: davidI@whoisq.com) and Maryana Kaplan is Chief Research Officer at Q Research Solutions, Inc., Old Bridge, NJ, USA (e-mail: maryanak@whoisq.com)