Mapping the bounds of incoherence: how far can you go and how does it affect your brand?

Consumers often have to evaluate products comprising a combination of attributes that is not expected by them, given their beliefs about how attributes normally co-vary in the product category. Such an attribute combination implies that the claimed level of a product attribute is then different from what the consumer might infer, given the level of another attribute, resulting in what we call product incoherence. We develop a model to calibrate the effect of incoherence on perceptions, uncertainty, preference, and ultimately purchase. Our model can allow managers to determine consumers' acceptance for different positions in the multiattribute space, so they can optimize their product's positioning. Our model implies that a product that combines positively valued attributes might increase some elements of preference for the product, but if those attributes occur in unexpected combinations, incoherence will also increase uncertainty which in turn might lower other elements of preference. The net risk-adjusted preference for a product in our model accommodates both the benefit from the expected attribute levels and the uncertainty associated with incoherence. We derive implications from the model and provide an empirical test that supports those implications.

Key words: product positioning; product management; new products; brand management History: This paper was received September 28, 2005, and was with the authors 4 months for 2 revisions; processed by Duncan Simester.

1. Introduction

We examine the situation in which a consumer receives information about two attributes whose levels are expected to be correlated, but where the claimed level of one attribute is far from what the consumer would expect given the claimed level of the other. Consider the 2005 Honda Accord hybrid, which claims to be a "surprisingly fuel efficient 255 horsepower" car (Newsweek 2004). A car with high fuel efficiency and power is called surprising because the commonly observed negative correlation between these attributes makes this new combination unexpected. Specifically, the consumer must reconcile the car's claim of high fuel efficiency with an inference of low fuel efficiency, based on the belief of negative correlation between fuel efficiency and power (and conversely). We call this phenomenon product incoherence, where unlikely attribute combinations provide the consumer with discrepant attribute information about product performance. Our goal in this paper is to develop a methodology to calibrate the effect of incoherence and its antecedents on preference, allowing managers to make better decisions about how to optimize product positioning in practice.

While incoherence has been studied in the behavioral science literature in various forms--concept incoherence (Murphy and Medin 1985), schema incongruity (Meyers-Levy and Tybout 1989), and category incoherence (Rehder and Hastie 2004)--that literature has not produced a quantitative model of the effect of product incoherence or its antecedents on consumer preference. In the absence of such a model, it is difficult to calibrate the effect of unexpected attribute combinations on preference. For example, if powerful cars are thought to be fuel guzzlers, safe cars are believed to be boring, and cola sodas are known to be brown, then to what extent will the market accept a "powerful and fuel efficient" car (2005 Honda Accord hybrid), a "safe and stylish" car (Volvo V70, see http://volvocars-pr.com), or "a blue-colored cola" (Pepsi Blue, see http://www.spudart.org/pepsiblue/)? Our model can provide a manager with information about the effects of different positions in multiattribute space on market acceptance.

As a result, such a model can inform several important marketing problems, including (1) improving product design and selection of optimal attribute combinations, (2) choosing a credible position for a new product launch into an existing market structure, and (3) developing effective communication strategies that emphasize coherent attributes.

Our model integrates a utility-maximizing consumer's (i) beliefs of interattribute covariation, (ii) attribute perceptions, (iii) uncertainty about those perceptions, and (iv) risk aversion, leading to an expression for the risk-adjusted preference for the product. Our main result shows that while combining positively valued attributes may increase expected preference for the product even if those attributes occur in unexpected combinations, incoherence will increase uncertainty about the product's attributes because of the conflict with the consumer's interattribute covariation beliefs. If the utility loss due to increased uncertainty (adjusted by the consumer's risk aversion) is greater than the expected benefits offered, the consumer's net preference for a product suffers. Our model provides managers with the capability to understand the risks associated with unique and unexpected attribute combinations that might be offered in pursuit of differentiation.