Identifying Attributes and Attribute Levels

The experimental design for a choice-based conjoint starts with the identification of the attribute levels to be included in the study.

The attributes and attribute levels used in a choice-based conjoint study should be determined by the objectives of the study. If the only objective is to understand the brand price relationship, you should only include brand and price attributes. If you want to understand the drivers of sales, you need to include all the attributes that influence choice behavior.

Types of attributes

There are three basic types of attributes: product features, benefits, and images, as defined in the table below. Price, for example, can be represented as a product feature (i.e., the actual price), a benefit (inexpensive), or an image (cheap). “0 to 100kph in 6 seconds” is a product feature, “fast acceleration” is a benefit, and “for rev-heads” is an image.

It is almost always possible to map out the relationship between features and benefits. The relationship of images to features and benefits is more problematic. In the late sixties and early seventies, the focus of research into how people choose focused on understanding the relationship between benefits and choice. More recent efforts have focused almost exclusively on using product features as attributes. The issue of which type of attribute to use depends:

• Consumer processing.
• Anticipate implementation.
• Whether the research is a tactical or strategic orientation.
• The number of attributes.
• Ease of measurement

Consumer processing

Consumers generally think in terms of benefits and images. Tasks structured using benefits and images may be easier for the consumer. The flip side is that benefit and image statements can be ambiguous. Consider the women's hair shampoo benefit decompositions shown below (adapted from Myers, James H. and Allan D. Shocker (1981), “The Nature of Product-Related Attributes”, Research in Marketing, 5, 211-236, p. 230; the original diagram referred to Farrah Fawcet instead of Jennifer Lawrence.) While consumers should not have too much difficulty in understanding a product feature attribute called "fragrance", they may struggle to process “leaves a deposit” in a meaningful way.

Anticipated implementation

Firms think in terms of product features. While it is easy to say that research should be consumer-focused, there's no point in doing this if the research cannot be implemented. Consider the general-purpose household cleaning example shown above. While the benefits and imagery may be in consumer language, they are too ambiguous to provide clear guidelines for product development. Product features are likely to be much more useful attributes for engineers and chemists charged with the responsibility of building the new product.

Whether the research has a strategic or tactical application

The ease of implementing the recommendations of research conducted using a conjoint where the attributes were defined as features means that except in exceptional circumstances all tactical studies should be conducted using product features as attributes. By contrast, when a study is to be used for long-term strategic planning – such as when conducting segmentation research – a focus on benefits is likely to remain relevant for a much longer time.

Consider the value map of the wine cooler market shown below. (Reynolds, Thomas J. and Jonathan Gutman (1988), “Laddering Theory, Method, Analysis, and Interpretation,” Journal of Advertising Research, 26 (February/March), p. 19). The decline of the cooler market since the map was drawn in 1988 means that a study of choice behavior using the product features shown would be of interest to marketing and alcohol historians only. However, a study that understood how consumers evaluated the different cooler benefits would still be of value today, providing insight into consumption behavior in markets that have replaced the wine cooler market, such as the alcoholic soft drink market.

Of even greater potential long-term strategic worth would be a study that understood the relationship between values and choice, although such a study will struggle with measurement difficulties.

The number of attributes

Sometimes many features will relate to one benefit, while in other situations multiple benefits will arise from one feature. In the diagram above there are seven features and twelve benefits. Similarly, features such as leather seats, inset lights, and paint finish all relate to whether an automobile is “stylish”. Conversely, the feature Japanese is related to benefits such as fuel economy, styling, and reliability. Assuming all else is equal, in general, the best thing to do is choose the type of attribute which results in the smallest number of attributes.

Ease of measurement

Research designs for measuring consumers’ attachments to different product features are very sophisticated, making valid measurement possible. The valid measurement of the benefits that consumers seek is somewhat harder but possible. However, at the time of writing, no rigorous methodologies have been developed for understanding the link between image and brand choice. The consequence of this is that in situations where it is believed that image is the only type of attribute that is appropriate, it is questionable as to whether there is currently any value in using choice-based conjoint methods.

Techniques for identifying relevant attributes

Typically, attributes should be identified in two stages:

1. A brainstorming session with key stakeholders. Such a brainstorming session will be most effective if all of those involved independently come up with lists of attributes, these lists are combined, and then the group collectively goes through the process of culling the list (see Rossiter, John R. and Gary L. Lilien (1994), “New ’Brainstorming’ Principles,” Australian Journal of Management, 19 (June), 61-72.)
2. By research, typically qualitative research, with people in the market.

The brainstorming should always occur first, allowing the research stage to concentrate on testing and improving on what has been identified in the brainstorming.

Research techniques for identifying attributes and attribute levels include:

• Identifying factors relating to purchase/non-purchase. One recommendation is to conduct a focus group and ask (Louviere, Jordan J. (1988), Analyzing Decision Making: Metric Conjoint Analysis, SAGE Publications Inc, p. 51.)
  • Which products (services) in this product class do you buy (own), or would you consider buying (owning)?
  • Which products (services) in this product class do you not buy (own), or would you not consider buying (owning)?
  • You said that you would buy (own) or would consider buying (owning) brand(s) (products that in Question 1 are now repeated). What is it about these products that make them attractive to you?
  • You said that you do not buy (own) brand(s) (products in Question 2 are repeated) or would not consider buying (owning) them. What is it about these products which are unattractive to you?
  • Thinking about the products (services) that you buy (own) or would consider buying (owning), what would the producers or sellers of these products (services) have to do to them to influence you not to buy (own) them?
  • Thinking about the products (services) that you said you do not buy (own) or would not consider buying (owning), what would the producers or sellers of these products (services) have to do to them to influence you to buy (own) them?
• Observing purchasing patterns. How do consumers actually purchase products? Marketing gossip has it that 60% of grocery purchase decisions are made in the store, and only take a few seconds. If this is true, it is extremely unlikely that the factors influencing purchase are utilitarian benefits. They're much more likely to relate to point-of-sale factors and price.
• Observing how consumers use products. In the early 1970s, Pepsi gave 350 families the opportunity to order home-delivered Pepsi and competitive soft drinks at discount prices. No matter how many bottles were ordered, the consumers always drank them, leading Pepsi to conclude that the volume of soft drink consumption was in part driven by consumers' ability to get products home. Pepsi tapped into this attribute by developing the plastic bottle, completely re-shaping the soft drink market by enabling consumers to buy more products (MacMillan, Ian and Rita Gunther McGrath (1996), “Discover Your Products’ Hidden Potential,” Harvard Business Review, 74 (May-June), 58-73.).
• Identifying consumers' perceptions of risk. Midas has built a successful brake repair business by reducing consumers' fears of being exploited by leading them through a checklist of all that can be wrong and is wrong (MacMillan, Ian and Rita Gunther McGrath (1996), “Discover Your Products’ Hidden Potential,” Harvard Business Review, 74 (May-June), 58-73.).
• Using Kelly's personal construct theory. Consumers are presented with sets of three existing products and asked to identify the attributes on which the products are similar and different. Content analysis – commonly researcher judgment – is used to identify the underlying attributes. This approach is generally only useful when researching product categories that little is known about, or when attempting to elicit image-based attributes.
• Laddering. Laddering methodologies, such as means-end chains, can be a useful means of identifying relevant attributes. The figure above showing values, benefits, and features in the wine cooler market is an example.
• Consulting the academic, trade, technical and consumer literature. A good literature review can save substantial time and money. The coffee, automobile, and toothpaste markets, for example, have received much attention in the theoretical and methodological marketing literature. For consumer products, magazines such as Consumer Reports and online product comparison services like com often contain comparisons of products on what are believed to be the key attributes. Similarly, in industrial markets, technical reports and magazines often compare products based on key performance criteria.
• Include a label attribute if a choice in the market seems likely to be driven by difficult-to-describe attributes. For example, in the cola market, it is much better to use a brand as an attribute than to attempt to create a set of attributes that explain the differences between why people purchase Coke versus Pepsi. Similarly, if modeling transport, it is usually better to include an attribute of the mode (Bus, Train, Car), than attributes describing the fundamental technical differences between these attributes.