Editor’s note: Philip Moore is market research manager, CarMax, the Auto Superstore, Glen Allen, Va.
I hope one of three things will happen as a result of this article:
1. Someone with better knowledge of the current state of vendor offerings will direct me to the right place, or
2. Someone smarter than me will tell me why this approach is methodologically unsound and should not be pursued, or
3. Someone will recognize the power of fully leveraging computer-administered discrete choice and build the tool.
Discrete choice is a terrific tool for isolating brand equity. As far as I am concerned, that issue is resolved. Consider, however, a retailer like Sears trying to measure its brand equity in a discrete choice study with consumers of appliances. Now you have Whirlpool, Maytag, Kenmore , Amana, Sunbeam, Frigidaire and many other brands complicating my ability to isolate the contribution of the Sears brand to the choice decision. If you fail to specify the brand of appliance in the design, i.e., the discrete choice card just says “a washing machine,” then a respondent loyal to the Kenmore brand will likely have higher utility for Sears than a fan of Maytag. If you do specify Kenmore in the design, then the Maytag fan will probably generate lower utility scores for Sears, even if they would love to buy a Maytag from Sears.
The interaction of retailer brand and product brand gets even more complicated if brands within the product category have staked out different price space. Consider automobile brands. A respondent whose relevant range of products is Geo, Kia and Hyundai will balk at the price points on discrete choice cards constructed with BMW or Porsche. Their negative utility scores for ABC Automotive would have nothing to do with the retailer and everything to do with the research design.
What I need is a tool that will leverage the power of the computer to construct and administer a discrete choice card set, using table queries based on responses to initial screening questions. The tool would allow the respondent to make choices based on relevant product brands and price points, with the retailer’s brand equity expressed in percentage of price paid.
In case this tool does not exist or someone wants a more concrete example to critique, here is exactly what I have in mind. My discrete choice has three attributes: vehicle class, retailer and price. The levels for the vehicle class attribute are: new, certified used, non-certified used. The levels for the other two attributes are all constructed based on respondent input.
Q1. What is your five-digit zip code?
Q2. What type of vehicle are you most likely to purchase the next time you buy? (small sedan, large sedan, SUV, minivan, etc.)
Q3. What make of vehicle are you most likely to purchase the next time you buy? (all the brands that offer the specified type in Q2: Acura, Audi, etc.)
A query against my dealer database will give me the closest franchised dealer for any brand and my biggest competitor for market share in the respondent’s zip code. This gives me my retailer levels.
A query against my make/model database will give me the average sales price for any make/type combination. For example, the respondent tells me they are likely to buy a large sedan (Q2) made by Buick (Q3) so my query goes and finds the average new selling price for a Buick LeSabre (P). My price points are simply 0.75(P), 0.85(P), P, and 1.15(P).
Now I have the following attributes and levels for the discrete choice:
Vehicle class
New
Certified used
Non-certified used
Retailer
Closest franchised dealer of specified brand to respondent’s zip
Largest market share competitor in respondent’s zip
Client retailer (me)
Price
15 percent above average price of specified make/model
Average price of specified make/model
85 percent of the average price of specified make/model
75 percent of the average price of specified make/model
Now I interpret the utility score for client retailer as the independent contribution of my brand to the choice decision. At least that is what I will do once I find the application that makes this possible. I look forward to hearing from anyone who has assistance to offer!