Asking consumers what they want is so passé.
There was a time when brand and marketing managers would not think of making a decision about the product, its features, packaging, and even pricing without asking the consumer first – mostly through surveys. Lengthy questionnaires were always being designed, tested, administered, analyzed, and interpreted. Management meetings were called to discuss survey results, and make decisions. No self-respecting brand manager would launch a product without commissioning a major consumer survey, and few VPs of marketing would permit a major 4P decision without “the data.”
But asking consumers is becoming a mug’s game. For one, consumers may not really know what they want, especially if the product or feature is new (did consumers really want multi-touch before they saw it?). And even if they do know, the way we formulate the question may have as big an impact on their responses as their inherent preferences. Furthermore, their preferences may be context dependent – they may say blue is their favorite color, but not, it turns out, for baked goods.
As early as the 1960s, market researchers recognized many of these problems. Conjoint analysis was developed as a “revealed preference” technique in which consumer choices between products with different bundles of attributes revealed how important each feature was in their choice. But walking consumers through dozens of choices is cumbersome. It is expensive to conduct conjoint analysis on large samples.
Twenty-first century market research does away with asking the consumer – it solves problems associated with asking consumers their preferences in two elegant ways.
First, in many product categories it has become cheaper to make the product and put it on the shelf to see if consumers will buy it, than to do the market research before launching the product. Zara produces experimental runs of a few hundred units of a specific dress, and ships it to select stores. If it sells well, they can produce a few thousand units and have them on the shelves of all of their stores within a few weeks. Given the speed and low cost of producing the actual product, there is no need to develop prototypes and survey consumers to see if they like the product.
Second, the virtual shopping environment offers enormous amounts of data about consumers, their media and shopping habits, as well as their actual purchases. As consumers make phone calls, surf the internet, and use their credit and loyalty cards, elfin bots are busy collecting and processing the digital trail of billions of crumbs and cookies that they leave behind. Telecom carriers analyze call frequency, duration and timing. Credit card companies assess which rewards offers are most likely to spur purchase by any given cardholder. Facebook captures data about connections between consumers to find hidden patterns that anticipate consumer behavior. Amazon knows which products each consumer browsed before they bought what they did, and so much more. These new developments reflect a revolution in the way companies collect and assemble consumer data.
In a virtual shopping environment, it is possible to vary product features, prices, product presentation, assortment and context, and measure actual consumer choice to determine preference.
As Tesco, the largest retailer in the United Kingdom, put it in one of its recent annual reports: “We have spent many years developing our customer insight …through careful analysis of sales and loyalty card data we can better understand what is important to our customers…we don’t have to guess what our customers want, we know.”[i]
[i] Tesco Annual Review 2009, 8.