At Keio University, Professor Takuya Satomura, in the Faculty of Business and Commerce, is researching scientific methods for evaluating consumer behavior, marketing strategy, marketing mix, and customer strategy.
"Recently, our group has been focusing especially on utilizing visual data in marketing research. Today, I'd like to talk about some of the research methods we've developed."
Nowadays, a great many leading brands and copycat brands are seen in and outside Japan. When two brands are visually similar, consumers are likely to get the brands confused, and lawsuits may arise as a result. So, there's a strong demand for methods that objectively assess whether visual similarity makes consumers confuse brands.
Accordingly, Professor Satomura has developed methods to measure, assess, and predict how much consumers will confuse brands that have visually similar product packages. This has been achieved by combining experimental methods with image processing technology.
"In typical research until now, measurement has been done by presenting consumers with copycat brands and the original brand. With that method of measurement, consumers have a comparatively long time in which to judge the brands. But when we think about how consumers actually behave, especially in shops, consumers make judgements in a shorter time, taking just a few seconds."
Professor Satomura uses a method called the triangle test, where people are shown a package, plus two similar ones. Consumers are asked to choose the one that's different. To replicate conditions in an actual store, the packages are shown for 0.3 seconds. Consumers say which goods they think are different in that very short time, and the time they take to respond is measured.
"When something is shown for an instant to get a response, if people are confident, they can respond right away, but if people aren't confident, they can't make up their minds. It's known that, in a test that measures reaction times like that, the faster people respond, the more accurate their judgement is."
The analysis is done by combining data obtained from such consumer tests with data on how similar the package colors and shapes are, which is obtained using image processing technology.
"We combine two types of information: the degree of confusion obtained from consumer tests and the degree to which images are similar. To do that, we utilize reaction models used in mathematical psychology, called competing accumulator models. The idea was to identify what kind of information makes consumers confuse packages."
To assess the degree of confusion using prediction results, a three-tiered metric can be used, namely, "copy alert", "copy watch", and "copy safe".
"Until now, this type of analysis method based on human perception mechanisms hasn't been applied to confusion regarding packages. But by bringing in this kind of metric, we've made it possible to do analysis that's more firmly based on theoretical principles."
Using this method will enable businesses to not only evaluate confusion of brands after a product is released, but also to predict whether consumers will be confused about packages, even at the product development stage.