Determining the relative importance to wine consumers of sensory and non-sensory attributes on liking and choice: a cross-cultural study

Abstract

The objective of the project was to understand the relative impact of sensory and non-sensory (e.g., packaging, pricing) attributes on consumer wine choice, and to develop methods capable of measuring and predicting consumer reaction to changes in these variables. Four main methods were used in this project: sensory evaluation, chemical analysis, simulated choices of wines, and actual sales based on AC Nielsen data.

Summary

This project focused on developing, testing, and demonstrating new methods to understand how consumers make wine choices, and the relative impact of sensory and non-sensory factors. The methods developed are science-based and validated against actual sales in the market, rather than on consumer attitudes towards the taste or packaging of wines. The project demonstrates to the wine sector that it can potentially anticipate consumer responses to changes in the product and its marketing and by doing so can design products more likely to succeed in the competitive wine market. These new methods overcome the issue that consumers cannot introspect their own response to packaging and taste, and therefore their responses do not predict their actual behaviour.

The project was conducted in two phases, one in Australia and one in the United States. The first phase in Australia focused on Shiraz wines and used three different experiments: an online choice experiment using 21 real wines, an online choice experiment using simulated wine bottles varying a wide range of packaging and prices, and a tasting of the 21 wines (each person tasted 5 of the 21) followed by an evaluation and purchase intention measure. Overall, we found that consumers’ choices online of real wine bottles did reflect the actual sales of those wines using AC Nielsen scanner data. Our experiments were good predictors of actual sales. However, after tasting the wines, consumers had higher liking for the more expensive wines, but did not actually choose these wines for repurchase. Their repurchase was linked more strongly to their online wine choices. Consumers could only taste 5 wines, so it is not practical to include actual tastings in future predictive research. We also found that it is impractical to manipulate packaging variables, because there are too many different combinations to test reliably. We found some indication that wine chemistry was useful in predicting consumer purchases. The most important variables predicting choice were (in order): objective ratings of the wines using 1-5 stars, the brand name, the price, medals, price discounts, followed by alcohol level, region, label style, label colour, and finally closure.