Comparing Lab, Virtual, and Field Environments in Sensory Product Acceptance Testing: An Abstract

Sensory marketing research has recently attracted massive attention in theory and practice. Every year, companies spend billions of dollars on the sensory evaluation of food products, particularly in the contexts of new product launches and sensory product differentiation. Such experimental sensory test procedures are typically conducted in highly controlled sensory labs and involve a segmentation of acceptance ratings in order to disclose consumer groups that differ significantly regarding their preferences for various sensory experiences (Ernst et al., 2010; Moskowitz & Rabino, 1994). Such segmentation then constitutes the starting point for shaping market segmentation strategies (Armstrong & Kotler, 2017; Wendin et al., 2015). Despite all these efforts, food product marketing failure rates are approximately 50% (Ogawa & Piller, 2006; van der Panne et al., 2003), which calls into question the validity of the clustering solutions and sensory test carried out in lab environments (Carbonell et al., 2008).

Recent research has started to investigate whether virtual consumption environments allow mimicking expensive and time-consuming field acceptance tests in lab experiments (Bangcuyo et al., 2015; Kim et al., 2016). While these studies offer important insights for sensory marketing research, they do not allow for any conclusion as to whether results from virtual consumption environments correspond to those generated in real settings. Addressing this gap in research, our study is the first to compare the results of a consumer acceptance test conducted in (1) a sensory lab, (2) a virtual consumption environment, and (3) a field setting. Our analyses show that segments derived from answers in the virtual coffeehouse better correspond to the segments derived in the real coffeehouse than those derived in the sensory lab. This finding is robust with regard to varying segmentation algorithms, segment numbers, and preprocessing strategies for acceptance ratings. Considering that researchers and practitioners routinely rely on clustering of acceptance scores in an effort to disclose consumer groups that significantly differ in their preferences, our findings are compelling for improving sensory marketing research.