Comparing results of CBC and ACBC with real product selection

We examined consumer preference for a computer accessory product line with an online survey. Every respondent completed both a choice-based conjoint (CBC) procedure and an adaptive choice-based conjoint (ACBC) procedure using the same attributes. Our goal was to predict market share and to answer both methodological and managerial questions. In evaluating within-subjects results, CBC and ACBC gave generally similar estimates, with ACBC estimating greater price sensitivity and giving a smaller standard deviation of respondent utilities. When estimated head-to-head preference between two products was compared to actual market data, ACBC agreed closely with actual market data. In a 4-product portfolio, ACBC predictions of preference share were in somewhat closer agreement to observed market data than those using CBC data; the ACBC error proportion was 15-25% lower than the error rate with CBC data.