Explore the vast range of titles available.

An accurate prediction of what a customer will purchase next is of paramount importance to successful online retailing. In practice, customer purchase history data is readily available to make such predictions, sometimes complemented with customer characteristics. Given the large product assortments maintained by online retailers, scalability of the prediction method is just as important as its accuracy. We study two classes of models that use such data to predict what a customer will buy next, i.e., a novel approach that uses latent Dirichlet allocation (LDA), and mixtures of Dirichlet-Multinomials (MDM). A key benefit of a model-based approach is the potential to accommodate observed customer heterogeneity through the inclusion of predictor variables. We show that LDA can be extended in this direction while retaining its scalability. We apply the models to purchase data from an online retailer and contrast their predictive performance with that of a collaborative filter and a discrete choice model. Both LDA and MDM outperform the other methods. Moreover, LDA attains performance similar to that of MDM while being far more scalable, rendering it a promising approach to purchase prediction in large product assortments. Data, as supplemental material, are available at http://dx.doi.org/10.1287/mksc.2016.0985 .

Consumers are widely adopting Artificially Intelligent Voice Assistants (AIVAs). AIVAs now handle many different everyday tasks and are also increasingly assisting consumers with purchasing decisions, making AIVAs a rich topic for marketing researchers. We develop a series of propositions regarding how consumer decision-making processes may change when moved from traditional online purchase environments to AI-powered voicebased dialogs, in the hopes of encouraging further academic thinking and research in this rapidly developing, high impact area of consumer-firm interaction. We also provide suggestions for marketing managers and policymakers on points to pay attention to when they respond to the proposed effects of AIVAs on consumer decisions.

Many consumer decisions are assisted by product recommendations. When retailers provide such recommendations, there is an inherent tension between (1) presenting a set of products that are close in attractiveness (fine product set granularity) and (2) presenting a wider range of products that are more different in attractiveness (coarse product set granularity). While the former can maximize the attractiveness of the recommended set of products, the latter makes it easier for consumers to determine which of the recommended products is most attractive, thus boosting consumer response. Evidence from a large-scale field study (with naturally occurring variation in the granularity of online recommendation sets) provides strong support for this tension and shows that less fine-grained product recommendation sets promote consumer response. We also find that, in line with our theorizing, coarser set granularity increases the time consumers spend processing detailed information about individual products relative to time they spend comparing products at the set level. These effects are less pronounced when consumer engagement in the decision process is low. The key insights from the field study are replicated in a tightly controlled experiment (using a different product domain). The findings of this research have important implications for how best to integrate large online assortments and product recommendations to stimulate consumer response.

Digital technology in financial services is helping consumers gain wider access to investment funds, acquire these funds at lower costs, and customize their own investments. However, direct digital access also creates new challenges because consumers may make suboptimal investment decisions. We address the challenge that consumers often face complex investment decisions involving multiple funds. Normative optimal asset allocation theory prescribes that investors should simultaneously optimize risk–returns over their entire portfolio. We propose two behavioral effects (mental separation and correlation neglect) that prevent consumers from doing so and a new choice architecture of virtually integrating investment funds that can help overcome these effects. Results from three experiments, using general population samples, provide support for the predicted behavioral effects and the beneficial impact of virtual integration. We find that consumers’ behavioral biases are not overcome by financial literacy, which further underlines the marketing relevance of this research.

This research examines the impact of defaults on product choice in sequential-decision settings. Whereas prior research has shown that a default can affect what consumers purchase by promoting choice of the preselected option, the influence of defaults is more nuanced when consumers make a series of related choices. In such a setting, consumer preferences may evolve across choices due to \"spillover\" effects from one choice to subsequent choices. The authors hypothesize that defaults systematically attenuate choice spillover effects because accepting a default is a more passive process than either choosing a nondefault option in the presence of a default or making a choice in the absence of a default. Three experiments and a field study provide compelling evidence for such default-induced changes in choice spillover effects. The findings show that firms' setting of high-price defaults with the aim of influencing consumers to choose more expensive products can backfire through the attenuation of spillover. In addition to advancing the understanding of the interplay between defaults and preference dynamics, insights from this research have important practical implications for firms applying defaults in sequential choices.

Many firms rely on choice experiment-based models to evaluate future marketing actions under various market conditions. This research investigates choice complexity (i.e., number of alternatives, number of attributes, and utility similarity between the most attractive alternatives) and individual differences in decision time as key factors that affect the predictive performance of models based on choice experiments, both within and between complexity conditions. The results show that complexity and individual decision time not only affect the error in consumer choice models but also consumers' decision strategy and systematic utilities. The authors introduce a complexity-adjusted mixed logit (CAM logit) model to capture the various influences of complexity in choice experiment-based models. They illustrate the consequences of complexity on choice behavior with market share predictions of the CAM logit model for different complexity conditions.

Background and Objective Non-participation in colorectal cancer (CRC) screening needs to be decreased to achieve its full potential as a public health strategy. To facilitate successful implementation of CRC screening towards unscreened individuals, this study aimed to quantify the impact of screening and individual characteristics on non-participation in CRC screening. Methods An online discrete choice experiment partly based on qualitative research was used among 406 representatives of the Dutch general population aged 55–75 years. In the discrete choice experiment, respondents were offered a series of choices between CRC screening scenarios that differed on five characteristics: effectiveness of the faecal immunochemical screening test, risk of a false-negative outcome, test frequency, waiting time for faecal immunochemical screening test results and waiting time for a colonoscopy follow-up test. The discrete choice experiment data were analysed in a systematic manner using random-utility-maximisation choice processes with scale and/or preference heterogeneity (based on 15 individual characteristics) and/or random intercepts. Results Screening characteristics proved to influence non-participation in CRC screening (21.7–28.0% non-participation rate), but an individual’s characteristics had an even higher impact on CRC screening non-participation (8.4–75.5% non-participation rate); particularly the individual’s attitude towards CRC screening followed by whether the individual had participated in a cancer screening programme before, the decision style of the individual and the educational level of the individual. Our findings provided a high degree of confidence in the internal–external validity. Conclusions This study showed that although screening characteristics proved to influence non-participation in CRC screening, a respondent’s characteristics had a much higher impact on CRC screening non-participation. Policy makers and physicians can use our study insights to improve and tailor their communication plans regarding (CRC) screening for unscreened individuals.

Acquisition channels are important predictors of customer loyalty in the first stages of a business–consumer relationship. Although some researchers have provided examples of the differences in the value of the customers businesses acquire via different channels, they have not considered the impact of acquisition channels on loyalty and cross-buying. Using probit-models we explored how retention rates and cross-selling opportunities differ among the various acquisition channels a financial-services provider uses. Our results indicate that the direct-mail acquisition channel performs poorly on retention and cross-selling, while radio and TV perform poorly for retention only. The firm's Web site seems to perform well for retention. The theoretical and practical implications of our results are discussed.

Background: Discrete choice experiments (DCEs) are rarely analyzed with choice models that include a full set of two-way interactions between the attribute levels: the resulting model would be (too) difficult to interpret and the sample size requirements (far) beyond what is feasible in applied research. Therefore, an alternative modelling approach is introduced that allows for interactions between the attributes-as opposed to interactions between the attributes levels. Methods: DCEs often comprise at least a subset of attributes for which monotonically increasing or decreasing preferences can be presumed, e.g. costs, benefits, risks, etc. Without imposing linear preferences, an optimal scaling approach can be used to transform the levels of these attributes onto continuous latent scales, which can be interacted with each other and with the levels of categorical attributes. This results in a very parsimonious model specification. Results: The proposed model with and without interactions is fitted on an existing dataset of N = 3699 respondents who each completed 16 EQ-5D-3L discrete choice tasks. As shown, the interactions between the attributes are straight-forward to interpret and their inclusion greatly improves statistical (WAIC) model fit statistics, while requiring 97% fewer parameters compared to a standard MIXL model with a full set of 2-way interactions between the included levels. Conclusions: The proposed interaction model is parsimonious, produces estimates that are straight-forward to interpret, and accommodates the estimation of interactions in DCEs with more attractive and feasible sample size requirements. The model has one major disadvantage: it is not straight-forward to transform preferences for attributes with categorical levels onto a continuous latent scale.