An Algorithm for Designing Value Propositions in the IoT Space: Addressing the Challenges of Selecting the Initial Class in Reference Class Forecasting

Abstract

Developing products and entering new markets in the Internet of Things (IoT) and industry 4.0 space are challenging because value propositions and business models are complex. IoT products are fundamentally platform products, whose embedded components can measure a disparate set of characteristics for several user types. Deciding in advance which combination of characteristics and users to focus on is daunting, and further complicated wherever the user is not the payer. It requires prediction. A common predictive technique, such as reference class forecasting (RCF), often suffers from the biased selection of initial reference classes. In this article, we propose combining RCF with our new biclustering algorithm, ECrfBimax, built from 200 use cases we distilled. ECrfBimax uniquely makes biclustering predictive. Our combined biclustering/RCF approach helps managers and designers design value propositions for IoT devices through a data-driven selection of initial classes. Applying the algorithm to readily available qualitative data from a high-dimensional dataset comprising 10 competitors of a real-life IoT company, 200 customers, and 230 product characteristics, we identify solution characteristics deployable by that company in different sectors. ECrfBimax identifies how far competitors are targeting users using similar/dissimilar combinations of characteristics. Quantitative proxies for size, success, and performance based on funding raised empirically support and statistically validate the algorithm. Uniquely too, our bicluster hierarchically groups customers with characteristics which either should or should not be jointly developed to target them. We also uniquely compare RCF and biclustering as methods for analyzing objects’ similarity.