Michael Reinhold

Recommendations systems are coveted intelligence systems in e-business that provide benefits such as automated cross- and upselling. However, they were impractical to adopt in SME contexts characterized by scarce transaction data. Based on an SME case study (dayzzi.com), we demonstrate how SME's can overcome three essential barriers to adopt recommendation systems. Large scale simulation and modelling techniques is the key.

Recommendations systems are artificial intelligence systems that have found a wide range of applications notably in e-business contexts, providing benefits such as automated cross- and upselling in addition to a reduction of transactional costs on both the demand and the supply side. They have been widely accepted by large enterprises but SMEs trying to adopt and implement similar technologies must face the problem of scarcity of transactional data. As a case in point we present the obstacles that a recent SME venture ran into and needed to overcome in order to provide the customers with a satisfying sales experience We will touch upon both the marketing and machine learning aspects of the platform and discuss several techniques that have been key in the successful implementation of expert knowledge extracted from the experience of the company's founders.