Artificial Intelligence-Based Consumer Communication by Brick-and-Mortar Retailers in India Leading to Syllogistic Fallacy and Trap – Insights from an Experiment

Abstract

It is observed that a majority of organized brick-and-mortar (B&M) retailers in India believe
that they have adopted the latest Artificial Intelligence-based consumer communication
(AIBCC) tools/solutions and are yielding accurate outputs that can be used for interpretation,
conclusion, and decision-making concerning consumer communications. This
belief/assumption in itself is a classic example of a syllogistic trap. This study reveals that the
B&M retailers in India are least worried about AIBCC tools/solutions repeatedly sending
promotional/campaign messages to consumers based on their past transactional data till they
come back again to the store without knowing the ‘Purpose of Previous Purchase’. This is
mere because the cost of such communications is negligible (Just costs about 1 US dollar for
sending 500 messages to a mobile phone number). We have also observed that the B&Mretailers are unaware of the potential negative impacts of false/fake/artificialpromotional/campaign messages being sent to consumers as a result of syllogistic fallacycaused by the AIBCC tools/solutions on the overall brand image in the consumers’ minds.Experimentation results demonstrate that the existing belief of the organized B&M retailers inIndia which assumes that the AIBCC tools/solutions are accurate is just a misconception anddoes not hold. On the other hand, when we experimented by identifying two main gaps (inputand output-level) in their existing AIBCC tool/solution for six months at over 35 percent storesof a select retailer, the real treatment effect indicated that the experimental group of stores hasshown (i) two times higher rate of conversion to any promotional/campaign messages; (ii) 19times better in capturing the ‘Purpose of Purchase’ field; (iii) 22% lesser consumer communication expenses; (iv) 22.80% higher revenue generation; and most importantly; (v) 4.25 times higher store-level profits in comparison with the control group of stores. We have also noted that in the control group of stores about 36% of the customers/consumers who have received the promotional/campaign messages from the automated AIBCC tool/solution were not real consumers. Besides finding evidence of the syllogistic fallacy and trap, our results are also consistent with our ‘Theory of B&M Retailing in India and the concept of ‘Debiasing by Instruction’ by Evans et al.