AI for Customer Journeys: A Transformer Approach.

The article focuses on a transformer-based artificial intelligence (AI) framework designed to model and predict customer journeys by analyzing sequences of customer interactions across multiple marketing channels. This model employs a heterogeneous-mixture multihead self-attention mechanism to capture both population-level trends and individual heterogeneity in touchpoint effects, enabling more accurate predictions of customer behaviors such as channel visits and purchase conversions. Empirical application using hospitality sector data demonstrates the model's superior predictive performance compared to benchmarks like hidden Markov models (HMMs), point process models, and long short-term memory (LSTM) networks, particularly in long-term forecasting and identifying high-potential customers for targeted marketing. The approach also provides descriptive insights into the timing and impact of different marketing touchpoints, offering managers actionable guidance for optimizing marketing interventions. Beyond marketing, the model's flexibility suggests broad applicability in other domains involving sequential customer or user interactions.