Data-enabled learning, network effects and competitive advantage
Updated: Mar 8
With machine learning firms are continuously improving their products via customer data (often in real time). How much competitive advantage do firms get from this type of learning?
In my recent paper with Andrei Hagiu, we call this type of learning "data-enabled learning" and model competition between firms that enjoy data-enabled learning. We explore the implications for competitive dynamics of three new features of data-enabled learning compared to traditional types of learning (e.g. learning-by-doing): (i) learning increases willingness-to-pay rather than reducing marginal cost, (ii) firms can improve their products for each customer based on their individual usage experience, (iii) products can improve while customers are still consuming them. The model allows us to analyze data-sharing, the shape of the learning curves, and other factors affecting an incumbent’s competitive advantage. We also show when and how network effects arise endogenously from data-enabled learning.