Machine learning will empower eCommerce retailers to adapt and react to changing market conditions with unparalleled accuracy.
Unless you’ve been living under a rock for the past couple of weeks, you’ll have heard about the defeat of the world’s greatest Go player at the hands of an artificially intelligent algorithm developed by Google.
A Go tournament is a great exhibition of the distance AI research has come, but it’s in industry and business that AI will have the greatest impact over the next decade. While they’re nowhere near creating the human-like intelligences promised by science fiction, computer scientists are applying machine learning and other so-called cognitive technologies with human-like abilities within specific domains, including eCommerce.
Developments in cognitive technology will provide adaptive, interactive, and contextual systems for improving automated decision-making across the eCommerce domain, allowing retailers to create eCommerce experiences tailored for individual consumers in a way that has only been possible in a clumsy and inaccurate fashion up to now — we’ve all seen the wildly inappropriate product suggestions that current eCommerce stores can produce.
Demand forecasting is a classic problem in retail. How does the retailer know how much stock to buy? Overstock and the surplus will often have to be sold at a loss. Understock and demand will exceed supply — potential profits will be lost. Retailers already employ an array of big data analytics tools in an attempt to forecast demand, but it’s likely that machine learning and cognitive technology more broadly will be able to do a better job by ingesting huge amounts of information and learning from that data to provide a probabilistic prediction of future demand.
As I’ve already said, the effectiveness of product recommendations for upselling and cross selling is not all that a retailer might hope. To be sure, current systems are better than nothing, but with improvements in machine learning and an investment in devising specific solutions applicable to business, it’s likely that artificially intelligent algorithms with deep insight into buying patterns, customer sentiment, and the individual history of specific shoppers will make a significant impact. Companies like RichRelevance have already made progress in this area.
Price optimization is a complex field and it’s all but impossible for humans to comprehensively monitor and respond to price changes in a market that is prone to constant fluctuation in response to a bewildering array of factors that range from customer demand and competitor pricing to supply constraints and weather patterns. Price optimization is a perfect application of cognitive technologies that are able to make by-the-second predictive pricing changes tailored to specific customers.
The application of cognitive technologies to eCommerce is in its infancy, but many companies are making significant investments towards creating practical implementations of the theoretical advances exemplified by Google’s AlphaGo.