Steven Keith Platt is Director and Research Fellow at the Platt Retail Institute (PRI). He is an Adjunct Professor at Northwestern University and serves as the Research Director at the Retail Analytics Council, an initiative between the Medill School, Integrated Marketing Communications Department, Northwestern University and PRI.
He is also the Co-Founder & Chief Development Officer of RetailPredict.ai, a company focused on enabling sustainable revenue and profit improvement by providing AI-powered prediction models that are seamless to implement and can be deployed rapidly
What initially attracted you to Retail AI?
I have been working in retail analytics for over 25 years. The industry always had lots of data, but the analytics being applied to learn from this vast trove of information to operationalize the business was lacking. The ability to manage big data was the first major change and then as AI became more mainstream around five years ago, it was a natural progression to move to more advance computational methods.
Could you share the genesis story behind RetailPredict.ai?
RetailPredict.ai was an outcome of my lab work at Northwestern University, where I teach a retail AI course. Each quarter we work with a retailer to solve a business problem by applying AI solutions. Those use cases proved that there is an existing demand to solve those problems and that we can do this by applying AI. So, in the lab we run POCs; at RetailPredict.ai we take those findings, industrialize the models and put them into commercial production.
Why did you choose to focus on Retail AI?
Various reasons including:
Lots of data. Lots of addressable issues. Once you get past the majors (i.e., Walmart, Target, the Home Depots), many retailers with under $10B in sales do not have the resources to develop solutions in-house and are challenged to find the talent to help. So, we see lots of opportunities to help.
How can companies best leverage AI in a retail environment?
Success requires adoption/embracing at the leadership level. AI can require new ways for companies to accomplish things, and cultural impediments to change can present challenges. So, a roadmap is required. Also, an understanding of what it can and cannot do. Finally, a focus on short-term wins to establish credibility, rather than a boil-the-ocean approach, is helpful.
What type of productivity improvements have been seen from the implementation of AI in retail?
The range of solutions is pretty unlimited. From online order estimates to supply chain, the range of use cases to solve is vast. At RetailPredict.ai, our initial focus is around labor optimization (predicting store traffic for up to five weeks in advance) to better match labor with customers. For example, reduce staff when fewer customers are anticipated, perhaps add more when customer traffic is expected to increase conversion. As well, our hourly predictions enable store managers to better task manage (i.e., we expect a rush at this time, so let us have some extra folks at the checkout). Our other product predicts product demand to reduce out-of-stocks, eliminate excess investment in slow-moving products, test demand for new products, and manage markdowns around expected demand and price elasticity.
Is there anything else that you would like to share about RetailPredict.ai?
Our approach is unique in the market. Very inexpensive, easy to onboard use case-specific models that do not require extensive integration and can be launched quickly. We couple this with user-friendly dashboards for easy data interpretation. Alerts can be programmed to inform managers about changing conditions. Finally, confidence in the models is important, so we incorporate a variety of performance metrics.
Thank you for the interview, readers who wish to learn more should visit RetailPredict.ai.