Digital Retail Is Drowning in Demand Data. AI Can Help
Demand data is everywhere, but supply planning response hasn’t kept up. Rebalancing is key for keeping customer promises.
When Amazon.com came of age in the 1990s, some traditional retailers dismissed e-commerce as a niche channel appropriate only for specific product categories. In late 2023, BCG projected that online sales would account for 41% of all retail worldwide by 2027. Today’s reality is that anything can be bought online and across multiple channels, none of which are mutually exclusive in the eyes of the customer.
Retail supply chains are drowning in demand data. Zero100 estimates that demand chain data is at least 6x greater than supply chain data overall and growing faster. Upstream logistics, manufacturing, and sourcing operations looking for better efficiency and agility need some way to handle all this data.
AI may be the answer.
Digital Retail Has Exploded Demand Data
It is common for brands to have a half-dozen digital customer touchpoints before purchase. Between online search, marketplaces, social media, company websites, and physical store visits, customers create hundreds of demand data elements before even making a purchase. Amazon’s buy page, for instance, generates dozens of data points for every item a shopper considers before an order is even placed.
Plus, after-purchase data elements, including product reviews, merchandise returns, YouTube videos, and loyalty programs, add still more complexity to demand data, especially in terms of customer satisfaction and willingness to buy again from the same brand. Layer on the compliance and reporting demands of regulators, investors, and assorted watchdogs, and the data puzzle facing supply planners can be overwhelming.
The challenge for leading consumer brands like Kroger, Nike, and Starbucks is a Goldilocks problem: how to get digital retail “not too hard, not too soft, but just right”? 75% of customers say technology makes shopping better, but the same number still prefer talking to a human for customer support. Self-checkout at the grocery store is great, but abandoned carts are still too common. Demand data is everywhere, but supply planning response, especially at the customer-specific level, hasn’t kept up.
Balance Demand Data with Supply Detail
Solving this problem is a matter of finding balance with the right supply-side data so that customer promises can be made and profitably kept. Until recently, this has been handled in batches with demand forecasters using POS data and assorted modeling tools to set inventory levels and lead times with supply planners in an S&OP process.
More sophisticated systems, like one built jointly by Unilever and Walmart Mexico, connect demand data with supply data at a super-granular level, including item, location, and date, to automate CPFR (aka Collaborative Planning Forecasting and Replenishment). The system in this example generates more than 3.1 million forecast combinations every day, and its neural network model makes 12.5 billion computations per day using AI to automate what was once a labor-intensive process.
But this kind of demand data is still limited to optimizing supply for products that already exist. Data that captures what customers might want is another huge area of potential value that hasn’t been well connected with the supply side. AI now allows a more direct connection between demand data from products being used by customers back to sourcing and innovation.
Demand Data Feeds Faster Sourcing and Innovation
Fast fashion provides an example of how this kind of product use data has been and still is shaping supply chains in apparel. Zara pioneered this thinking with sales associates feeding shopper insights directly back to sourcing to turn styles faster than competitors. More recently, SHEIN has gone full virtual with an AI-driven model to do the same, but even faster, using an Uber-style network of manufacturing partners and TikTok influencers.
Just this year, Walmart introduced a program called Trend-to-Product, which plumbs social media data and uses genAI tools to create mood boards, tech packages, and other product documentation to feed to suppliers. The tool is built specifically to speed up new product innovation and allows Walmart to get new fashions on the shelf in as little as six weeks.
AI Is the Unlock
The six-to-one demand-to-supply data imbalance has been a problem. AI can turn it into an opportunity.