Inventory Optimization with Artificial Intelligence

Physical retail is not dead; although headlines regularly trumpet the large number of store closings, over 4000 net new retail stores opened in the USA last year. Retail is a $27 trillion-dollar industry worldwide with more than four out of five transactions still happening in the store.

"Retail is a $27 trillion-dollar industry worldwide with more than four out of five transactions still happening in the store."

However, the rise of Amazon and on-demand e-commerce is causing the retail sector to undergo tremendous shifts and being compelled to invest in accommodating changing consumer behaviors.

Retailers are increasingly recognizing that the data they have on products, trends, locations, and customer visits needs to be used to deliver better value to their consumers, and to retaining their loyalty.

That is why NGP Capital invested in Celect, a cloud-based machine learning platform, helping retailers optimize inventory decisions at procurement, store allocation, and consumer fulfillment levels.

Situation today: Fast-changing consumer needs and cost pressures

The rise of competition has not only challenged revenue growth but also put pressures on margins as well. Jeff Bezos famously said, “Your margin is my opportunity” and retailers are aggressively trying to cut costs, often tactically (store closures) rather than strategically (differentiation) to stay relevant.

Consumers have come to expect a product to be available in the right size, on the right shelf (or website), at the right time. We expect our store experiences to provide us the same level of speed and service as online, yet with the flexibility of experiencing the product and returning it in store.

"Consumers have come to expect a product to be available in the right size, on the right shelf (or website), at the right time."

Additionally, there is a significant shortening of the product and trend lifecycles, requiring retailers to significantly cut down planning, design, and production times (weeks instead of months).

Finally, retailers need to allocate their inventory intelligently at their e-commerce warehouse and at individual stores so that they have enough stock to satisfy their customers on every channel (avoid stockouts), but not overstock (to lead to excessive markdowns).

Unfortunately, poor inventory decisions are putting tremendous cost pressure on retailers today. In the US alone, over $250 billion-dollars are lost each year due to over- and understocking merchandise.

Challenges to overcome

Retailers understand the evolving customer needs. However, to satisfy the customer across channels, they need to improve their ability to predict their customer needs in advance.

Historically retailers have relied on instinct, experience, and stale data to make their planning, buying, and allocation decisions. Merchandisers have yet to capitalize on the rich and real time data from customers including online and point of sale systems.

Our research showed that in 2018, a well-known national department store was sending identical merchandise to five different stores in the Bay Area, even though the demographics, economics, and even the weather of these locations were very different.

Although retailers have made significant investments in cloud computing, data science, and data infrastructure, a significant amount of data is still siloed and in need of integration.

Additionally, data governance, data management, and quality control initiatives are still nascent. Therefore, even in progressive retailers, the models remain on the shelf and not in production environments.

Celect: Helping retailers optimize inventories from planning to fulfillment

Enter Celect — a predictive analytics and optimization engine, spun off from MIT and recognized by MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) as one of the 50 greatest innovations it has ever produced.

Using its patented algorithms, which works even with sparse data, Celect integrates with retailers’ existing historical and real-time data to create a comprehensive view of demand prediction for each SKU (stock keeping unit) at a national and hyper local store level.

Merchandisers can then use this data to make data-driven buying and allocation decisions for each store based on its unique demand profile.

"Merchandisers can then use this data to make data-driven buying and allocation decisions for each store based on its unique demand profile."

As customers expect seamless experience with shopping and home delivery, Celect’s fulfillment capability informs the retailer from which store or warehouse to deliver the goods to the consumer.

Retailers like Neiman Marcus, Saks Fifth Avenue, ALDO, and Urban Outfitters are already leveraging Celect within their organizations to power their predictions and take back their margins.

Celect’s platform provides many of the capabilities that retailers need to become more competitive and retain the loyalty of their customers.

We are excited to join this journey with Devavrat, John, Vivek, and the Celect team as they leverage their innovative technology to help solve one of the most important challenges facing retailers today.

‍