Celect raises $5M to help retailers stock their shelves

Boston-based startups creates a recommendation engine based on customer options

If you've ever shopped at Amazon (and who hasn't, really), you'll know that recommendations are a big part of its customer service. If you bought book A, then you'll be shown books purchased by other customers who also bought book A.  

It's a great way for Amazon to encourage additional purchases. But there's a lot more to knowing what to put in front of customers than that.

Celect hopes it can enhance a retailer's ability to stock its shelves with an assortment of products that a customer may want to buy. And it's raised $5 million in Series A, led by August Capital, with participation from Activant Capital Group, to do so. 

Boston-based Celect, founded by MIT professors Vivek Farias and Devavrat Shah - both of whom are experts in the field of "Customer Choice Modeling," essentially looks at purchasing habits based on options vs opinions. 

A purchase is likely to occur due to the choices before a buyer vs a buyer's preferences and tastes.  

For instance, if a person buys Pecorino cheese, it may be that it's not because they like the taste as much as it's a substitute for what's not on the shelf. Or in my case, when I purchased black flip flops, I did so mainly because the store didn't carry dark brown. Now if the store started offering me variations of black sandals, they'd totally be off the mark.  

"Customers choose in very complex ways - customer decisions are governed by choice: what a customer purchases is both constrained and influenced by what that customer is offered - that is, what their options are," said John Andrews, CEO of Celect. "Celect uses comparisons between products derived from transactional, inventory, and online browse data to model choice."

Preferences based on choices are signals that aren't captured in traditional recommendation engines, Andrews explained.

"Traditional 'recommendation engines' leverage behavioral targeting techniques which group different customers together - Celect goes to the granular customer level at real time, using choice modeling to provide much more precise predictions of the optimized assortment of products to present to customers."

With the new funds, Celect plans to roll out its software across The Bon-Ton Stores, a regional department store with 270 locations. While Andrews wouldn't discuss how well his software worked at the store, he did say that results showed increase in in-store revenue. Recently, the CEO of Bon-Ton gave a shout out to Celect in their earnings call, saying: "We are testing new software called Celect, a predictive modeling tool that allows us to leverage customers' omnichannel purchase and browse behavior to identify merchandise localization opportunities. We believe localization can allow us to further increase market share in our low-volume doors."

Andrews couldn't give an example of how the software is actually making better recommendations, but he did share an example of how Celect helped one store in Lancaster, PA, which never sold Michael Kors handbags as the store owners didn't believe their demographic would support it. In reality, 15% of the population would buy Michael Kors' handbags, said Andrews. "Based on the buying patterns of customers, not just high-level demographic data, we can find opportunities like this where there is latent demand."