Saurabh Parikh, VP/Innovation, Architecture & Data at CONA Services, a unit of Coca-Cola that serves the organization's bottlers across North America, addressed this topic at a recent New York conference.
“Our business is dependent on making sure the right products, in the right stores, in the right packaging are available to consumers,” he explained. “And bottlers play a very critical role in bringing brands to life in the marketplace.”
But with more than 1,000 stock-keeping units (SKUs) in its portfolio and a near-constant flow of new products, retailers can struggle to keep up. “The challenge becomes: How do you know what’s the right product assortment for a given outlet based on who the shopper is?” said Parik.
“And this is where AI comes in [and] where we put a lot of data together and build a specific portfolio that makes sense for that outlet and that outlet’s shopper.” (For more details, read WARC’s report: How Coca-Cola is using AI to generate fresh insights.)
He further reported that a “cosine similarity-based algorithm” now enables CONA Services to compare every convenience store in a set market with each another. If 53 such outlets are present in a particular geographical region, for instance, Parikh’s team crunches the numbers to identify those retailers with the most favorable equivalent sales patterns, customer profiles, and seasonal trends.
“What can I learn from those 53 stores?” asked Parikh. “What products are over-indexing there that I can bring here as well as predict how much business that will grow for the customer?”
The answer was to build a couple of algorithms, including one based on cosine similarity which worked particularly well.
“We put it in a couple of markets [and got] fantastic results,” he reported. “And I can tell you, we are working as hard as we can to roll it out across the [North American] market.
“The business opportunity is in millions. So, this is an example where AI is delivering value to our business, top line as well as bottom line.”
Sourced from WARC