How Reckitt Keeps Retail Execution Timely and Cost-Efficient With AI
Reckitt CIO of North America Varun Kakaria is well aware that many organizations are overusing artificial intelligence, and his approach to the technology has been to first focus on the basics and then use AI as a tool to make processes smarter.
One business area in particular that he said has untapped opportunity for this type of automated efficiency is retail execution — a naturally fragmented area due to the unique needs of varied retail partners and the markets in which they operate.
During the opening keynote presentation at CGT's Analytics Unite, held April 7-9 in Chicago at The Drake, Kakaria said retail execution requires an archetype-based approach, but not so much that strategies are differentiated so much for each market that they become siloed again.
"So how can we think of a hybrid approach where you don't give the same stuff to everybody, which might not be relevant to them, but we can still have a central spine," he said.
For this reason, Kakaria has focused on a three-pronged approach that takes into consideration channel structures, data maturity and cost to serve. As part of this, Reckitt has optimized efforts for its field reps, supporting them with AI so they can move away from time-intensive, manual Excel workflows and make the entire process more predictive so they can focus on selling work to improve execution.
Getting Market-Specific
In its U.S. market, the largest for Reckitt, Kakaria experienced challenges around millions spent on displays with no way to accurately measure value and return on investment.
"Display compliance is one of the biggest drivers in our promotional seasons where people kind of look for our products, so these are areas where we want to improve our accuracy and actionability," he said.
By using AI, Reckitt has been able to sell more displays and increase compliance through improved planning and data entry management.
Additionally, the company looked to optimize processes for its field reps through AI tools.
"That's where we had an idea … in this age of AI and algorithms and compute becoming higher, could we use data to better predict?" he said, adding that the approach pulled in insights such as customer data, cost data, Nielsen data, demographics data, seasonality data and promo data to determine which stores might be in trouble in the future and where stock problems could arise.
Also: Reckitt's Varun Kakaria on the new expectations for CIOs
For some of its larger customers, such as Walmart, Reckitt has deployed technology that consolidates action items by store priority so shared reps can make the most of the 30 minutes they are allotted per store visit.
"We have developed an automated compliance engine, and we have, in tandem, designed it to be integrated with the broker systems. So even if we don't have our brokers, those 10 actions will pop up," said Kakaria.
Previously, reps had to manage a workload of 30 to 50 items on their checklist at a time.
And in areas that do not have rep presence, the company now has enough actionable analytics (and not just dashboards for the sake of dashboards) to identify problem areas that need to be addressed.
Prior to this implementation, Reckitt was at 20% accuracy for out-of-stock predictions. In the four or five months after the launch, this number increased to 45%. Now it gets closer to 75%.
While there's been clear advancement, Kakaria understands there's more work to be done, emphasizing that he always chooses progress over perfection.
