Connecting Splunk’s Use of Predictive Machine Learning with Intelligent Assistance

dunkin-donuts-710x400Operational data aggregator and business analytics provider Splunk announced this week the integration of machine learning into the core platform capabilities. In the context of intelligent assistance, Splunk’s product portfolio doesn’t include customer-facing web self-service solutions and the company doesn’t position itself as an intelligent assistant vendor. But it’s valuable to look at the recent announcement, made at the company users conference, and the evolution of Splunk‘s predictive business analytics software to get a sense of where intelligent assistance might be heading.

In a keynote presentation from the conference, Shawn Ahmed, senior director of IoT & Business Analytics with Splunk, demos analytics tools used at Dunkin’ Donuts (about 27 minutes in). The doughnut maker, with a market cap of $4.6 billion, collects and analyzes real-time point-of-sale data from its many retail locations.

Ahmed shows how Dunkin’ Donuts is applying Splunk’s machine learning toolkit to create a linear regression model against historical sales data. What the company ended up with was the ability to predict when people would be buying donuts, and in what volumes, from its stores. These predictions help ensure that inventory is available and doughnut racks are stocked when hungry customers arrive.

The use of machine learning also enabled Dunkin’ Donuts to create and gauge the effectiveness of one-to-one marketing campaigns. If the predictive algorithms foresaw a surplus of sweet treats at a specific location, a mobile marketing campaign targeted loyalty customers with coupons, trying to coax them into the store for a spontaneous doughnut purchase. Splunk analytics software also measured how many customers actually responded to the enticing offer.

It doesn’t take much imagination to envision how Splunk’s predictive functions could be used to enhance the capabilities of an intelligent assistant. If a loyal Dunkin’ Donuts customer has enabled the brand’s virtual agent or chatbot, the bot could send a notification to the customer when there’s an overabundance of fresh hot doughnuts at their neighborhood store. The bot could be the one to offer the customer a discount coupon, helping to build the relationship between customer and doughnut bot.

As companies use machine learning tools to predict customer behaviors, intelligent assistants can leverage those insights. Some of the insights might help the intelligent assistant predict what the customer wants to know about. Others can enable the assistant to tailor offers to the customer that are aligned with both the customer’s past purchase behaviors and the company’s current deals. We’ll see many more intersections between predictive analytics and intelligent assistance as technologies evolve.



Categories: Conversational Intelligence, Intelligent Assistants, Articles

Tags: , , , ,