Exciting New Ways to Analyze your Data...
Here at Medallia we firmly believe the information needed to win in the market is within your CX and EX data. This feedback is rich with insights that help organizations...
Your message has been received and we will contact you shortly.
Get the best in Customer Experience content delivered straight into your inbox.
Artificial Intelligence (AI) has the potential to add a tremendous amount of value to organizations looking to transform their Customer Experience. I sat down with Gregor Stewart, VP of Data Science at Medallia, to discuss Medallia’s approach to Artificial Intelligence in CX and how it enables companies to leverage AI for action to achieve tangible business results.
Priya: In a recent blog post we learned about your role at Medallia and the problems your team solves. Could you tell us more about Medallia Athena (the AI layer of the Medallia Experience Cloud), and how we’re making use of it?
Gregor: We’ve long been making use of what people now call AI, in various Medallia Experience Cloud (MEC) applications, most notably Text Analytics. Though sophisticated, the models and algorithms were siloed, which obscured their true potential. Medallia Athena provides a scalable and efficient way to deliver a set of foundational AI components (for us and our partners) that simplify the construction of intelligent features across the MEC platform. These foundational components include the multilingual text understanding models that we have been using to deliver Theme Discovery and Sentiment Analysis, the mechanisms we used to build those models, as well as other components that are entirely new. We can now quickly deliver intelligent, real-time features that adapt to each client and user, in all MEC applications, which is incredibly exciting.
Priya: Exciting stuff! Do you have an example of a new application powered by Medallia Athena?
Gregor: Absolutely – let’s take one of the Medallia Athena-powered features we’re working on. We observed that many customers and employees offer suggestions in their feedback, about what companies should do or should do differently – some more useful than others. We asked ourselves, how could we help someone like a regional manager at a global retailer or bank quickly discover the ideas that are worth pursuing or further investigating? We created a system of models that identifies all the bits of textual feedback that are suggestions, weeds out the ones that are too generic, too specific, or obviously not actionable, and uses the body of feedback to deduplicate and roughly order those that remain by their expected impact. For example, if even a few people are suggesting “offer a battery replacement at a discount” and many people are talking negatively about battery life and similar topics, this is probably an action a company should consider.
At scale, this would be incredibly time consuming for even a group of people to do. But once these suggested actions are surfaced and ranked, it takes just seconds to go through the list, prioritize those that make sense to action, and mark the ones that aren’t useful. We can then make use of these judgments in adapting the behavior of the system, so that the quality of these lists gets better over time.
Priya: What are some of the principles we’re following as we build additional applications?
Gregor: The core values of Medallia Athena are: action, collaboration, adaptation, and respect. Above all else, we focus on helping our customers take wise action in three key contexts: immediate actions to take with customers; longer term improvements that are evaluated and put into effect by employees; and strategic changes for the enterprise that can allow for large-scale business transformation.
We believe that AI should foster broad engagement in discovering, experimenting with, refining, and scaling actions that improve overall experiences. Just as important is that our systems be sensitive to differences between people and organizations, and (perhaps even more importantly) that our use of customer data in doing these things is up front and easy to understand.
Priya: Are there other applications where you’re applying these principles?
Gregor: We have another application that enables companies to better direct their attention in “closing the loop”, to focus on those customers most likely at risk to churn or reduce loyalty. Recently, I had my tires changed at a dealership. The manufacturer sent me a survey. I gave them a 6 and commented positively. I got a call on New Year’s Day to ask why my experience had been bad. I had to explain that I had no issues – it was just a tire change. This happens all the time: people who give “low” scores often aren’t at risk, while some score high (for whatever reason) but are disappointed or even frustrated by their experiences. We designed this application so that if people have to make calls on New Year’s Day, they are necessary; and that if they do happen to talk to a happy customer, the system will learn from that to better focus follow-up going forward.
Priya: The applications we’ve talked about are in the works today. What can we expect in the future?
Gregor: Our vision for Medallia Athena in Experience Management is a prescriptive one, where we (more often) recommend or nudge action, engage employees and companies in the use of these recommendations, and learn to do better from the user behavior and outcomes surrounding them. Here’s how this might play out in the two applications I’ve been talking about:
For our application that bubbles up customer suggestions, we have plans to enable impact experiments that will help our customers determine whether changes affect customer experience enough to be worth implementing at scale; and to assist them in scaling those actions that worked across their organizations by recommending those actions in similar contexts. For example, if a “self checkout” pilot in our our flagship UK store was successful, maybe we should recommend this to the managers of stores that have similar size/traffic/demographic in the US. Similarly, our vision for assessing risk is to graduate from doing a great job of recommending which customers to pay attention to, to suggesting what to do and perhaps even doing some of the associated work, e.g. drafting appropriate communications, using richer models of customers that capture their preferences and propensities.
Finally, we will be working to ensure that recommendations like these are fused into a simpler, day-to-day experience for all users. Rather than having users move from one tab to another looking for things to action, we aim to deliver a prioritized feed of things to do, try, investigate further, and reflect on, for each person.
Learn more about how you can leverage AI to transform your CX with Medallia Athena.