An Interview with Kevin Kennedy, Business Systems Manager at Terminix, a ServiceMaster Brand

Pulling together large sets of good data is the essential first step in any AI project. What is just as important, however, is having someone who truly understands the data itself from a technical and business perspective. For the Jedox AI project at ServiceMaster, this data expert is Kevin Kennedy, who has served as business systems manager at its division Terminix since 2005. We recently spoke with Kevin about the project that is looking into customer loyalty using AI, the experiences gained so far, and why he would recommend the Jedox AI technology to other FP&A teams.

Tell us about your work and previous experiences with Jedox at ServiceMaster and why Jedox was the right tool for your AI project.

I have helped with the original queries, setup, and data integration as the subject matter expert during the deployment and further development of the Jedox solution over the past three years. Jedox is a great tool for doing ad hoc analysis and bringing in huge amounts data. With Jedox, the FP&A team also has more self-service for its daily work.

For our AI project in particular, we are managing masses of data from several sources. Jedox is a great solution to assemble the right data as well as set, track, and monitor it over time. Instead of using an ERP system and different budget ledgers, we are using Jedox because it is much easier to change and manipulate the data. There is also quicker turnaround because FP&A is in control of it.

What kinds of information are being pooled into Jedox for your AI project assessing customer loyalty?

Jedox currently holds all our customer sales agreement information, which pretty much covers the entire customer experience. This includes customer purchases, frequencies, billing, customer balances as well as all the attributes about our products that they have purchased. We are now ready to make additional queries that require information about work orders along with revenue and any customer allowances that have been given. Lastly, we want to include all customer requests, which is when they call in with a schedule, billing, or other type of inquiry.

What experiences have you already made two months into the AI project?

From a standpoint of someone providing the raw data for the AI model, I would say we have learned a great deal about the system itself – how it is processing our data, how long it takes, and what capabilities the integrated AI technology has. I would say we are a fairly large business with a considerable number of customer agreements, transactions, and data elements. So, we are very happy with the Jedox system processing our data in a short amount of time. We are also constantly brainstorming about all the different data elements that we have available and could throw into the AI tool to help discover a new pattern or insight.

So far, we have predicted customer behavior with a very good confidence rate using the Jedox AI engine. Now it is all about understanding why. If this is our list of customer features that influence customer loyalty, which decisions is the AI making out of these features to think these specific customers are going to behave in a certain way?

What questions still lie ahead in your AI project?

The next step is going to be a control test to follow these customers into the branches and see how that plays out over time. I think the way that we can analyze the contribution of the AI results is to create two groups of branches with a similar geographic and economic setup – one that uses the AI results and one that does not – and compare the two.

We are already looking at many different aspects when it comes to customer loyalty, improving the customer experience, and increasing customer retention. Therefore, it is also important to show that any improvements in the branches actually stem from the AI information and nothing else is skewing the picture.

Why would you recommend Jedox to other companies looking to add AI capabilities in FP&A?

Based on what we have seen so far, Jedox gives a clearer, easier path to get started with AI instead of having to work from the ground up and needing specialized data scientists or developers on your staff. We have the ability to feed a dataset to the Jedox AI engine and then the “magic” happens accessing pre-built algorithms based on industry-standard AI-libraries. With Jedox, the finance department gains a direct solution pathway to information that it could not access before.