Deploying AI to Produce Real-World Business Value

  • 4 min to read
Deploying AI to Produce Real-World Business Value

Aleksandar Velkoski, Ph.D., Director, Data Science, IT at National Association of REALTORS

Executives are increasingly focused on enhancing artificial intelligence (AI) adoption within their organizations. According to the 2020 Gartner CIO Survey, AI ranked number one on CIOs’ list of top 10 game changing technologies. Data analytics, which is typically discussed within the context of core AI and supporting technologies, ranked second on the list. 

Although increased AI adoption is central to modern organization strategy, organizations continue to fail to deliver business value from AI. Of the biggest challenges to AI adoption, a lack of use cases, as well as an inability to measure or value success, continues to permeate . In addition, there is often misalignment among technologists who are responsible for AI innovation and business line leaders who are responsible for decision-making. The misalignment is so predominant that through 2021 80% of business line leaders are expected to override AI. 

To increase AI adoption, executives should take tangible steps toward crafting projects that leverage AI, with a heightened focus on business implementation. One path commonly traveled, as described in Figure 1, is to leverage AI to improve the efficiency and accuracy of human actions and decision-making.

Deploying AI to Produce Real-World Business Value chart

Given the challenges outlined above, though, how can we identify prediction problems that translate to real-world business value? The key is to promote cross-functional collaboration, identify clear business opportunities that can be transformed into AI solutions, and implement AI solutions throughout the business. 

Cross-Functional Collaboration. Start the journey by convening a cross-functional team. It may seem like a trivial starting point, but for AI innovation to succeed, teams must come together from the beginning and work to breakdown silos if silos exist. In fact, the most digitally advanced companies — those successfully deploying digital technologies and capabilities to improve processes, engage talent across the organization and drive new value-generating business models — are far more likely to perform cross-functional collaboration.

At the National Association of REALTORS®, we leverage cross-functional collaboration to accelerate AI adoption. The Data Science Group (DSG) and the Marketing, Communications and Events (MCE) Group recently worked together to identify a high-impact business opportunity related to increasing member engagement. To ensure that the process went smoothly, and that disparate needs were met, we openly and honestly discussed everything from technical capabilities and limitations to business unit goals and objectives. 

Through our collaboration, we identified a specific business opportunity to enhance member engagement. We agreed that increasing attendance at the association’s REALTORS® Conference & Expo, which is the largest annual event for real estate professionals, was a great way to deliver real-world value. When members attend the conference and consume conference educational and networking opportunities, they’re also more likely to be engaged and retained, and that serves as a win-win for members themselves and the association. 

From Business Opportunities to AI Solutions. It’s one thing to identify a business opportunity, and it’s another to establish an AI solution. In order to go from business opportunity to AI solution, you must have a good sense of your organization’s AI capabilities. This is one of the most important reasons why you need to establish a cross-functional team, particularly one that integrates technology stakeholders as partners in the process.

In our case, we transformed our business opportunity into a simple prediction question; namely, “Who is likely to attend the REALTORS® Conference & Expo?”. We gathered a significant amount of data, including core member engagement data, demographic data and behavioral data, and then we leveraged techniques in AI to build a robust predictive model. Once the model was built, it was deployed in production via an API. Finally, we used the API to append to each member record the output of the predictive model, which was an individualized score that represented each member’s likelihood to attend the conference. 

Through leveraging our predictive model, the DSG was able to identify an important segment of members that not only were likely to attend the conference, but also had not attended the conference previously. The DSG provided the member segment, including information about each member’s unique decision-making styles,communication tastes and conversion preferences, to the MCE Group to develop a comprehensive strategy for implementation. 

Business Implementation.Value comes from action. At minimum, business implementation is about augmented decision-making; using insights derived from machine intelligence to make better ad hoc decisions. At best, it’s AI at scale. 

In our case, the MCE Group took the DSG’s findings and crafted a sophisticated strategy to encourage members within the segment to attend the conference. The strategy included a campaign with personalized messaging that spoke to members’ unique needs and interests. The MCE Group divided the overall segment into five distinct categories, mainly by decision making styles, communication tastes and conversion preferences. Then the team created custom messaging for those groups. Groups included personas like Proud to Be an American Pam, Glory Seeking Gloria, Discount Deadline Dave, Rose-Colored Glasses Rosie and Generic George.

Ultimately our goal was to encourage members to attend the conference. After our campaign was complete, and the conference itself had finished, we observed a segment attendance rate of nearly 1.5% and an increase in revenue of roughly $100k. That may not seem significant, but given the complexity of the call to action — that is to say it required that members incur a collection of direct costs (conference fee, flight and hotel) and indirect costs (time commitment and lost business) — the rate and revenue gain was significant. This is especially true given that segment attendance outperformed almost all others. Based on this and a variety of other factors, it was clear that the effort was successful. 

Increasing AI adoption is a formidable goal. Although it comes with its own set of unique challenges, the rewards can be notable. Throughout your journey, you will surely encounter breakdowns in communication, differences of opinions and questions about the true value of the work. Push through those challenges and stay on course. Also, remember that perfection is the enemy of progress.

Promote cross-functional collaboration, identify clear business opportunities that can be transformed into AI solutions, and implement AI solutions throughout the business. If you do so, you’ll be well on your way to generating real-world business value with AI. 

 

Dr. Aleksandar Velkoski serves as the director of Data Science at the National Association of REALTORS®. In addition, he is an adjunct professor in the College of Computing and Digital Media at DePaul University, and co-founder and organizer emeritus of Chicago ML, a community of machine learning scientists. He serves as a distinguished faculty advisor and mentor at the Innovation Development (iD) Lab at DePaul University, and as a mentor at Second Century Ventures.