By Michael Kim, Chief Information Officer at MultiPlan and finalist in Best Hybrid Cloud Solution category in the 2021/2022 Cloud Awards

In the excitement to implement machine learning and AI, many organizations lean heavily on technology and machine-driven models but ignore the need to balance automation with human intelligence. When a company has years of quality data managed by a team of successful data scientists, it might seem that fueling a successful AI or machine learning effort would be a “sure thing.” But failing to acknowledge the significant role that human intelligence plays in the process, especially surrounding change management efforts that accompany the project, will undoubtedly impede success. According to Harvard Business Review, only 8% of firms engage in core practices that support widespread adoption. Most firms have run only ad hoc pilots or are applying AI in just a single business process.

But when everything is left to a machine model, vital business operations and needs can get lost, and user adoption can tank when internal teams aren’t adequately involved with and educated about how the new machine learning-based technology will make their jobs better and achieve better outcomes for their business unit and for customers. Companies can overcome the limitations of the machine model by leveraging their teams’ experiences and their commitment to client relationships alongside an innovative data-driven solution.

Where to start?

It’s important to first examine a company’s culture, organizational structure, and determine ways that best support broad AI and machine learning adoption. It becomes incredibly important to understand that some stakeholders might perceive machine learning and AI implementations as a threat to their routines and possibly their jobs. Why is this so important? Because organizational culture is routinely cited as one of the top roadblocks to adopting AI and machine learning within organizations. An implementation team needs to take steps to bring the negotiations team along with them throughout the process, get their buy-in, and ensure implementation.

To scale up and fully integrate its AI and machine learning efforts, teams should consider the following seven steps for success:

  • User Focus Groups: A business team should conduct user focus groups with its end-users to understand the challenges from the outset. End-users should be continually engaged throughout the development and deployment process and encouraged to provide input.
  • Proof of Concept Pilot: Pilot testing groups should be developed early to provide feedback on the tools in development. The business team should ensure that groups are diverse with a mix of both challengers and change embracers. This should be an intentional strategy that secures feedback from all types of users and identifies potential pushback before rolling out a new system. The pilot effort should also include technically-challenged users to test for ease of use. Ideally, pilot users can test the new process for at least thirty days to iron out issues and address challenges before a full-team rollout occurs.
  • Internal Campaigns: Creating a comprehensive internal marketing campaign to educate and persuade end-users about the new machine learning or AI tool can help build excitement about its use and improve speed of adoption. The goal is to make sure end-users understand how AI and machine learning works, how it helps them achieve their goals, and how it makes their jobs easier. Case studies and testimonials from the pilot should be developed and used to tell the story so that users learn about the solution directly from their peer group versus a push from leadership. The team can also use relatable examples of machine learning from daily life like targeted ad recommendations and Spotify music recommendations, demonstrating how machine learning can make life easier.
  • Training: A “given” in most new technology tool rollouts, training is a critical element. But in addition to having users receive traditional training on a new tool or system, pilot users should be engaged and encouraged to share their real-world examples and experiences of working with the technology, challenges they overcame, and positive ways it has impacted their work. Creating mentors and smaller working groups will help reinforce training efforts and ensure success.
  • Feedback Loops: After launch, project teams need to stay close to user feedback and offer multiple channels for end-users to communicate. For instance, each day end-users could be encouraged to complete a brief, simple survey. Questions could focus on their comfort level with the new system, any challenges they encountered, “aha” moments they had about machine learning, and even temptations they faced when using the new tool. With this information, a team should be able to quickly identify any issues or potential barriers to adoption. Machine learning project teams need to act on this feedback, communicating workarounds to challenges and any plans for solutions.
  • Tracking and Measurement: Finally, tracking and communicating well-defined performance indicators with end users creates a transparent, open, and collaborative environment where everyone involved understands how success is measured and is able to see the direct impact that machine learning is having on their business unit and individual performance. Compliance reports can be used to monitor employee usage and adoption, track feedback, and watch for new opportunities for improvements or additions.

How to use AI?

Balancing a solution capable of leveraging as much machine learning as possible with end-user needs, client demands, and business requirements is critical. Collaboration among operations teams, technology teams, and other key stakeholders ensures success throughout planning, formula adjustments, and ongoing buy-in efforts. While these aren’t always easy hurdles to overcome (and often add additional time to deployment) the final results can be exceptional and build the foundation for teams who value and embrace machine learning and AI.

Obviously, the human element of machine learning isn’t the only key factor to success, but it is the easiest to overlook, often the first to be eliminated, yet often the most critical for successful results. When everything is left to a machine model, essential business operations and needs can get lost, end-users and customers can become frustrated, and the next machine learning project may be met with skepticism. Limitations of machine models can be overcome by leveraging years of industry expertise, internal business and technology experts, and commitment to client relationships, creating a culture of energy and passion around the promise of AI and machine learning.

Visit MultiPlan to learn more about our award-winning machine learning success.