Selecting the Right AI/ML Problems

Use Case Prioritization

A crucial initial step in a digital transformation effort can be to perform a use case prioritization exercise to identify a portfolio of high-priority AI/ML problems appropriate for an enterprise, business unit, or division. Setting priorities for use cases involves thinking through all of the dimensions mentioned in this chapter, including problem tractability, economic value, and ethical considerations.

At C3 AI, we have reviewed hundreds of enterprise AI problems over the last decade. A typical first step involves a full value-chain exploration of high-potential AI/ML use cases. The following figure depicts an illustrative, high-level value chain map of AI use cases for a financial services company.

Infographic

Figure 24: High-potential AI/ML use cases for a financial services company. Example from a C3 AI strategic workshop

After developing a case map, business leaders typically want to perform additional exploratory work in certain areas to further flesh out the most tractable and valuable use cases for their organizations.

Most organizations can conduct a deep-dive exploration and understanding of AI/ML use cases quite rapidly, without requiring a prolonged strategy phase. The leadership team usually already has the relevant business knowledge with the help of SMEs. At C3 AI, we have developed a playbook over the past decade that lets us rapidly identify a portfolio of high-potential AI/ML use cases through a series of screening and scoping exercises and workshops. The basic principles are quite simple.

First, we ask business leaders and their management teams to fill out a template of the top business problems that they think could benefit from the application of AI. This activity is performed as pre-work before more detailed and in-depth workshops and discussions take place. The high-potential use cases outlined in the figure above can serve as inspiration for such an exercise. But we have found that, in most cases, business leaders and SMEs have already given significant thought to areas that can benefit from the application of AI/ML. The following figure shows an example of a pre-work template for use case identification and prioritization.

Use Case Overview (describe in simple terms what value an AI solution would bring to the business):

  • Identify early warning signals of clients who are likely to move their investments to another financial services company to enable proactive and timely engagement of those clients and take proactive action to intercept churn.

Infographic

Figure 25: Illustrative template to be filled out as pre-work, ahead of use case prioritization workshops

This pre-work activity is then followed by one or more use case prioritization workshops. The workshops can take many forms. One of the most productive formats involves presentations made by individual managers proposing their candidate AI/ML use cases to a leadership steering committee. In such a workshop format, individual managers explain the reasons why they consider their use case to be high-potential and a top priority.

This format accomplishes two objectives simultaneously. First, it ensures that the business requirements and value around a specific use case have been thought through well and peer-reviewed by both leadership and the AI/ML steering committee. Second, this format ensures that the business has bought into the opportunity’s value and benefits. All too often, enterprise leaders delegate AI/ML initiatives to digital or IT teams, stepping away from direct involvement. Ultimately, however, the entire business needs to incorporate AI/ML technology as part of their day-to-day operations in order to unlock value. Incorporating AI/ML involves business process change and challenging change management activities. A format in which the business actively asks for investment, early in the process, ensures that there is strong buy-in from business managers, plus interest and alignment in wholeheartedly implementing the AI/ML technology as part of daily business operations.

Following one or more use case presentation and discussion workshops, most businesses can assemble a portfolio of AI/ML initiatives to prioritize, resource, and put into production to unlock significant business and operational benefits to the full enterprise.

The following figure shows an illustrative example of a portfolio of high-potential AI/ML use cases for a business unit within a financial services company. Time-to-value, tractability, and the actual economic value are plotted on the chart. Business unit leaders can use this portfolio analysis to plan out their AI/ML transformation roadmap. In the example below, for instance, they may start with AI projects for customer churn or anti-money-laundering – efforts that may require a “medium” effort or time to implement, but are very tractable and have high economic value. These initial projects then can fund additional efforts as part of an enterprise AI transformation roadmap.

Infographic

Figure 26: Illustrative portfolio of AI/ML use cases for a business unit at a financial services company

Once the initial use cases are prioritized, enterprises can then prototype those use cases and scale them into production. The next chapter focuses on best practices in managing AI/ML prototyping efforts.