A Simplified Approach to Generating ROI from AI Apps

It’s not surprising that the primary focus of most of the CIO roundtables I am part of these days is getting AI apps planned, built, and deployed. But more recently, executive management has asked IT to justify these projects by documenting the benefits and value to the business. Gone are the days when you could simply do something cool that incorporated some obvious AI functionality.

Nowadays, management wants return on investment (ROI) calculations as part of any AI proposal. But how do you calculate ROI on something completely new and different—or on something as complex as AI, which brings with it lots of issues such as data privacy concerns, regulatory compliance complications, and all-new security risks? These considerations have forced many dev teams to reduce the scope of their AI plans—which is something of a relief for those charged with delivering on the promise of AI.

This is putting a new emphasis on building AI apps that can meet ROI goals by improving metrics that are already being measured while reducing the scope of AI projects to focus on the most compelling improvements (including delivering the largest ROI) for that digital system. This is a smart move. Dev teams can use existing metrics as guideposts for application design, evaluating the current apps to identify the most beneficial ways to use AI. And when a project is aimed only at improving current processes, not inventing something completely new, the scope is greatly reduced.

That new approach brings some compelling benefits:

• When projects aim only to improve metrics and processes, there is no need to create a new set of ROI calculations, circulate them, and get the various stakeholders to agree. The team can get to work and deliver the prototype much faster.
• Leveraging existing application logic and using the same or mostly the same data shortens the process of ensuring data compliance and security. AI projects with a larger scope are often held up as various teams outside of IT, such as legal and compliance, sort through the potential issues.
• Starting with more clarity around the application design and its value-add shortens the design process and reduces the number of changes needed after each code commit. Quite simply, it takes less time to enhance what is known than it does to create something new.

This change in approach to ROI may not work for every AI project. However, many organizations are finding that unknowns surround the design and justification of a brand-new AI application. Focusing on improving the most important metrics that are already being measured can be a valuable shortcut. And many of CIO’s roundtable attendees are already seeing the value of this approach and plan to implement it soon.