09-22-2026IMPACT COUNCIL

20 ways to measure ROI from AI initiatives

How leaders define value beyond adoption, speed, and experimentation.

[Photo: Getty Images]

The Fast Company Impact Council is an invitation-only membership community of top leaders and experts who pay dues for access to peer learning, thought leadership, and more.

BY Fast Company Impact Council

Adoption of artificial intelligence seems to be happening at, well, the speed of AI. Instant everything doesn’t always translate to instant return on investment. Jumping on the AI bandwagon without knowing what outcomes you want and how you plan to measure them makes for much noise without any real traction.

So, how can you tell if all the new AI tools you’ve added or want to amount to sound investment? We asked members of the Fast Company Impact Council how they measure ROI on AI initiatives. Here, 20 of them share what matters when deciding whether AI is worth the buzz or has fallen flat.

1. OUTCOMES FOR SPECIFIC PROBLEMS AT A SUSTAINABLE COST

Bigger frontier models don’t automatically create bigger business value. The real benchmark for AI is whether it delivers measurable outcomes for specific problems at a sustainable cost. One area where we really see this playing out for our customers is in internal applications and dashboards. Many Software-as-a-Service solutions are being replaced by custom AI-built tooling. It’s cheaper, faster, and adaptable. — Eric Simons, Bolt.new

2. PROGRESSION OF WHERE YOU NEED TO GO

I break our initiatives into advanced, maturing, emerging, and nascent. For advanced programs, they have clear outcomes like higher throughput and workflow velocity and are scaling. Mature efforts show clear business outcomes but are not yet reaching full potential even at their current scale. Emerging initiatives have a clear outcome in mind, but no measurable impact yet. Nascent effort is largely individual experiments. AI usage and token consumption are not proof of ROI and often confuse what you want to impact. I recognize some efforts are in the early stage, but you must show the progression of where we need to go. — Thomas Scott, Wrike

3. STANDARD BUSINESS OUTCOMES

Our approach to AI has always been about creating concrete value. It’s pretty simple: We measure our standard business outcomes, not AI usage. The idea is to focus on high impact use cases and evaluate whether AI is augmenting teams, speeding up innovation, and tightening security. — Khozema Shipchandler, Twilio

4. QUALITY AND WHAT THE NEXT TEAM CAN REUSE

Usage volume is the easiest AI metric to pull, but it tells you almost nothing. Same with speed. Anyone can make something faster now. What I look at is two things: quality and what the next team can reuse. Does the work still meet the quality bar we’re known for? Can the next team use what we learned to reach a better result faster? We recently designed, built, and launched a mobile app in 72 days, compared with six months or more using the old approach. That........

© Fast Company