The Most Dangerous AI Metric Is The One That Says You’re Successful
The Most Dangerous AI Metric Is The One That Says You’re Successful
Publish Date: 2026-06-28 09:26:00
Source Domain: www.forbes.com
Don’t measure the wrong metrics.
getty
Artificial intelligence (AI) has a CX measurement problem. It’s not that you can’t measure success it’s that many companies are measuring the wrong things.
A new study from Laivly, a pioneer in applied AI for the contact center, found that 65% of organizations considered their AI initiatives to be successful. That may seem like a successful number until you dig deeper into the data, which tells a different story. I had a chance to interview Jeff Fettes, founder and CEO of Laivly, for an episode of Amazing Business Radio, and he shared the danger in measuring the wrong metrics. My first take on his comments was that the wrong data will give you a false sense of security.
It turns out that 43% of AI projects don’t meet time deadlines, and more than half have exceeded their original budgets. And this finding summed it up:
Nearly three in 10 leaders (28%) said AI had directly contributed to lost revenue because it could not effectively handle complicated customer support issues, thereby frustrating customers and increasing churn.
In addition, one in five leaders (20%) said they know there is lost revenue but can’t quantify the damage.
While my focus is typically on customer service and CX, these findings go beyond the contact center. They will especially resonate with C-Suite decision-makers under pressure to deliver successful AI projects. In fact, Laivly found that 43% of boards of directors are dissatisfied with AI progress and are pushing the C-Suite to prove its success.
Defining the wrong success criteria is a major problem for any company implementing AI solutions intended to streamline processes and reduce costs.
Don’t Confuse Deployment with Success
Fettes says, “Organizations often celebrate milestones that are easy to see, such as implementation dates, reduced staff, chatbot launches, or how many tools are being deployed across the business. These metrics create the appearance of progress, but appearances…
