Gartner: only 13% of CIOs see real value from AI
A Gartner survey of 2,002 IT leaders shows few companies get strong results from AI. What sets the ones that do apart: knowing when to stop a project.

Only 13% of CIOs say their company has already achieved significant or transformational value from artificial intelligence. The figure comes from a Gartner survey released on Monday (October 19), during the Gartner IT Symposium in Orlando. The firm heard from 2,002 CIOs and technology executives between May and July 2026. According to Gartner, AI use cases have multiplied over the past two years, but the odds of any one of them succeeding are still about one in five. Analyst Alicia Mullery summed it up: “volume is not value”.
What separates those who get results
Gartner compared the best-performing leaders with everyone else, and the difference it points to is management discipline. Having clear rules for deciding when a pilot should not go into production makes a CIO 2.3 times more likely to be in the top-performing group, yet only 31% have those rules. Having criteria to switch off a low-value AI that is already running counts 2.6 times, and only 17% have them.
Cost is also a concern: 58% of CIOs feel more pressure to show savings from AI, and 51% expect it to raise the total cost of their technology solutions. CIO Dive highlighted another figure: only 4% have a plan to redesign roles around mixed teams of people and AI agents, and those who do are 2.4 times more likely to be among the best.
What this changes for your company
1. Define the value before you start. Every AI project should be born with a measurable goal: hours saved, shorter turnaround, errors avoided. Without that, there’s no way to know whether it worked, and the pilot becomes a never-ending project.
2. Agree on the stopping criterion on day one. “If customer service time hasn’t dropped by X% in 60 days, we shut it down.” Stopping a weak pilot isn’t failure. It frees budget and people for what works.
3. Track usage cost, not just implementation cost. With AI, spending varies with volume of use. A project that looked cheap as a pilot can weigh heavily once the whole company starts using it.
The survey confirms something I’ve seen for years: the problem is rarely the tool, it’s the lack of criteria for choosing, measuring and giving up. In 29 years of programming, the projects that delivered the most were the ones with the courage to cut what wasn’t delivering.
Originally published on IWA7 Insights, IWA7 Digital’s blog, which is now part of Escovando Bytes.


