ROI measurement gap
Most teams feel faster. Fewer can defend the number.
AI tooling is now a line item in every engineering budget. This survey asks engineering leaders how they are measuring AI return: productivity, quality, spend, and confidence in their metrics.
Complete the survey and opt in to receive the benchmark when it's published.
What leaders are unsure about
The next wave of AI adoption is a measurement problem.
01
“It's really hard to get from the day-to-day use of these AI tools to that bottom-line number.”
02
“Using tokens is not a measure of productivity per se. It's hard to measure productivity.”
03
“It's difficult to measure the impact of something when you're doing something you haven't done before.”
04
“What we can measure is roughly 20 percent productivity gain-ish, and the expectation was at least double that.”
05
“Some power users will see exponential gain like 2 to 3X, but it's still in pockets. And then there'll be a long tail of people still using AI in a very basic way. The organization as a whole caps out at 20 to 30 percent improvement.”
From AI-native adoption to ROI
The 2026 report showed adoption. This survey asks what came back.
Our State of AI-Native Engineering report found teams moving quickly into AI-assisted development while carrying new questions about trust, comprehension, and role change. The next benchmark focuses on the executive question behind that shift: how are leaders measuring productivity and ROI now that AI is part of the SDLC?
Read the State of AI-Native Engineering 2026 report219
engineering leaders surveyed
48%
of all code is now AI-generated
63%
say engineers are voicing skill-relevance concerns
01AI is already changing how teams write and review software.
02Leaders still need shared metrics for productivity, quality, and spend.
03The ROI benchmark will help compare what teams measure against what they feel.