AI Has Outpaced How Companies Measure Developer Productivity, Report Finds

News

1. OpenAI Explores Legal Options Amid Tensions With Apple

2. AI Has Outpaced How Companies Measure Developer Productivity, Report Finds

3. AI Dispatch | May 8 - 14

4. du Launches Sovereign Industrial AI Platform for UAE Manufacturers

5. Huang Foundation Donates $108M CoreWeave Compute for AI Research

6. Aramco Deploys Industrial AI as Strait of Hormuz Crisis Threatens Fuel Supplies

Artificial intelligence is reshaping software engineering faster than many organizations can measure it. While AI coding assistants are helping developers ship code more quickly, much of the promised productivity is offset by work that shows up in no metrics.

Harness has released The State of Engineering Excellence 2026 report, showing that AI coding tools have transformed the day-to-day work of software developers faster than the industry’s measurement frameworks can keep up. 

Across the United States, the United Kingdom, India, France, and Germany, 700 engineering practitioners and managers were surveyed for the report. Nearly 89% of engineering leaders said developer productivity improved after adopting AI coding tools.

Yet the report also highlights a widening disconnect between how organizations measure productivity and how engineering work is actually changing.

According to the findings, 81% of engineering leaders said developers now spend more time reviewing AI-generated code. Nearly a third of developer time is now consumed by “invisible work” such as reviewing AI-generated code, fixing bugs, and context-switching between tools.

“It is not the work organizations are trying to accelerate; it is the overhead attached to the work,” the report noted.

The findings add to growing concerns that enterprises may be overestimating the efficiency gains from generative AI by relying on legacy software metrics focused on output volume and delivery cycles. While AI-generated code can shorten development timelines and increase throughput, engineering teams are increasingly being asked to take on new responsibilities in governance, quality assurance, and risk management.

“AI coding is the first technology shift in modern software that has changed not just what developers build, but how they spend their hours,” said Trevor Stuart, SVP and General Manager at Harness. “Cloud and the internet were infrastructure revolutions layered underneath the developer. AI is reshaping the developer’s job entirely, and the measurement frameworks that the industry has relied on for the past decade weren’t built for this new unit of work.”

The report argues that many organizations continue to evaluate engineering performance using systems designed before generative AI entered workflows. Metrics such as DevOps Research and Assessment (DORA), cycle time, and velocity remain widely used, but they often fail to capture other factors which they are not trained to measure.

That contradiction is reflected in the survey data. While 89% of leaders said their current metrics accurately reflect AI’s impact, 94% simultaneously acknowledged that critical factors — including developer fatigue, code quality, and technical debt — are missing from those frameworks. Only 6% said existing systems were fully capable of addressing the gap.

The biggest challenge organizations identified was not AI adoption itself, but visibility into its real impact. Respondents cited measuring true productivity gains, maintaining code quality, and proving return on investment to leadership as the top concerns.

The pressure on developers is also intensifying. Harness’ findings suggest developers are increasingly uneasy about how AI-driven productivity data may be used. More than half of respondents feared individual performance evaluations based on AI metrics.

The perception gap between leadership and practitioners was particularly stark. Managers were nearly four times more likely than frontline developers to say they had no concerns about how AI productivity data would be used.

To adapt, the report recommends that engineering organizations begin measuring the new units of work created by AI-assisted development. That includes tracking validation time, AI-agent accuracy, code acceptance rates, burnout indicators, and cognitive load alongside traditional delivery metrics.

It also urges companies to separate improvement-focused analytics from employee performance evaluations and involve developers directly in defining how productivity should be measured in AI-enabled workplaces.

“Engineering leaders are being asked to make multi-year AI investment decisions using dashboards built for a different era of software development,” Stuart said.