Why did the hours-saved number fall?

Because the work changed shape. In 2025 the agent wrote a function and the saved hour was visible. In 2026 the agent writes the function and opens the PR, and the developer spends the hour reviewing and re-prompting, which the same survey counts as managing AI. The question kept asking about the place the hours left.

Is shipped a good enough replacement?

It is better than hours saved and Atlassian deserves credit for publishing counts. But shipped is a count of output. Landed, the behavior that changed for a customer, is the column that says whether the decision to build it was right.

Where does the column fit in Atlassian's tasks, skills, roles sequence?

Before tasks. Once shipped and landed are on the dashboard, widening contribution is safe, because the 100 changes will carry a second number next to them that is not 100.

Hours Saved Is the Wrong AI Metric. Count What Landed.

Should hours saved be the AI adoption metric?

Answered by Falk Gottlob · Reviewed Nov 11, 2026 · 1 min read

THE SHORT ANSWER

No. Atlassian's 2026 State of Developer Experience research, as Avani Prabhakar reported it on November 10, 2026, had 68% of developers saying AI saved them 10 or more hours a week in 2025 and 26% in 2026, with 49% now spending four or more hours a week managing AI. The hours did not disappear; they moved to managing the agent, and a self-report about time cannot see that. Atlassian's three PDLC lessons count better things, fixes prepared for 52% of vulnerabilities, two features in 2 and 4 developer weeks against estimates of 30 and 40, and 100 production changes from 50 PMs and designers, but all three count what shipped and none counts what landed. The metric is two columns per shipped item: shipped, with the reviewer's name, and landed, with the behavior that changed for a customer. Of 39 agents I appointed, 13 produced output nothing outside the fleet read, and I only knew because I counted.

No. Hours saved is a self-report about time, and the time moved to managing the agent, which is why Atlassian's own figure fell by two thirds in a year. Count shipped with a reviewer's name and landed with the behavior that moved.

The three lessons, their counts, and the missing column are in Atlassian's Developers Went From 68% to 26% on Hours Saved. Kill the Hours. Count What Landed.. The two columns come from Reddit Has 700 Prototypes on a Dashboard. The Column It Needs Is Shipped., and the same company's created-versus-closed gap is in Atlassian's Agents Write 50% More in Jira. Count What the Humans Closed..

Should hours saved be the AI adoption metric?

SOURCES

THE LONG VERSION

RELATED ANSWERS