Cycle time and review time are not the same metric

The distinction between cycle time and review time is crucial when discussing the impact of AI reviewers on PR (Pull Request) processes, as highlighted by Atlassian's Rovo Dev AI reviewer. Atlassian reported that Rovo reduced PR cycle time by up to 45% internally and 32% for customers, according to a source dated 2026-01-29. Cycle time refers to the entire PR lifecycle, from the moment a request is opened until it is merged.

Conversely, review time specifically measures the duration a human spends reviewing a pull request, which is distinct from cycle time.

Salesforce, in response to growing code volume and the frequent occurrence of PRs with over 1,000 changed lines, emphasized the importance of addressing review latency and setting latency goals, as noted in a source published on 2026-01-29. It is essential to recognize that neither cycle time reduction nor review time improvement necessarily equates to a faster or more accurate human review of the diff (difference).

The core of the issue lies in the method gap, where AI can enforce acceptance criteria and coding standards prior to human review, thus shortening the cycle time.

Automating aspects of the review process can lead to improvements in one metric without necessarily enhancing the other. For instance, an AI reviewer can ensure that code adheres to specified standards before it reaches a human reviewer, thereby reducing cycle time. Alternatively, automating baseline checks can alleviate the workload on human reviewers, potentially decreasing review latency by shortening the queue of human-reviewed PRs.

While both approaches lower a running clock, they do not guarantee that the human component of the review process becomes faster or more accurate.

A real-world example illustrating this concept can be found in a related post discussing the Rovo run; when review time plateaued, reviewers began taking shorter, less thorough readings of the diffs. This scenario shows that the two clocks—cycle time and review time—can operate independently and in opposite directions. When teams report percentage improvements, it is vital to clarify which clock they are referring to, as the implications for team workflows and productivity can vary significantly.

In summary, while AI reviewers like Atlassian's Rovo Dev AI can effectively cut PR cycle time through automated enforcement of code standards, their impact on review time, which measures human attention to the diff, is less impactful. Claims about improvements in either metric without specifying which one were measured lack validation. As of 2026-09-20, it remains a best practice to ask teams which clock they are highlighting to ensure the claimed benefits are substantiated.