Skip to main content

Bottleneck Insights

Find where work slows down across your delivery lifecycle and target the root cause.

Bottleneck Insights highlights where work slows down across your delivery lifecycle. It helps teams move from symptom reporting to root-cause targeting.


​

Where to find it


​

In the sidebar, under Actions Hub → Insights. It used to be its own entry; it now sits with the other things that are waiting on a decision — Alerts, the rules that notify you when a metric crosses a threshold, and Agents.


​

Inside the page, the detected bottlenecks are on the Inbox tab.


​

How insights are classified


​

Every insight carries a type, and the Inbox filters by it:


​

  • Anomalies — a metric deviated negatively from what is normal for you. Cycle time spiked far above the recent average.

  • Early Warnings — individual items, not metrics: a PR or task at risk of dragging a metric down while there is still time to act. Fresh PRs that have been open too long.

  • AI overspend — AI consumption above what your organization normally spends, or consumption that leaves no measurable return. Unattended agent runs that went off the scale in cost.

  • Agent — insights written by the Research Agent rather than detected by a rule. The filter lets you focus the list on these.


​

Each one also carries a risk level: High, Medium or Low.


​

What it typically surfaces


​

  • Review pickup delays — pull requests that wait too long for the first review.

  • Long waiting states — items sitting in non-active workflow states (blocked, awaiting QA, awaiting deploy).

  • Excess WIP accumulation — too many items in flight versus throughput, slowing everything.

  • Team-specific throughput constraints — one team's flow is capped while others have headroom.


​

How teams use it


​

  1. Isolate the stage where delay concentrates — review, waiting, or deploy.

  2. Segment by team, project, or workflow type to localize the problem.

  3. Validate whether mapping or process drift is affecting results before assuming a behavioral cause.

  4. Prioritize one operational fix and re-check the trend in the following review.


​

Best for


​

  • Engineering managers running continuous improvement.

  • Delivery operations identifying systemic friction.

  • Cross-team optimization initiatives where you need to know where to start.


​

Related articles


​

  • Cycle Time

  • Work in Progress (WIP)

  • Waiting Time

  • Review latency

  • What is Context?

Did this answer your question?