Kanban column with a growing queue IS the bottleneck — no further analysis needed
When cards accumulate in one column of a kanban board creating a visible queue, treat that column as the current bottleneck without requiring separate cycle time analysis.
Why This Is a Rule
A kanban board is a physical model of your workflow, and queues are the visible signature of constraints. When cards pile up in one column — "Code Review" has 12 cards while "In Progress" has 3 and "Done" has 2 — the accumulation is the diagnostic. That column is processing cards slower than they arrive, creating a queue that grows over time.
Formal cycle time analysis can confirm and quantify this, but it's not needed for identification. The visual pattern (cards piling up) provides the same diagnostic information as a statistical analysis (throughput < arrival rate) with zero computational overhead. A kanban board that makes queues visible is itself a constraint-detection instrument.
This matters because teams often delay constraint intervention while gathering "enough data." "Let's run a cycle time analysis next sprint." Meanwhile, 15 cards accumulate in the review column and three engineers sit idle waiting. The visual queue is data. Act on it.
When This Fires
- Glancing at a kanban board and noticing one column with significantly more cards than others
- During team retrospectives when discussing flow and throughput
- When deciding where to focus process improvement efforts
- Any time you're trying to identify what's limiting team throughput
Common Failure Mode
Ignoring the visual signal because the column has "always looked like that." A chronically full column is a chronic bottleneck — the fact that it's familiar doesn't mean it's not constraining. The other failure mode: demanding quantitative analysis before acting. If 12 cards are sitting in "Code Review," you don't need a spreadsheet to confirm that code review is the bottleneck. You need to address the queue.
The Protocol
When you see cards accumulating in a kanban column: (1) Name the bottleneck: "[Column name] is our current constraint." (2) Ask: why is the arrival rate exceeding the processing rate? (Too few reviewers? Too large PRs? Reviews taking too long?) (3) Intervene: stop starting new work and redirect capacity to clear the queue. In kanban terms: "stop starting, start finishing." (4) Monitor: does the queue shrink? If yes, the intervention worked. If not, investigate deeper causes.