# Kanban Board Audit

> One question decides most of this: **are there WIP limits, and do they ever stop anyone?**
> Without that, you have visibility and none of the flow benefit.

**Team / board:** _______________  **Date:** _______  **Audited by:** _______________

## 1. Does the board show the real workflow?

- [ ] Columns match how work **actually** moves, not an idealised process
- [ ] **Waiting states are columns** — waiting for review, for test environment, for another team,
      for a decision
- [ ] Steps outside the team's control are shown, even though the team cannot fix them
- [ ] A blocked item is visibly blocked, with the reason and who is chasing

🔴 Delay lives in the queues, not in the work. A board that omits the waiting states measures the
small part of the process you control and leaves everyone puzzled about the lead time.

**Columns currently on the board:** _______________
**Waiting states NOT shown:** _______________

## 2. WIP limits — the mechanism

- [ ] A limit exists per column (or across the board)
- [ ] Limits are **visible on the board**
- [ ] The limit is **reached regularly** — if never, it is doing nothing
- [ ] When it is reached, people help finish rather than starting something new
- [ ] Limits were adjusted based on evidence, not set once and forgotten

| Column | Limit | Times reached last month | Times exceeded |
|---|---|---|---|
| | | | |

**If "times exceeded" is high, the limit is decorative.** A limit that is routinely broken is a
suggestion, and the board has quietly become a push system with nicer columns.

## 3. Explicit policies

- [ ] "Ready" is defined — what must be true before an item can be pulled
- [ ] "Done" is defined **per column**
- [ ] Prioritisation rule written down
- [ ] Expedite policy written down: what qualifies, how many at once, what it displaces
- [ ] Policies are visible, not folklore

Unwritten policies get applied inconsistently and argued about weekly.

## 4. Expedites

- [ ] Expedite lane exists and is limited (usually to one)
- [ ] Used rarely — count last month: ____
- [ ] Each expedite has a recorded justification

If more than a small fraction of work is expedited, priority is being set by whoever asks loudest,
and the ordinary queue has stopped moving.

## 5. Is the board true?

- [ ] Every item in progress has actually been worked on this week
- [ ] Nothing has sat in one column for more than ____ days without a flag
- [ ] Work being done is on the board — no invisible side work
- [ ] Items are updated as they move, not in a weekly tidy-up

**Oldest item in progress: ____ days.** **What is stopping it?** _______________

Those two questions do most of the work in a weekly review.

## 6. Metrics

- [ ] **Cycle time** measured (start of work → done)
- [ ] **Lead time** measured (request → delivered) — usually much longer than the team expects
- [ ] Throughput tracked (items per week)
- [ ] Forecasts use the **distribution** of past cycle times, not an average
- [ ] Metrics describe the **system**, never individuals

| | Value |
|---|---|
| Median cycle time | |
| 85th percentile cycle time | |
| Throughput per week | |
| Current WIP | |
| Flow efficiency (working time ÷ elapsed) | |

Flow efficiency is frequently under 20% on first measurement. That is uncomfortable and useful: it
says the problem is queues, not effort.

Forecast as "85% of items finish within ___ days" rather than "average ___ days" — the average
hides the tail that causes every escalation.

## 7. Improvement

- [ ] Someone looks at where work accumulates, regularly
- [ ] A constraint has been fixed in the last quarter — which: ____________
- [ ] The team has a scheduled improvement conversation (kanban does not mandate one; teams
      without one usually stop improving)

## Verdict

| | |
|---|---|
| WIP limits exist | yes / no |
| WIP limits ever stop anyone | yes / no |
| Waiting states visible | yes / no |
| Policies written down | yes / no |
| Cycle time measured | yes / no |

**"No" to the first two means this is a to-do list on a wall.** That still has some value, and it
is not kanban — and the flow improvement everyone is expecting will not arrive.