Ideas you can
put to work.

Practical field guides

Choose a reading path from a recurring work problem

Use the catalogue to resolve one practical question rather than reading every guide before acting. This route begins with a fictional problem: “Requests reach review, then nobody can say what they are waiting for.” It links current field guides without rewriting Richard Durnall’s historical articles or treating a reading list as an implemented improvement.

1. Find the boundary: read Map one piece of work. Produce one trace from request to useful delivery, with known states, dates and unknown intervals. If the team cannot reconstruct the item, stop and obtain those observations; do not invent a bottleneck.
2. Make a decision at capacity: read Set a WIP limit. Write what “started” and “finished” mean and what happens when the agreed review boundary is full. Keep blocked items inside the count where they remain within that boundary. A chosen cap is a trial agreement, not an optimal number discovered by the article.
3. Test the explanation: read An A3 for a recurring problem. Separate the observation from the proposed cause, choose a reversible countermeasure and agree what would count against it. Leave the result unresolved until the trial runs.

Completed route card — fictional: the trace identifies an item waiting for an acceptance decision, but does not establish why. The team’s working hypothesis is that review requests lack an explicit decision owner. Its proposed test is to name that owner before the next agreed set of comparable handovers enters review, while keeping the existing capacity policy visible. If owners are named but the decisions still stall, the hypothesis is incomplete; inspect authority, availability or unclear criteria before repeating the same fix.

The next reading follows the unresolved question. If the item history is missing, return to the trace. If capacity exceptions hide unfinished work, return to the WIP agreement. If the evidence contradicts the proposed cause, revise the A3. The route is finished when it yields a reviewable next action, not when every linked page has been opened.

Route card: recurring problem ___; chosen item and boundary ___; observation still missing ___; first guide and output ___; policy decision needed ___; hypothesis ___; proposed test ___; evidence that would contradict it ___; owner/review date ___; next reading or stop ___. No business outcome is inferred from completing this card.

The Kanban Guide, May 2025: explicit workflow boundaries and improvement