Trace the waits

Follow one outcome

Choose a customer or operational outcome that is moving too slowly. Follow it from the first decision through design, build, testing, release, operation and measurement. Record every point where another team must act before learning can continue.

For each wait, capture the request, sending team, receiving team, queue time, rework and missing authority or skill. Do not assume every hand-off is waste. Safety assurance, regulation, scarce expertise and shared infrastructure can justify separation.

Conway’s paper argues that communication structures constrain system design. It supports inspecting the organisation as part of the delivery system; it does not supply a universal team design.

Failure mode: mapping the organisation chart instead of the path the work actually follows.

Worked example and transfer practice

All examples below are fictional teaching material. Use synthetic notes; no workplace or personal data is required or uploaded.

Measure the path rather than the chart. A synthetic maintenance request needs 1 day of diagnosis, waits 3 days for routine testing, needs 1 day of testing, then waits 2 days for an authorised safety review. Its observed elapsed path is 7 days; 5 are waits. The arithmetic describes this one item, not a stable average. Routine test capability may be movable; an independent safety decision may need to remain separate.

Try before reading the guide: Trace one fictional request with entry and finish definitions, active work, waits, reasons and owners. Name the evidence needed to decide whether a recurring hand-off can move.

Practice review guide

The map follows the outcome and preserves necessary assurance. Counting departmental activity alone can miss the waiting that determines elapsed time. Do not remove a safeguard simply because it adds a queue.