Use this companion exercise to document one recurring workflow so another team member can carry it out and recognize when to ask for help. It is an original implementation worksheet for the documentation topic, not an excerpt or a claim about a specific framework in the book. Revised November 21, 2026.
Start with one observable handoff
Choose a task that happens often enough to test, has a clear beginning and end, and regularly sends someone back to the owner for clarification. “Run the business” is too broad. “Prepare a completed project for the billing team” is bounded enough to examine. Write down the trigger, the person who receives the work and the evidence that the handoff is complete.
Observe a real performance of the task before drafting the procedure. Ask the person doing the work to explain each decision, including what they check and when they stop. Capture the current process separately from proposed improvements. Otherwise, a polished document can describe a workflow the team has never actually used.
Write the procedure around decisions
- Purpose and boundary: state what this procedure accomplishes and what remains outside its scope.
- Trigger and inputs: identify the event that starts the task and the records that must be available.
- Responsible role: name the role doing the work and the role authorized to resolve exceptions.
- Ordered actions: give each step an observable action, its required input and its expected output.
- Stop conditions: say what happens when evidence is missing, instructions conflict or the person lacks authority to continue.
- Completion evidence: define the record or status that confirms the task is finished.
- Maintenance: identify the document owner, version, last test date and event that triggers another review.
Prefer “check that the approval record names the project and authorized reviewer” to “ensure everything is approved.” The first instruction gives a new operator something to inspect. Avoid embedding passwords or unnecessary customer information in screenshots; refer to the approved system and the authorized access role.
Fictional example: completed project handoff
Imagine a fictional service business that sends completed project records to a separate billing team. The handoff begins when the project lead marks delivery complete. Required inputs are the project identifier, approved scope, completion evidence and any approved changes. This example does not authorize an invoice or replace the business's own approval policy.
The operator first checks that each required record refers to the same project. Next, the operator checks whether the delivered work matches the approved scope and whether changes have a recorded decision. If scope or approval is missing, the operator marks the handoff “needs review,” names the missing record and routes it to the project lead. Guessing the missing scope is not an acceptable completion step.
When the records are complete, the operator sends a link to the handoff packet through the approved workflow and records the receiving role. Completion means the billing team has acknowledged a usable packet, not merely that someone sent a message. A returned packet is an exception to investigate, not a completed handoff to count again.
Test with someone who did not write it
Ask a second authorized team member to walk through the procedure using a safe example. The writer should observe rather than explain missing steps. Record every clarification, missing input and point where the operator had to infer a decision. Test both an ordinary case and an exception, such as a project with an unapproved change.
Define acceptance before the test: the operator can identify required inputs, complete the ordinary handoff, stop the exception case correctly and produce the specified completion record. Speed alone is insufficient. A faster handoff that silently omits approval evidence has failed the procedure's purpose.
Keep the working version usable
Store the approved procedure where the team starts the work, give it a named owner and link it from the relevant task or checklist. Retain prior versions according to the organization's records policy. Update the procedure when tools, responsibilities or approval rules change, then repeat the affected test cases. A review date is a reminder to check usefulness, not evidence that someone actually tested the workflow.
For a broader look at work that still depends on the owner, use the founder bottleneck audit. Return to the Built to Run overview to explore the book and other chapter topics. If the team needs help turning its own workflow into an operating process, the related consulting scope below provides a starting point for defining that engagement.
Turn documentation into a working operating process
For implementation support from the author’s consulting firm, review Elixir’s operations consulting scope. Bring one recurring workflow, its owner and the errors or delays you want to reduce; agree what needs to be documented and how the team will maintain it.