1
Complete the process header before anything else
Enter the process title, a unique process ID, version number (start at 1.0), today's date as the effective date, and the name and title of the process owner. This header is the document's identity β every field matters for version control and audit trails.
π‘ Use a consistent process ID format across your organization (e.g., OPS-001, FIN-002) so documents are searchable and cross-referenceable in your document management system.
2
Write the purpose and scope statements
Draft a 2β4 sentence purpose that explains why this process exists and what outcome it reliably produces. Then define the scope: where the process starts, where it ends, which roles it applies to, and what is explicitly excluded.
π‘ Write the out-of-scope statement before you finalize the scope β listing what the process does not cover forces clarity about where adjacent processes begin.
3
Assign roles using a RACI table
List every role involved in the process and assign R, A, C, or I for each major step. Confirm there is exactly one Accountable role per step β the person who owns the outcome, not just the task.
π‘ If you find yourself assigning A to a team rather than a named role, the process lacks a real owner. Name the role title, not the department.
4
List inputs, triggers, and required formats
Document every input the process requires before it can begin β name the specific document or data, its required state (e.g., approved, not draft), and the system or person it comes from. Then state the trigger condition that starts the process.
π‘ Walk through a recent real execution of the process and note every file or piece of information you actually used β this surfaces hidden inputs that never make it into documentation.
5
Write the step-by-step procedure
Number each action sequentially. Each step should be a single discrete action β one person, one system, one decision. Include decision branches (if/then logic) and note the role responsible for each step.
π‘ Shadow someone performing the process in real time rather than asking them to describe it from memory. Live observation captures the four or five informal steps that practitioners do automatically but never mention.
6
Define outputs and controls at each stage
For each major step or handoff, name the output produced and where it goes. For each control point, name the check, who performs it, and what the escalation path is if the check fails.
π‘ If a control has no named owner and no escalation path, it is not a real control β either assign it properly or remove it and acknowledge the gap.
7
Set KPIs with targets, owners, and measurement cadence
Add 2β4 metrics that indicate whether the process is working. For each, state the target value, who measures it, and how often they report it. Tie at least one metric to cycle time and one to quality or error rate.
π‘ Avoid vanity metrics. 'Number of processes documented' is not a KPI for the process itself β 'on-time completion rate' and 'defect rate per 100 executions' are.
8
Complete the version history and set the next review date
Record this as Version 1.0, enter today's date and your name, and note 'Initial release.' Set a review date no more than 12 months out, and add at least one trigger condition β such as 'review immediately if the supporting system changes.'
π‘ Store the finalized document in a location where all relevant staff can find it and where access is logged β shared drives with version history, not email attachments.