Begin with Actual Work
Choose a recent order that completed normally and another that needed intervention. Follow each from purchase to delivery and any customer follow-up. Ask the people doing the work to show the records, queues, messages, and decisions involved.
Draw what happens now. Separate the documented process from the shortcuts that keep the company moving. A useful map includes waiting, duplicate entry, approval, correction, and escalation—not just the happy path.
Make Every Handoff Explicit
For each step, record the trigger, owner, input, system, decision, output, and receiving owner. Ask how the receiving person knows work is ready and how they know it is complete. Where a handoff relies on memory or a private message, make that dependency visible.
Look for unclear ownership between ecommerce, fulfillment, support, and finance. An exception can pass through several teams without anyone owning resolution. Name one owner for the outcome even when several people contribute.
Find the Cost of the Constraint
Measure a small, representative set of work. Record elapsed time, active effort, rework, exception frequency, and customer impact. Be careful with averages: a few severe delays can disappear inside an acceptable average cycle time.
Choose the constraint with meaningful business impact and a plausible path to improvement. A visually untidy spreadsheet may matter less than orders waiting for an approval that only the founder can give.
Change One Workflow and Test the Exceptions
Design the smallest change that can remove the constraint: a clearer standard, a decision threshold, a queue, a better input, or an integration. Define what should happen for missing data, duplicate events, cancellations, and unavailable owners.
Run the new process with its owner. Compare the result against the baseline, document what changed, and schedule a review. A workflow is not finished because a diagram is approved; it is useful when the team can operate it and recover when something goes wrong.