Use a realistic task with an observable result
Choose tests that match the person's actual responsibility. A seller should demonstrate commercial judgment; an administrator should demonstrate access and recovery; a researcher should demonstrate evidence and disqualification. Do not use one generic quiz for every role.
Include a normal case and an exception. Define acceptance in advance and record what the operator did, what evidence they used and where help was required. A failed test is useful while there is time to coach or repair the system. Avoid declaring independence because the person can follow a memorized example.
| Decision | Evidence to use | What changes next |
|---|---|---|
| Research | New account with source, fit and rejection reasoning | Operator can reproduce the ICP rather than copy a list |
| Selling | Ambiguous buyer case with next action and escalation | Operator distinguishes evidence from optimism |
| Administration | Controlled error with recovery guide | Operator restores continuity without unsafe improvisation |
Work through the decision
Illustrative assessment: the seller receives a case with an enthusiastic user but no active buying project. They should record interest, clarify the decision and avoid inventing pipeline. The administrator receives a failed routing test and should restore the record without duplicate outreach.
Review the artifacts against the agreed rules. If the seller needs missing product context, update the playbook; if the administrator cannot recover because the workflow is agency-owned, fix ownership. Independence depends on both people and system design.
The operator knows the example but not the decision
A familiar demonstration can conceal dependence. Use a new account and an exception that requires judgment. Ask why the action is appropriate and what would change it, not only whether the person clicked the right buttons.
A concrete next step
Schedule one independent task for each future owner. Review the resulting record and reasoning, then repeat only where a specific unresolved gap justifies it.
Sources and research notes
- GitLab customer-ready shadow programCompany operating handbook
- Attio: Back up workspace dataProduct documentation
Primary sources reviewed October 6, 2026. The operating recommendations and worked scenarios are Daavid’s analysis. Illustrative numbers are assumptions, not measured client results. Company marks identify sources and prior experience; they do not imply a customer relationship or endorsement.
Schematic