Reconcile overlapping document edits from a common base
Merge parallel proposals while checking that deletions, renamed roles, and independent additions preserve the agreed intent.
Kiguri Blog
Prioritize review tests with residual uncertainty visible
Choose checks by the failures they can reveal, then record what a limited review leaves untested.
Test a support knowledge tree for complete conditions
Trace fictional cases through every branch to find loops, missing unknown paths, and advice that outruns its evidence.
Observe a prototype task attempt without coaching the route
Log an attempted task in order, separate assistance from independent completion, and tie follow-up questions to observed steps.
Negotiate a session schedule when constraints conflict
Test a fixed-duration session against hard availability limits, then seek an explicit change when no interval is feasible.
Map who a process change affects before announcing it
Compare old and proposed obligations, notices, and decision rights so a change does not silently assign work to absent people.
Rehearse a disagreement without deciding the underlying issue
Practice accurate paraphrases and useful questions while keeping the underlying project choice explicitly unresolved.
Review an alternate format for information equivalence
Check that order, relationships, and exceptions survive when a visual instruction becomes a text version.
Organize a small knowledge collection with tested categories
Sort representative documents, separate topics from status, and test category rules on material nobody used to design them.
Check what a chart actually measures before discussing it
Inspect denominators, time windows, units, and visual encoding before turning a chart into a claim about change.
Pilot survey options before asking people to answer
Test whether fictional respondents can answer honestly and reach the right follow-up before distributing a survey.
Review fictional sample data against a shared field contract
Distinguish inconsistent rows from missing rules before correcting a synthetic planning dataset.