Judgemental versus random samples in expense testing
Random samples look impartial; judgemental samples find known pressure points faster. Most Hong Kong compliance checks need both.
Sampling design is the quiet decision that shapes every expense policy compliance check. Controllers sometimes ask for a pure random sample so the review “cannot be accused of targeting.” That request is understandable — and often incomplete.
When random helps
A random layer across the full population supports statements about overall exception rates. It is useful when the audit committee wants a broad temperature check and no single category is already under suspicion.
When judgemental is stronger
If entertainment, overseas lodging, or a new subsidiary is the known risk, a judgemental overweight finds more actionable exceptions per testing hour. Cloudapi Pro usually pairs a judgemental core with a smaller random overlay so unexpected categories still appear.
Document the choice
Whatever mix you pick, write it into the scope letter. After findings land, nobody should invent a story about why certain claims were selected. Clarity at kick-off protects both the reviewer and the finance team.
For how this sits inside a full engagement, see our engagements page.