Planning a Control Review
A practical framework for defining the purpose, boundaries, evidence, sampling approach, and responsibilities of a control review.
Define the Review Before Testing
Start with purpose, not paperwork
A control review asks whether a control is appropriately designed, has been put in place, and operates consistently enough to address its intended risk. So begin by identifying the risk or requirement the control should address and the criteria for successful performance. This keeps the review focused on what the control is meant to accomplish, rather than only on documents that happen to be available.
Define the review before testing
Set the scope and period: specify the business units, systems, transactions, requirements, and dates included. Identify the control and its owner, including who performs, reviews, and oversees it, and whether it is manual, automated, or a combination.
Next, define test procedures and evidence: what steps and evidence will let the reviewer assess whether the control is designed appropriately, implemented, and operating effectively? Then establish the population (the full set of events that could be tested), how items will be selected for the sample, and what will count as a deviation. Decide these boundaries before testing so findings are judged against stated criteria. The sample should fit the review’s objective and risk; a small or judgmental sample does not prove that every instance complied.
Finally, identify the roles and approval path: who will review, who will provide evidence, who should approve the plan, and who should receive the results.
Worked example: quarterly privileged-access reviews
Suppose the control is a quarterly review of privileged access. Define the population as all reviews due during the period, then identify the required reviewer and the criteria each review must meet. Select review events to inspect and specify the evidence and procedures needed to evaluate them. Before inspecting the sample, state what counts as a deviation: for example, a review was late, or a flagged account was not investigated. This plan makes the test’s boundaries and judgments explicit; results from the sample still do not establish that every review in the population complied.