How would you use data analytics in the audit of revenue recognition, and how do you evaluate the results?
A core Audit & Assurance interview question — asked in analyst and associate interviews across IB, PE, and the Big 4.
THE SHORT ANSWER
Used well, analytics test the whole population rather than a sample. Concrete techniques: a three-way match of invoices to dispatch/delivery and cash; journal-entry analysis on revenue accounts filtered by risk criteria (manual entries, period-end, round numbers, unusual users); cut-off testing around period end; trend and ratio analysis (revenue vs receivables/deferred revenue, margin by product/customer); and duplicate/gap detection on invoice sequences. Crucially, analytics produce exceptions, not conclusions: I'd set expectation thresholds, then investigate every flagged item to root cause with corroborating evidence — an anomaly isn't a misstatement until explained. I'd also validate the completeness and reliability of the data extracted (reconcile the dataset to the ledger) before trusting any output, and document how the analytic addresses the specific revenue assertion (occurrence, cut-off, accuracy).
WHAT INTERVIEWERS LISTEN FOR
- ✓Full-population: three-way match, JE testing by risk criteria, cut-off, trend/ratio, duplicates/gaps
- ✓Validate data completeness/reliability (reconcile to ledger) first
- ✓Analytics yield exceptions to investigate to root cause, not conclusions
- ✓Map each analytic to the specific revenue assertion
COMMON MISTAKES
- ✗Treating an anomaly as a misstatement without investigation
- ✗Not validating the underlying data
- ✗Vague 'trend analysis' with no assertion linkage
Reading isn't the same as answering under pressure.
Interviewers don't hand you the model answer — you deliver yours on a clock. Practice this and 1,000+ questions with AI feedback on every answer.
RELATED QUESTIONS