SimplyCPA
CPA/AUD/IT Controls in Audit

IT Controls in Audit

General and application controls, auditing around vs. through the computer, and using data analytics.

Medium 50 minArea III: Performing Further Procedures and Obtaining Evidence

General controls vs. application controls

TypeScopeExamples
IT general controls (ITGCs)The whole IT environmentAccess security, change management, program development, IT operations/backup
Application controlsA single process or systemInput edit checks, validity tests, three-way match, automated approval limits

IMPORTANT: Application controls can only be relied on if the ITGCs supporting them are effective. If access or change management controls are weak, an automated control that looks perfectly designed cannot be relied on — anyone could have altered the program or the data.

Auditing around vs. through the computer

  • Around the computer — trace inputs to outputs, ignoring processing. Only appropriate for simple systems with clear audit trails and low risk.
  • Through the computer — test the processing itself, using techniques such as test data (auditor's fictitious transactions run through the client's system), integrated test facility (a dummy entity within live processing), and parallel simulation (auditor's own program reprocesses client data).

Segregation of duties in IT

Classic separations: systems development from operations, and both from data control/security administration. A programmer with production access is the standard exam red flag — it enables unauthorized, untested code to reach live financial data.

Data analytics (ADAs)

Audit data analytics can test 100% of a population rather than a sample — for example, recomputing every sales invoice, matching every payment to an approved vendor, or identifying all journal entries posted on weekends or by unusual users. Analytics identify items warranting further investigation; they don't replace the auditor's judgment about whether an exception is a misstatement.

EXAM TIP: Testing 100% of transactions eliminates sampling risk but not nonsampling risk — the data could be incomplete or the auditor could apply the wrong criteria.