COMPARISON

Continuous Monitoring vs Periodic Sampling

Sampling is statistically sound and structurally late. By the time an exception is found, the money has usually gone.

Sample-based testing exists for a good reason: examining every transaction by hand was never possible, and a well-designed sample supports a defensible conclusion about a population. That logic is sound and remains the basis of most audit methodology.

What has changed is the cost of testing the whole population. Once that cost falls, the two real limitations of sampling become harder to accept — it cannot find the specific items you most want to find, and it reports them long after they occurred.

SIDE BY SIDE

How the two approaches differ

Dimension	Periodic sampling	foretale.ai
DimensionPeriodic samplingforetale.ai Population tested	A statistically selected sample, commonly tens to low hundreds of items.	Every transaction in the period.
Purpose	Concluding on control effectiveness across a population.	Identifying the specific exceptions that occurred.
Timing	After period end, often weeks or months later.	On an ongoing schedule, while correction and recovery are still possible.
Rare high-value items	Likely to be missed unless the sample is stratified toward them.	Found, because the whole population is tested.
Recovery potential	Often expired — vendor relationships and periods have closed.	Frequently intact, which is where most measurable value comes from.
Control degradation	Visible only as a difference between periodic test results.	Visible as exception-rate trend by process, entity, and approver.
Effort profile	Recurring manual effort each cycle.	Configuration effort up front, low marginal effort per cycle.

Periodic sampling remains appropriate when

Methodology or regulation requires a statistical conclusion about a population.
The control being tested is judgemental and cannot be expressed as a data test.
Evidence exists only on paper or outside accessible systems.
The population is small enough that full testing offers no advantage.

Continuous monitoring is appropriate when

The control can be expressed as a test against transaction data.
Exceptions are individually recoverable if found in time.
Volumes are high enough that a sample cannot represent rare events.
You need to show control performance over time rather than at a point.

Frequently asked questions

Does continuous monitoring replace audit sampling?

No. They answer different questions and most programmes run both. Sampling supports a statistical conclusion about control effectiveness, which is often a methodology requirement. Continuous monitoring identifies the specific exceptions that occurred. Where continuous monitoring covers a control well, sample sizes for that control can often be reduced and better targeted.

How often should continuous monitoring actually run?

It depends on whether the exception is recoverable and for how long. Payroll analytics are most valuable per pay run, accounts payable analytics before payment runs close, and journal entry analytics around period end. Monthly is a reasonable default where recovery windows are longer.

Won't testing everything produce too many exceptions?

It will if exceptions are reported flat. Usable output requires ranking by risk and materiality, and it requires data integrity issues to be resolved first, since those generate most false positives. The goal is a prioritised population an investigator can work through, not every technical breach.

What is the main measurable benefit?

Timing. The same exception found before a payment run rather than after year-end is often the difference between prevention and a write-off.

NEXT STEP

See it on your data.

Request a focused walkthrough—usually within one business day.