Audit sampling means selecting part of a population of transactions, records or items to draw a conclusion about the whole. It works only if the sample is representative.
Why sampling matters
Checking every transaction is impractical, especially in cloud and ERP systems. Sampling saves time and cost, improves efficiency and still provides enough evidence for an opinion. It is used to detect errors or fraud, evaluate internal controls, test the accuracy of the statements and check compliance.
Two categories
The methods are either statistical or non-statistical (judgemental). The choice depends on the risk, the size of the population, the control environment and the audit procedure.
Statistical methods
These use random selection and give a measurable sampling risk.
- Random: every item has an equal chance. It suits large populations without patterns.
- Systematic: every nth item, for example every 50th invoice. It is unsuitable where the data has cyclical patterns.
- Stratified: items are grouped by value, risk, customer category, department or location.
- Monetary unit sampling: larger amounts are more likely to be selected. It is common for receivables, inventory, revenue and purchases and is strong for finding overstatements.
- Probability proportional to size: similar, with higher-value items sampled more often.
Non-statistical methods
These rely on professional judgement, and the reasoning must be documented.
- Haphazard: no structured technique and no intended bias. It suits low-risk populations or strong controls.
- Block: selects a group, such as one month's invoices. It is easy but may not be representative.
- Judgemental (selective): targets high-value, unusual or risky items, such as year-end adjustments, large payments, new vendors and manual journal entries.
How auditors choose
The factors are materiality, the risk of misstatement, the control environment, the variability of the population, the audit objective, the degree of automation and past errors or fraud. High-risk audits generally call for statistical sampling, and simpler low-risk audits may use judgemental sampling. Combinations are common.
The eight steps
1. Define the population. 2. Determine the sample size, based on risk, confidence level and variability. 3. Select the method. 4. Choose the items by software, randomisation or judgement. 5. Perform the tests and document the results. 6. Evaluate whether errors are isolated or widespread. 7. Project the findings to the population and assess materiality. 8. Conclude and document, which forms part of the audit opinion.
Advantages
Faster work, lower cost, higher efficiency, testing of large populations, reliable conclusions, better fraud detection and better use of resources.
Common mistakes
- Samples that are too small
- Judgemental sampling where statistical sampling is needed
- Undocumented selection criteria
- Ignoring high-risk items
- Failing to project the errors
- Relying on a single method
Sampling and data analytics
Analytics can review whole populations, flag outliers and predict high-risk transactions. Sampling is still needed for reliable audit evidence, and a hybrid approach works best.
Good sampling leads to good audits. See our audit services and our article on risk assessment techniques.
Key takeaways
- Sampling lets auditors reach a conclusion on a whole population without testing every item.
- Statistical methods give measurable risk; judgemental methods need documented reasoning.
- High-risk audits lean on statistical sampling.
Frequently asked questions
What is audit sampling?
Testing a selected part of a population to draw a conclusion on the whole.
What are the two categories?
Statistical and non-statistical (judgemental).
Does analytics replace sampling?
Not entirely. A hybrid approach is recommended.
Talk to a chartered accountant
Questions about your own situation? Reach us directly and we will reply with clear next steps.
This article is general information and not advice for your specific circumstances. Rules and thresholds change, so confirm the current position before you act.
