Audit sampling is how auditors test financial data without examining every single transaction. Instead of reviewing all 10,000 invoices in your purchase ledger, the auditor selects a representative sample, tests those, and draws conclusions about the whole population. It’s a practical necessity – testing everything would take months and cost a fortune.
ISA (UK) 530 governs how auditors design, select and evaluate audit samples. The key requirement: the sample must be large enough and representative enough to give the auditor reasonable assurance that the population isn’t materially misstated.
What are the sampling techniques for audit?
Auditors use two broad approaches to select their sample:
Statistical sampling
The auditor uses mathematical methods to select items and evaluate results. Each item in the population has a known probability of being selected. The main statistical sampling methods are:
- Random sampling. Every item has an equal chance of selection. The auditor uses a random number generator to pick items from the population. Simple and unbiased, but may not focus on higher-risk items.
- Systematic sampling. The auditor selects every nth item from the population (e.g., every 20th invoice). Quick to apply, but can be biased if the population has a recurring pattern.
- Monetary unit sampling (MUS). Also called probability-proportional-to-size sampling. Each pound in the population has an equal chance of selection, which means higher-value transactions are more likely to be picked. This is the most common statistical method for substantive testing because it naturally targets the biggest balances.
- Stratified sampling. The auditor divides the population into subgroups (strata) based on a characteristic like value, type or risk level, then samples from each stratum separately. A typical approach: test all items over materiality, sample items between performance materiality and materiality, and test a smaller sample of items below performance materiality.
Non-statistical sampling
The auditor uses professional judgement rather than statistical theory to select the sample. The selection methods include:
- Haphazard sampling. The auditor picks items without a structured method, but tries to avoid bias. Not the same as random sampling – there’s no mathematical basis for the selection.
- Judgemental sampling. The auditor deliberately selects specific items based on risk, value or other criteria. For example, selecting all journal entries posted on weekends, or all transactions with a new supplier. This is targeted rather than representative.
- Block sampling. The auditor selects a contiguous block of transactions – all invoices from March, for instance. Useful for testing periods but risky because it may miss issues that only arise at certain times of year.
How do auditors determine sample size?
The right sample size depends on several factors:
- Materiality. Lower materiality thresholds require larger samples because the auditor needs to detect smaller errors.
- Risk assessment. Higher assessed risk of material misstatement means more testing, which means bigger samples.
- Expected error rate. If the auditor expects errors in the population (based on prior year experience or interim testing), they’ll need a larger sample to quantify the total error.
- Confidence level. For statistical samples, the auditor sets a confidence level (typically 90-95%) that determines how many items to test.
- Population size. Larger populations don’t always need proportionally larger samples. A sample of 60 items can provide the same assurance for a population of 5,000 as for 50,000.
What happens when the auditor finds errors in the sample?
If the sample contains errors, the auditor must assess whether the population is likely to be materially misstated. For statistical samples, the auditor projects the error rate across the entire population. For non-statistical samples, the auditor uses judgement to estimate the total misstatement.
If the projected error is close to or exceeds materiality, the auditor will either:
- Extend the sample (test more items to get a better estimate)
- Ask management to investigate and correct the errors
- Conclude that the population contains a material misstatement and propose an audit adjustment
For your finance team, the practical implication is clear: errors in a small sample translate into large projected misstatements. Ten errors in a sample of 50 invoices doesn’t mean you have 10 wrong invoices – it suggests you might have hundreds.
How to prepare for sample-based audit testing
You can’t control which items the auditor selects, but you can make sure your records are in good shape:
- Keep source documents (invoices, contracts, approvals) organised and easy to retrieve
- Reconcile key account balances before the audit starts
- Fix known errors before year-end rather than leaving them for the auditor to find
- Maintain a complete audit trail from source document to ledger entry
If you’d like to understand what your auditor will be testing and how to prepare, get in touch with our team.
