One of the first questions on any fixed-asset audit is also one of the hardest: how much do we actually have to count? A register can hold thousands of line items spread across dozens of sites, and verifying every single one is rarely possible within an audit timetable. Fixed-asset audit sampling is how auditors answer that question — testing a defensible subset of the population, examining the highest-value items in full, and drawing a conclusion about the register as a whole.
The goal is not to count the most assets possible. It is to gather sufficient appropriate audit evidence about the existence and completeness of fixed assets while keeping sampling risk — the risk that the sample is not representative — at an acceptable level. This article walks through how that sample is sized, which assets are never sampled, and how the conclusion is documented.
How do you determine fixed asset audit sample size?
Auditors set sample size from the assessed risk of material misstatement, materiality (tolerable misstatement), the desired confidence level, the expected error in the population, and population size. Higher risk or confidence and lower tolerable error push the sample up; individually significant, high-value assets are tested 100% as key items rather than sampled.
Why Auditors Sample
Auditing standards do not require an auditor to examine 100% of a population. Audit sampling exists precisely because a full census is usually neither efficient nor necessary to support a reasonable conclusion. For a fixed-asset population spanning manufacturing lines, warehouses, vehicles, and IT equipment across many locations, a complete physical verification of every item would consume time and cost out of proportion to the assurance it adds.
Sampling lets the auditor apply procedures to fewer than 100% of the items in a population while still drawing a conclusion about the entire balance — provided the selected items give a reasonable basis for that conclusion. The trade-off is sampling risk: the chance that the auditor’s conclusion based on the sample differs from the conclusion they would reach by testing every item. The whole discipline of sample design is about controlling that risk, not eliminating it.
It is worth separating two things that are easy to conflate. An organization performing a wall-to-wall physical verification of its own register may well count everything — that is a management exercise to clean and prove the register. The external auditor, by contrast, usually samples from that verified register as part of forming an opinion. The cleaner and more complete management’s own count, the more efficient the auditor’s sample can be.
Statistical vs. Judgmental Sampling
Auditing standards recognize two broad approaches to selecting a sample. Both are legitimate; the choice depends on the objective, the population, and the degree to which the auditor wants to quantify the result.
- Statistical sampling uses random (or otherwise probability-based) selection together with probability theory to evaluate the results. Its defining feature is that it lets the auditor measure and control sampling risk and mathematically project the observed error rate to the whole population at a stated confidence level. Methods such as random selection, systematic selection with a random start, and monetary-unit (dollar-value) sampling fall in this category.
- Judgmental (non-statistical) sampling relies on the auditor’s professional judgment to decide which items to select and how many. It is faster to design and can be highly effective when directed at the items that matter most, but it does not produce a statistically projectable conclusion or a measured confidence level. The selection still has to be representative of the population to support a valid conclusion.
A common misconception is that statistical sampling is inherently "better." It is not — it is simply more measurable. Standards permit either approach, and many fixed-asset engagements blend them: full testing of key items, a judgmental selection of higher-risk categories, and a statistical sample of the remaining homogeneous population. What matters is that the method chosen gives a reasonable basis for the conclusion and is documented.
Need a verified register your auditors can sample with confidence?
The efficiency of an audit sample depends on the quality of the count behind it. CPCON physically verifies, reconciles, and tags fixed assets against your ledger — across 30+ countries — so auditors start from defensible existence and completeness evidence.
Factors That Drive Sample Size
Sample size is not a single number you look up — it is the output of several inputs the auditor assesses for the specific population. The factors below move sample size up or down. There is no fixed percentage of assets that "must" be counted; the same population can warrant very different samples depending on these drivers.
What moves fixed-asset sample size
- Population size: Larger populations can require larger samples, though the effect on a statistical sample is far smaller than intuition suggests — beyond a point, sample size grows slowly with population.
- Materiality / tolerable misstatement: The smaller the misstatement the auditor is willing to tolerate, the larger the sample needed to detect it.
- Risk of material misstatement: Higher assessed inherent and control risk — weak asset controls, prior ghost assets, poor reconciliation — drives a larger sample.
- Desired confidence level: The more assurance the auditor wants from the sample (lower acceptable sampling risk), the larger the sample.
- Expected error / deviation: If the auditor expects more exceptions in the population, the sample must be larger to conclude reliably.
These drivers interact. A clean, well-controlled register with low expected error and a high materiality threshold can be tested with a relatively small sample; a first-time or post-restatement audit of a register with known data-quality problems will demand far more coverage. The auditor weighs all five together rather than applying any one in isolation.
Full-Scope / Key Items (Tested 100%, Not Sampled)
Before any sampling begins, the auditor carves out the items that will be tested in full. In fixed assets these are typically individually significant or high-value assets — a single asset whose carrying value alone could be material, or an item the auditor judges to carry elevated risk. These are examined 100%; they are not part of the sample.
The logic is straightforward. If one production line or one building is worth more than the entire tolerable misstatement, sampling around it makes no sense — the auditor must verify it directly. Pulling key items out first also makes the remaining population more homogeneous, which is exactly the condition under which a sample of the residual items behaves well. The population that actually gets sampled is therefore the lower-value, higher-volume tail — furniture, fixtures, standard IT equipment, and similar items where no single asset is individually significant.
Two-Directional Testing
Sampling a fixed-asset population is not a single test — it is two, run in opposite directions, because different errors hide in different directions. This is the heart of how existence and completeness are evidenced.
Existence: Register → Floor
The auditor selects items from the fixed asset register and traces each one to the physical asset on the floor — confirming the asset exists, is in the recorded location, and matches its description and tag. This direction catches ghost assets: items still carried on the books that no longer physically exist.
Completeness: Floor → Register
The auditor selects assets observed on the floor and traces each one back to the register — confirming it is recorded, tagged, and depreciated. This direction catches unrecorded assets: items physically present and in use but missing from the books.
A register-to-floor test alone would never reveal an asset that was never recorded, and a floor-to-register test alone would never reveal a record with no asset behind it. Testing only one direction leaves half the risk untouched. Both selections are sized using the same sampling logic described above, and the results of each feed the corresponding assertion. For more on how these results become evidence, see our guide to the types of audit evidence used for fixed assets.
How Big Should the Sample Be?
There is no universal answer, and any source quoting a single "count X% of your assets" rule is oversimplifying. The correct sample size depends on the drivers described earlier — risk, materiality, desired confidence, expected error, and population size — combined with the key items already pulled for 100% testing.
In qualitative terms, the sample grows when:
- the assessed risk of material misstatement is higher;
- the tolerable misstatement (materiality) is lower;
- the auditor wants a higher confidence level / lower sampling risk;
- more error is expected in the population.
And it shrinks when the register is clean, controls are strong, expected error is near zero, and materiality is comparatively high. For statistical samples, these inputs are run through a sampling formula or software to produce a defensible number; for judgmental samples, the auditor reasons through the same factors to arrive at a size they can support. Either way, the size is a conclusion from the inputs, not a figure chosen in advance. Population size matters, but past a certain point a statistical sample grows only slowly as the population gets larger — which is why a 10,000-asset register does not need ten times the sample of a 1,000-asset register.
Documentation & Conclusions
A sample is only as good as the record that explains it. Audit documentation should make the sampling approach reproducible and the conclusion traceable. At a minimum, the workpapers capture the objective of the test and the assertion addressed, the population and how it was defined, the items removed for 100% key-item testing, the sampling method (statistical or judgmental) and the basis for sample size, how items were selected, and the results.
When exceptions are found, they are not simply noted and set aside. The auditor investigates each one to understand its cause, decides whether it is an isolated anomaly or indicative of a wider problem, and — for statistical samples — projects the error to the population to evaluate it against tolerable misstatement. If the projected error plus an allowance for sampling risk approaches or exceeds materiality, the auditor extends testing, requests adjustments, or reconsiders the assessed risk. The final step is a clear written conclusion: whether the evidence obtained, sample plus key items, supports the existence and completeness of fixed assets at an acceptable level of assurance.
How CPCON Approaches Sampling
CPCON works on both sides of this line. When the objective is a management fixed asset audit or register cleanup, we typically perform a wall-to-wall physical verification — scanning and reconciling every asset — because the deliverable is a fully proven register, not an estimate. When the objective is a risk-based test between full counts, we apply a structured sampling plan: key items verified 100%, higher-risk categories targeted with judgmental selections, and the homogeneous remainder sampled and projected.
In every case we run both directions of the test, document the population, selection basis, exceptions, and conclusions, and hand over scan logs, photos, and exception reports as evidence. That is what lets an external auditor lean on our verification to support the existence and completeness assertions — and, where they choose to sample independently, start from a register that is already clean enough to keep their sample small. You can see how the full process fits together in our guide on how to audit fixed assets.
Frequently Asked Questions
How do you determine fixed asset audit sample size?
Auditors set fixed-asset sample size from the assessed risk of material misstatement, materiality (tolerable misstatement), the desired confidence level, the expected error in the population, and population size. Higher risk or confidence and lower tolerable error increase the sample; high-value or individually significant assets are tested 100% as key items rather than sampled.
What is the difference between statistical and judgmental sampling?
Statistical sampling uses random selection and probability theory, letting the auditor quantify sampling risk and project the error to the whole population with a measured confidence level. Judgmental (non-statistical) sampling relies on the auditor's experience to choose items and size the sample; it is faster and still requires a representative selection, but it does not produce a statistically projectable conclusion. Both are acceptable under auditing standards when applied with appropriate professional judgment.
Should every fixed asset be physically verified?
Not always. A full census of every asset is often impractical for large, multi-site populations, so auditors test a representative sample. However, individually significant or high-value assets — and any items with elevated risk — are typically examined 100% as key items rather than sampled. Organizations doing a register cleanup or first-time audit frequently choose a wall-to-wall physical count for their own purposes, which then strengthens the existence and completeness evidence the auditor relies on.
What is two-directional fixed asset testing?
Two-directional testing checks fixed assets in both directions. To test existence, the auditor selects items from the fixed asset register and traces them to the physical asset on the floor (register → floor). To test completeness, the auditor selects assets observed on the floor and traces them back to the register (floor → register). The register-to-floor direction catches ghost assets; the floor-to-register direction catches unrecorded assets.
