What is an inspection audit: inspection audits explained

An inspection audit is a re-examination of a sample of completed inspections, carried out to establish whether the work behind the records really happened and whether the judgements recorded in them were sound. It looks at the records rather than the assets, sampling closed inspections and testing them against what a competent reviewer would have concluded.

The plain description is the useful one: it is the check on the checkers. Confusingly, the term is also used for an audit of an inspection function by an external body against a standard, which is a governance exercise. This entry covers the operational version that a quality manager runs internally.

What does an inspection audit look for?

An audit is testing three separate things, and they fail in different ways. Keeping them apart in the report is what makes the results actionable.

• Did it happen? Is there evidence that a person was in front of this asset on the date claimed? Capture timestamps, location signals and image content answer this, and a closed record on its own does not.
• Was it complete? Were all required steps carried out and does the submission carry the full photo requirement, or did the inspector skip the awkward angles?
• Was the judgement right? Given the evidence recorded, would a competent second reviewer reach the same finding? This is the slow, expensive part, and the part most audits skimp on.

A fourth question sits underneath all of them. Is the record consistent with itself? Four images captured within 90 seconds of each other, at the same GPS point, for a walkaround that takes twenty minutes, is a discrepancy worth opening.

How does an inspection audit work?

The mechanics are ordinary sampling, and the discipline is in the selection.

• Sample randomly, not conveniently. Auditing whichever records are easiest to open produces a flattering and useless result.
• Stratify deliberately. Pull across inspectors, sites, asset types and time of day, plus a slice from the last two days of the reporting period where period-end closures cluster.
• Add a targeted sample. On top of the random sample, deliberately pull inspections that recorded no defects at all, and inspections completed unusually fast.
• Score against written criteria. A pass or fail per question, not an overall impression. Impressions are not comparable across auditors.
• Feed the result back as a rate. The output is a verified rate you can set against the claimed inspection completion rate, plus a defect list against the process.

Sample size is usually a negotiation with capacity. Auditing 5 percent of a monthly population, thoroughly, tells you more than a superficial pass over 30 percent of it.

An inspection audit in practice

Of 288 closed condition inspections in a month, a quality manager audits 40. Four have no evidence that the asset was seen. Eleven more are missing at least one required image, and two record a pass on a component that the attached photograph shows as damaged.

The headline that reaches the operations director is that claimed completion was 90 percent and verified completion is closer to 81 percent. The more useful line is buried in the detail: nine of the fifteen problem records were closed on the last working day of the month.

The uncomfortable finding an audit usually produces

Audits rarely uncover a rogue inspector. What they almost always uncover is that a meaningful share of inspections recorded as complete were not really performed, and that the process quietly encouraged it.

The pattern repeats across industries. Failures cluster at period end, on the longest forms, on the least accessible assets, and among the inspectors carrying the heaviest schedules. That is a workload and process design finding wearing the costume of an integrity finding.

It lands badly, because the natural response is to discipline individuals. Do that and the next audit will look better while the underlying situation gets worse, since the immediate lesson learned is to make records harder to question rather than to inspect more assets. The productive response is to cut the required steps to the ones that carry decisions, fix the access problems, and pair the audit with an audit trail that makes the real sequence of events visible without anyone having to accuse anyone.

How an inspection audit differs from a second opinion inspection

Both re-examine completed work. They differ in scope and in what they can conclude.

• Scope. An audit samples a population and judges the process. A second opinion inspection re-inspects one asset and judges a single finding.
• Trigger. Audits are scheduled. Second opinions are usually triggered by a dispute or a value threshold.
• Evidence. An audit works from what was recorded. A second opinion generates new evidence by going back to the asset.
• Output. An audit produces a rate and a corrective action. A second opinion produces a decision on one case.

Related programmes carry their own names for the same instinct. A warranty audit is a manufacturer applying this logic to claims, and a rolling quality assurance inspection programme is the same sampling discipline applied to finished work rather than finished paperwork.

Making the audit itself credible

An audit run by the manager whose completion rate is being audited is not an audit. Separate the reporting line, publish the criteria before the sample is drawn, and record every audited case rather than only the failures. Then re-audit the same categories next quarter, because a corrective action nobody re-tests is a memo.