Preparing for a client or accreditation audit

Whether it is a principal contractor, an accreditation body or a client's own auditor, the exercise is the same: a small sample, followed through end to end, to find out whether the system behind it is real.

They follow threads, they do not read everything

An auditor given a folder will typically pick two or three structures and trace them — the design basis, the handover, the inspection history, any adaptations, the competence of the people involved. What is being tested is whether the chain holds, not whether documents exist. One broken link says more than a hundred filed certificates.

Which is why coverage matters more than volume. A firm that cannot say how many structures it has standing cannot demonstrate that its inspection records represent all of them, and a register is what turns a pile of inspections into a coverage figure.

Consistency beats perfection

Five files each complete in a different way suggests the outcome depends on who did the work. A consistent set with one clearly documented exception suggests a process with a decision in it. The second is a far better result, and it is achievable where perfection is not.

In practice: audit your own files first

The cheapest preparation is reading your own records the way an auditor would, a few weeks ahead. Pick two structures at random from different sites and try to assemble everything: design basis, handover, inspections, adaptations, who inspected and why they were appropriate.

Whatever cannot be produced in an afternoon is the finding, and finding it yourself is considerably better than being shown it.

Common mistakes

• –Assembling records when the audit is announced rather than maintaining them
• –No register, so coverage cannot be stated
• –Inconsistent practice between supervisors or sites
• –Exceptions handled without a record of who decided and why
• –Evidence spread across site folders, inboxes and personal drives
• –Never sampling your own files before somebody else does

What a good result actually looks like

Firms preparing for an audit tend to aim at having nothing found, which is both unlikely and not what is being assessed. An auditor finding two minor items in a consistent, well-documented system generally forms a better view than one finding nothing in a system they cannot follow.

What produces a poor result is not imperfection but inconsistency — evidence that the answer depends on which supervisor, which site or which week, because that means nothing can be inferred about the files they did not look at.

Getting this right

The cheapest preparation is sampling your own records a few weeks ahead: pick two structures from different sites and assemble everything — design basis, handover, inspections, adaptations, and who inspected and why they were appropriate. Whatever takes more than an afternoon is the finding.

• –Sample your own files ahead of time rather than assembling on announcement
• –Aim for consistency with documented exceptions, not for nothing found
• –Keep a register so coverage can be stated, not just inspections shown
• –Record who decided an exception and why
• –Consolidate evidence out of site folders and personal drives
• –Fix the retrieval problem, which is usually the real finding

Where this connects: what a good result actually looks like

Firms preparing for an audit tend to aim at nothing being found, which is both unlikely and not what is being assessed. An auditor finding two minor items in a consistent, well-documented system generally forms a better view than one finding nothing in a system they cannot follow.

What produces a poor result is inconsistency — evidence that the answer depends on which supervisor, which site or which week, because that means nothing can be inferred about the files they did not look at. A documented exception with a named decision-maker reads as a process working; five differently complete files read as no process at all.

• –Aim for consistency with documented exceptions, not for nothing found
• –Record who decided an exception and why
• –Sample your own files weeks ahead rather than assembling on announcement
• –Keep a register so coverage can be stated
• –Consolidate evidence out of site folders and personal drives
• –Treat retrieval time as the finding — it usually is

One practical test: pick two structures at random

Choose two standing structures from different sites, without warning anybody, and try to assemble everything about them: the design basis, the handover, every inspection since, every adaptation, and who inspected them and why they were appropriate for that type.

Whatever cannot be produced within an afternoon is your finding, and it will almost always be retrieval rather than existence — the records were created and live in a site folder, an inbox, a phone and a filing cabinet. That is exactly what an auditor will discover, and finding it yourself three weeks earlier costs nothing.