Several cycles without material nonconformities may describe a mature system. They may also describe a sample that repeats itself. There is no basis for asserting that the probability of a gap automatically increases by the third cycle. I use a clean history as a sample-renewal heuristic · a signal to change the questions, not a presumption against the organisation.
The operational question this criterion puts before the board is a single one · when the certificate arrives flawless for the third time, has the system improved, or has it trained itself to look better? Both hypotheses produce the same document. They produce different organisations.
Three mechanisms · none requires malice
Mechanism one · the client's adaptation to the auditor's method. Cycle after cycle, the organisation learns what the auditor expects · what questions they ask, what records they request, what evidence they review in detail. Without any intention to deceive, the team starts presenting itself in the language and format the auditor recognises. Processes the auditor does not examine in depth stop being prepared for audit; later, they stop being prepared at all. That is where dark zones emerge.
Mechanism two · the auditor's adaptation to the familiar organisation. The auditor returning for the third time brings a prior picture. That picture is an efficiency asset · they already know where the risks were, which processes work and whom to ask. Efficiency comes at the cost of a fresh perspective. What prior hypotheses do not flag is not examined in detail.
Mechanism three · mutually comfortable evidence. When the client trusts the auditor and the auditor trusts the client, the conversation becomes cordial and ceases to be adversarial. Fewer tough questions on one side · less uncomfortable information on the other. The audit closes with few findings and everyone satisfied. The management system was not evaluated · it was confirmed.
A case that demands asking the question again
The Ministerio de Transporte japonés recorded 142 irregularities in Daihatsu type-approval applications, covering both models in production and discontinued models, and ordered checks and corrections (MLIT, 2023-2024). This was not an ISO certification cycle. The case is relevant because it shows a more general boundary · a previous assessment does not replace a review of changes or turn an approval into permanent truth.
A subsequent incident does not demonstrate that previous audits were flawed. It requires reconstructing their scope, sample and evidence, and whether the issue already existed or arose later. The history is useful when it allows that sequence to be established without guessing.
Four patterns when someone breaks the cycle
Pattern one · undocumented evolution of scope. The organisation expanded operations, opened offices and added products. The management system was never formally updated. The certified scope ceased to match actual operations, and the regular auditor continued auditing the organisation they knew, not the one that existed.
Pattern two · critical processes outsourced without governance. The organisation gradually delegated critical system tasks to suppliers without formalising controls. The auditor assumed that control existed because it always had.
Pattern three · key staff who left without a handover. The person who designed the system left the organisation. The system runs on inertia · no one remembers why each control is the way it is. When a control fails, there is no one who knows how to fix it.
Pattern four · minor findings that concealed a major one. Previous audits detected minor findings, all closed on time. The list of closures served as evidence of maturity. What the list concealed was the underlying systemic pattern that, when it finally became visible, constituted a major nonconformity.
The algorithmic version of the paradox
A model that prioritises the sample using past findings may focus on areas where problems have already occurred and omit areas with no history. The corrective approach is designed before using it · document variables, reserve part of the sample for new hypotheses and periodically review which populations have never been selected.
The clean-cycle test
Four specific stages when two or more consecutive cycles show no material nonconformities.
Stage one · review of actual scope. Not only the scope declared on the certificate · also actual operations, confirmed with evidence (billing, contracts, staffing, offices, products), and a rigorous comparison between the two.
Stage two · interviews with new staff. Confidential interviews with staff who joined in the last two years, outside the processes typically audited. Three questions, phrased in practical terms · «what has changed since you arrived, what surprised you, what makes you uneasy». New staff's unease is the information that a mature system tends to normalise.
Stage three · review of accumulated minor nonconformities. Statistical analysis of findings closed over the last three cycles, looking for the pattern that individual closure does not show. Minor findings concentrated in the same area are a sign of an unaddressed systemic cause.
Stage four · review of unreported structural changes. Changes in ownership, management, critical suppliers, underlying technology or applicable regulation. Any of these changes without a formal system update is a candidate for a material finding.
If the four stages reveal nothing new, the result may be perfectly valid. The quality of the exercise is judged by its design, evidence and declared limitations, not by an obligation to find something.