Home / Journal / Audit findings

Crochet factory audit findings: how non-conformities actually get closed

Updated 9 November 2026 · Compliance, audits and corrective action

Closing an audit finding means producing evidence that the cause of the non-conformity has been removed and that the evidence has been accepted by whoever holds the finding — and on a social audit the surprising part is that the auditor and the audit scheme do not issue a pass or a fail at all: the buyer decides, and the finding stays open until the buyer's chosen verification route says otherwise.

Most guidance on corrective action plans is written as if the factory closes its own findings by promising to do better. That is not how the mechanism works. On a SMETA audit, Sedex states plainly that neither the auditor nor Sedex issues a pass or a fail, and that the report carries no final grade or score; the conclusion belongs to the customer. On an amfori BSCI audit, the buyer requests the follow-up and tracks the corrective actions on the platform. In both cases the factory supplies evidence, but it does not award itself the closure.

That distinction decides how you should read a report, how fast you should respond, and what you put in the purchase order. Below is the mechanism, the three kinds of finding and why they close differently, the four artefacts that survive a desk review, the fake closures buyers learn to recognise, and a seven-day plan for the first response.

In this article

1. What the report actually says · 2. Correction versus corrective action · 3. Who is allowed to close a finding · 4. Evidence that survives a desk review · 5. Four fake closures · 6. What to write into the order · 7. The first seven days

1. What the report actually says

A social compliance report is a list of observations against a standard, with photographs, document extracts and worker interviews behind each one. It is not a certificate and it does not rank the factory. Three consequences follow, and each one catches buyers out.

Assumption buyers makeWhat the scheme actually states
“The report has a grade, so I can compare two factories.”SMETA reports carry no final grade or score. Sedex describes SMETA as a continuous-improvement methodology rather than a certification, precisely because a certification gives a pass-or-fail snapshot.
“The report expires, so I know when to re-audit.”The report itself has no fixed validity period under SMETA; the customer sets the cycle. Full amfori BSCI audits run every two years, with follow-up audits in between covering only the performance areas that need new evidence.
“Findings close automatically once we send the plan.”On SMETA, closing a non-conformity requires contacting the audit firm that performed the full audit and arranging a desk review or a follow-up audit. The audit firm does not require a follow-up as a matter of course; which findings must be closed is decided by the customer.
Read that third row twice, because it is where schedules break. A factory can be finished with its corrective work and still be “open” in the buyer's system for weeks, waiting for an audit firm's desk review slot. If your purchase order or your onboarding checklist says “all findings closed before first shipment”, the constraint is not the factory's effort but a third party's calendar — which is why the verification route should be agreed at the same time as the closure deadline.

The practical translation: when a report lands, the first question is not “how bad is it” but “who holds this finding and what will they accept as closure”. Those are different questions with different owners, and confusing them is the most common cause of a stalled remediation.

2. Correction versus corrective action

ISO 19011, the auditing guidance standard, keeps three terms apart, and the distinction is the whole basis of how findings are graded and closed: a nonconformity is the failure to meet a requirement, a correction removes the detected failure, and a corrective action removes the cause so the failure does not recur. Fixing the pay slip that was wrong is a correction. Changing the payroll review step that allowed it to be wrong is a corrective action.

Auditors grade findings largely on how much of the system is implicated. An isolated slip that one person can fix is graded low. The same slip appearing across a sample of twelve files suggests a control that does not work, and is graded high — not because the error is bigger, but because the evidence points at the system rather than at the incident.

Type of findingWhat closes itTypical evidence
Isolated, administrativeA correction, usually accepted at desk reviewCorrected document, dated, with the original retained
Recurring across a sampleA corrective action: change the control, then show the new control operatingRevised procedure, training record, and two or three months of the new record showing it worked
Systemic or rights-basedCorrective action plus a period of monitoring; often requires a follow-up audit rather than a desk reviewRoot-cause analysis, retroactive remediation for affected workers, monitoring data over time

The trap in the second row is time. A revised procedure dated the day before the desk review proves that a document exists; it does not prove that the control operates. Reviewers know this, so the credible package is a procedure plus records generated after the change — the new checklist with three months of completed entries, the training sign-in with names that match the payroll, the calibration certificate for the scale that was previously uncalibrated.

For a crochet programme the recurring findings we see cluster in a few places, and none of them are exotic: piece-rate wage calculations where the recorded output and the paid output reconcile only loosely; overtime recorded in one book and paid from another; needle, hook and small-part control where the count is done but not signed; and age verification for home-based workers in the surrounding villages, where the paperwork exists for the workshop but not for the outworker network. Each of those is a control problem, not an incident, so each needs a control-level answer.

3. Who is allowed to close a finding

There are three routes, they cost different amounts of time, and they are not interchangeable.

Desk review by the audit firm

The factory submits evidence; the audit firm reviews it remotely and updates the report. This is the cheapest and fastest route, and it is the one to ask for first. It works when the finding is documentary and the evidence is self-proving: a corrected record, a new procedure with matching records, a certificate. It does not work when the finding is about conditions on the floor, because a photograph of a fixed fire exit dated after the audit is still only a claim until someone looks.

Follow-up audit

A shorter on-site visit, scoped to the open findings. On amfori BSCI this is the standard mechanism between full audits: the buyer requests the follow-up and the visit covers only the performance areas that need new evidence. On SMETA the audit firm arranges it on request. Budget the lead time honestly — auditor availability, not factory readiness, is usually the binding constraint — and expect the follow-up to be charged.

The buyer's own verification

Many buyers do not wait for the scheme at all. They run their own desk check, send a third-party inspector, or ask their sourcing office to walk the line during the next visit. This route is faster and entirely at the buyer's discretion, which is exactly why the acceptance criteria should be agreed in writing before the work starts: “photographs with visible date and location, the revised procedure, and three months of the new record” is a specification; “evidence of improvement” is an argument waiting to happen.

Ask one question early: which route will this buyer accept for this finding? The answer changes the work. A finding that can close by desk review needs documents; the same finding routed to a follow-up audit needs the floor to look right on a specific day, with the records and the people present to explain them. Preparing for the wrong route is the second most common cause of a stalled remediation, after not knowing who holds the finding.

4. Evidence that survives a desk review

Reviewers are not looking for effort; they are looking for four properties: relevance, dating, internal consistency, and a trail back to the original finding. An evidence pack that has all four closes quickly. One that has three of them comes back with questions, and each round trip costs a week or more.

Two smaller habits pay for themselves. Keep the superseded version of every corrected document — a corrected record with the original attached shows a working control, while a clean replacement with no history looks like a rewrite. And photograph with something in frame that fixes time and place: a dated whiteboard, a gate sign, a loading bay. An undated photograph of a tidy storeroom proves that a tidy storeroom existed at some point.

5. Four fake closures

Buyers who audit repeatedly learn to recognise closures that will not hold. They are worth knowing from both sides of the table, because a closure that fails on the next visit is worse than an open finding honestly reported.

PatternWhy it fails
The paper-only fix. A new SOP is written, no records follow.The next audit samples records, not documents. An SOP with no output is evidence that the control is not running.
The retro-fill. Missing records are completed in the same handwriting, in one sitting.Ink, pen, sequence and cross-references give it away, and the finding escalates from a records gap to a integrity issue.
The cosmetic pass. The floor is rearranged for the visit and reverts afterwards.Follow-up audits and unannounced visits exist for this. Worker interviews also contradict a floor that only looks right on audit day.
The promise with a date. “Will be implemented by Q2” with no owner and no milestone.Nothing to verify at the next review, so the finding reopens with a shorter deadline and less goodwill.

The counter-pattern is unglamorous: a dated root-cause statement, one control change, records produced after the change, and a named person who owns the record. It reads badly as a slide and closes reliably in practice.

6. What to write into the order

Because closure depends on a third party's calendar, the commercial documents should carry the mechanics rather than assume them. Five clauses do most of the work.

  1. Named findings, not a general condition. List the finding references that must be closed before shipment, and state explicitly that findings outside the list do not block the order. A blanket “all findings closed” gives a buyer an accidental right to hold any shipment for any minor observation.
  2. The verification route. Say whether desk review, follow-up audit or buyer inspection will be used, and who pays for it. Follow-up audits are chargeable and the cost is often discovered late.
  3. A realistic window, tied to the route. A desk review and a follow-up audit have different lead times; a single number for both guarantees a missed date. Anchor the date to the audit firm's booking, not to the factory's effort.
  4. What happens if it slips. Prefer a defined consequence — a re-inspection at the supplier's cost, a hold on the next release, a reduced allocation — over termination language that nobody wants to use and that therefore disciplines nothing.
  5. Evidence standards. State the four properties from the previous section as acceptance criteria. It is a short paragraph and it removes most of the argument.

None of this is adversarial. A supplier that knows exactly which findings block shipment, which route will verify them and what evidence will be accepted can plan production around it — and a buyer who writes those three things down stops discovering them at the loading bay.

7. The first seven days

The work that decides the outcome happens before any corrective action is written.

DayActionOutput
0–1Read the report as a buyer would: which findings are systemic, which are documentary, which are contestedA triage list, and a list of findings you believe are factually wrong
1–2Contest what is contestable, in writing, with evidence — not later, and not verballyA short rebuttal per contested finding
2–3Confirm the verification route and the closure window with the buyerAgreed route and date per finding
3–5Root cause per finding; one named owner per findingA CAP with owners and dates, not a list of intentions
5–7Start the control change, and start generating records under the new controlThe clock on “three months of records” starts here, not at the deadline

The last row is the one that gets deferred and the one that matters. Records cannot be back-dated credibly, so the interval between changing a control and the review is the real constraint on the calendar. Starting the new record in week one rather than week ten is usually the difference between closing a finding in this audit cycle and carrying it into the next one.

If you are commissioning the audit rather than receiving it, the same logic applies in reverse: the useful question to a supplier is not “are you compliant” but “show me the last report, the open findings, and the records generated since the control changed”. Our own 12-point factory audit checklist covers what to look at on the day; the seven checkpoints for choosing a crochet manufacturer cover the sequence before it; and the TÜV audit scope page shows what our own file contains.

Does a failed audit mean the factory cannot ship?

No, because on SMETA there is no pass or fail to begin with: Sedex states that neither the auditor nor Sedex issues a pass or a fail, and that the report has no final grade or score. Whether a shipment is held is a commercial decision by the buyer, usually expressed as a list of findings that must be closed first. On amfori BSCI the rating scale does run from A to E, but even there the consequence is set by the member's own sourcing policy rather than by the audit itself. The practical question is always which findings, verified how, by when.

Who actually closes the finding — the factory or the auditor?

The factory produces the evidence; the audit firm or the buyer accepts it. On SMETA, closing a non-conformity requires contacting the audit firm that performed the full audit and arranging a desk review or a follow-up audit, and the audit firm does not require a follow-up as a matter of course — which findings must be closed is decided by the customer. So the factory can finish its corrective work and still be open in the system, waiting for a review slot. Plan the review booking at the same time as the corrective action, not after it.

How long should a corrective action take to close?

Longer than the document and shorter than the record. A documentary finding can close at the next desk review, which may be a matter of weeks. A control finding needs records generated after the control changed, so the honest minimum is the review interval the buyer will accept — commonly a few months of the new record. Anyone promising closure of a systemic finding within a fortnight is either describing a correction rather than a corrective action, or describing a document that will not survive the next audit.

Can we close a finding with photographs only?

For a physical condition, sometimes, if the photograph fixes time and place and the condition is self-evidently corrected. For anything involving a process, no: a photograph shows a state, not a control. The combination that reviewers accept is a dated photograph plus the record that shows the new control operating over time. Undated photographs of a tidy workshop are the most common item in a rejected evidence pack.

What is the difference between a correction and a corrective action?

A correction removes the detected non-conformity; a corrective action removes its cause so that it does not recur. ISO 19011 keeps the two terms apart for exactly this reason. Fixing the one wrong payslip is a correction and closes an isolated finding. Changing the review step that let the payslip be wrong, then showing three months of payslips that passed through the new step, is a corrective action and is what closes a finding graded as systemic. Submitting a correction for a systemic finding is the single most common reason a closure is rejected.

Do we have to do a follow-up audit, and who pays for it?

Not automatically. Under SMETA the audit firm does not require a follow-up as standard; the buyer decides which findings need one. Under amfori BSCI the follow-up is the normal mechanism between two-yearly full audits and is requested by the member, covering only the performance areas that need new evidence. Follow-up audits are chargeable in practice, and the cost usually sits with the supplier unless the contract says otherwise — which is a good reason to put the route and the cost in the purchase order rather than discover them at invoicing.

Many "not as approved" findings are really missing references rather than missing control. Sealing and retention practice removes that class of finding before the auditor arrives.

Holding an open finding on a crochet supplier? Send the finding text and the report date, and we will tell you which route we would use to close it, what evidence we would submit, and how the interval affects your shipment date — before you set a deadline that the audit calendar cannot meet.

Email bianhaitao@kivastitch.com · WhatsApp +44 7796 847136 · See the 12-point audit checklist and our TÜV audit scope.