Skip to content
New season Shop the Collection

What social compliance audit documents on site: part 2

What social compliance audit documents on site: part 2
The short answer

The second half of on-site document review is where most social compliance findings originate: working hours, wages, and whether the two reconcile with production records. Auditors spend less time on policy binders at this stage and more time tracing individual workers from attendance logs through payroll to payment proof. Factories that keep these record families consistent usually clear document review in a single round, while gaps between them trigger deeper sampling and longer audit days. The checklist below follows the order auditors actually work in, so a factory can prepare files the same way they will be read.

Key takeaways

  1. Attendance logs, overtime approvals, and shift rosters form the hours side of the evidence, and auditors test them by sampling months rather than reading every page. [1]
  2. Payroll sheets, pay slips, and payment proof form the wage side, and every sampled worker must trace cleanly across all three without contradiction.
  3. Production and shipment records act as the reality check, because output volumes that could not be produced within the recorded hours raise immediate questions.
  4. The verification logic resembles a factory acceptance test in spirit: the workforce system is treated as equipment under test, and records must show it performs as claimed before the buyer accepts the result. [2]
  5. A short internal reconciliation before audit day, matching hours to wages to output for the likely sample months, removes the majority of document findings.

The first half of on-site document review covers the factory as it claims to be: business licenses, policies, contracts, and handbook rules. The second half covers the factory as it actually operates, told through numbers that must agree with each other. Auditors shift at this point from collecting documents to testing them, and the mood of the audit often turns on what they find here.

Three record families carry this stage: working hours, wages, and production output. Each family is ordinary on its own, and most factories keep all three in some form. The audit question is never whether the files exist but whether a single worker can be traced through all of them without contradiction, from clock-in entry to pay slip to bank transfer.

This article follows the sequence auditors actually use. It explains how sample months are chosen, what each record family must contain, which cross-checks decide findings, and how to present the files so verification finishes in one round. Nothing here requires new software or consultants, only an afternoon of reconciliation before the audit team arrives.

How auditors choose which records to read

No audit team reads twelve months of payroll line by line, and buyers should not expect them to. Instead the auditor defines the population, usually the full year of attendance and wage records, and selects a sample that must include at least one peak-production month. Peak months matter because hour violations concentrate there, so auditors deliberately oversample the periods when the factory was under the most pressure. Understanding this selection logic is the single most useful thing a factory can do before audit day.

The sampling approach follows the same reasoning that governs acceptance sampling in quality work: a well-chosen subset, tested rigorously, supports a defensible conclusion about the whole without inspecting every unit. [1] A factory that has already reconciled its own peak months therefore faces the sample with confidence, while one that only tidied an easy month is exposed the moment the auditor names a different period. Prepare every month to the same standard, because any month can be drawn.

Auditors also reserve the right to extend the sample when the first draw shows problems. A mismatch in one month typically doubles the months under review rather than ending the inquiry, so a small inconsistency discovered early can cost the whole afternoon. This is why the pre-audit reconciliation should cover the full year rather than a showcase quarter. The marginal effort of checking twelve months instead of three is far smaller than the cost of a second audit visit.

Working-hour records, line by line

Raw attendance data

Auditors want the raw material first: clock-in exports, biometric logs, or signed manual sheets, depending on what the factory uses. Summaries prepared for management are read second, and only trusted when they can be regenerated from the raw data in front of the auditor. Factories sometimes present only the summary and discover that the auditor treats the absence of source data as a gap in itself. Keep the terminal or system export function working on audit day.

Each record must identify the worker, the date, and start and end times, including breaks where the local rules require them. Anonymous totals by department prove nothing about individuals, and auditors will set them aside. Where corrections exist, they should carry a reason and an authorizer rather than appearing as silent overwrites. Clean correction trails signal an honest system, while scrubbed data invites suspicion.

Overtime approvals and rest days

Overtime entries need matching approvals that show the extra hours were authorized and voluntary rather than simply demanded by the schedule. Auditors compare approved overtime against actual clock-out times, and systematic gaps between the two suggest the approval is paperwork rather than practice. Weekly rest records get the same treatment: one rest day per seven-day period is the baseline expectation, and auditors count rest days across the sampled months worker by worker.

Holiday and leave records complete the picture, because statutory holidays paid at the wrong rate are among the most common wage findings. The auditor will ask how the factory tracks annual leave balances and whether unused leave was compensated according to local rules. Factories that keep leave cards per worker answer in minutes, while those reconstructing from memory lose credibility for the whole wage section.

Illustration: Working-hour records Decorative illustration for the section "Working-hour records"; visual only, carries no data.

The cross-checks that decide the outcome

Record ARecord BWhat auditors compareWhat a mismatch triggers
Monthly attendance summaryRaw clock-in exportTotals and overtime hours per workerDeeper sampling of additional months
Payroll sheetSigned pay slipsNet pay and deductions per workerWorker interviews on actual take-home pay
Payroll sheetBank transfer recordsAmounts and payee namesProof-of-payment requests for cash portions
Attendance logProduction and shipment recordsWhether output fits recorded hoursCapacity analysis and overtime recalculation
Leave cardsPayroll holiday pay linesHoliday rates and leave balancesBack-pay calculation for affected workers

Presenting records so verification moves faster

Organization matters as much as content, because an auditor who cannot find the file for the sampled month starts from a position of doubt. Arrange attendance, payroll, and payment proof in matching monthly folders with identical naming, so any worker in any month can be traced in under a minute. When buyers plan a social compliance audit in China, the factories that finish document review fastest are usually those that seat one clerk beside the auditor with authority to pull any record on request. social compliance audit in China That single staffing decision often saves more time than weeks of file polishing.

Illustration: Presenting records so verification Decorative illustration for the section "Presenting records so verification"; visual only, carries no data.

Treat the records the way engineers treat a system under acceptance testing: the claim is that hours are controlled and wages are correct, and the documents are the test evidence that must demonstrate it. [2] In a formal acceptance test nobody accepts a verbal assurance that the machine works, and auditors apply the same discipline to workforce claims. Each assertion in the handbook, from overtime limits to holiday pay, should point to records that prove it held in practice. Walk through two or three worker files end to end before audit day, and fix whatever breaks the trace.

Finally, brief the payroll clerk and the attendance administrator together, because auditors interview both and compare answers. If the clerk describes an overtime process that the attendance data contradicts, both lose credibility. A fifteen-minute joint briefing on how records are actually kept, including honest acknowledgment of any weak months with corrective action already taken, reads as maturity rather than concealment. Auditors penalize cover-ups far more harshly than they penalize disclosed problems with fixes underway.

A pre-audit file routine that takes one afternoon

  1. Sample your own records firstChoose the peak month and one quiet month yourself, and trace five workers from each through attendance, payroll, slips, and payment proof. Fix every break in the chain before the auditor finds it.
  2. Rebuild one summary from raw dataRegenerate one monthly attendance summary live from the raw export in front of a colleague. If the numbers differ, find out why before audit day rather than during it.
  3. Check hourly equivalentsDivide piece-rate or overtime earnings by recorded hours for ten workers and check the results against the local minimum. Adjust any shortfalls through proper back pay with documentation.
  4. Reconcile leave and holidaysPull leave cards for the sampled workers and confirm holiday pay lines match the statutory rates. Recalculate any month where holidays fell during peak production.
  5. Assemble the audit-day file setPlace the twelve monthly folders, the worker roster, and a one-page file index on one computer or in one cabinet. Test that a stranger can locate any worker-month in under a minute.

Sources

  1. Wikipedia: Acceptance sampling —— Overview of acceptance sampling including AQL concepts used in product inspection.
  2. Wikipedia: Factory acceptance test —— Overview of factory acceptance testing performed before shipment or installation.

Frequently asked questions

How many months of attendance and payroll records should a factory prepare?

Most auditors ask for a full twelve-month window so they can include at least one peak season. If the factory has operated for less than a year, bring everything since production started and say so up front. Shorter windows invite questions about why months are missing, while a complete year lets the sampling logic work as intended.

Do auditors accept electronic attendance systems?

Yes, electronic systems are widely accepted, provided the raw data is exportable and the edit history is visible. Auditors distrust attendance files that only exist as polished monthly summaries with no underlying clock-in data. Keep the terminal exports or system logs available so the auditor can verify that summaries were generated from real inputs.

What if the factory pays piece rates instead of hourly wages?

Piece-rate pay is acceptable, but the factory must still show how many hours each worker spent earning that pay, and the resulting hourly equivalent must meet the local minimum. Auditors divide piece earnings by recorded hours, so missing hour records turn an honest piece-rate system into a finding. Keep job tickets or daily output slips that tie each worker to specific dates.

Should payroll and attendance records be translated into English?

Full translation is rarely required, since audit teams working in China normally read Chinese records. What helps far more is a one-page index in English listing each file, the period it covers, and who prepared it. That index shortens document review noticeably because the auditor stops guessing which folder holds the payroll for the sampled month.