Tracing contractor coverage across four reporting cases

A contractor-coverage statement belongs to a particular measure, population and source version. Moving it into another table can change the claim being made. A reporting team needs evidence of the relationship between the two populations before describing them as equivalent.

This study preserves four historical coverage questions across five reports from Nitium’s fixed 0.12.18-local1484-directory-wave-d edition, frozen on 6 October 2026. One issuer contributes a paired original and later report. Six accepted observations anchor the cases; five were in the historical 1,040-record public projection and the remaining one was local-only at that freeze. No contractor total or new company-data record is created.

The unit of analysis is one coverage question per issuer, not a measurement of workforce size or safety performance. We distinguish a stated restriction from an unresolved mapping and retain the evidence supporting each decision. This is a source-reading study, not an examination of contracts, payroll rosters or individual employment status.

All selected disclosures are standalone; table-specific populations remain separately qualified.

Four bounded coverage cases

GIPCL’s FY2023–24 standalone BRSR covers its corporate office, manufacturing/power plants and own captive mining operations. The headcount includes only company-payroll employees and workers; no complete off-payroll contractor population is established. Source: GIPCL annual report, PDF 74–75, printed 72–73.

KRBL’s standalone FY2024–25 headcount lists 328 other-than-permanent workers. Its separate safety table includes contract workforce. These statements do not establish that 328 represents the entire safety population or an exposure-hours denominator. Source: KRBL BRSR, PDF/printed 2 and 22.

V-Mart’s standalone FY2024–25 workforce panel has employee categories and unexpanded Workers NA tokens; its safety table includes contract workforce. That note resolves neither the NA meaning nor the relationship between the two populations. Source: V-Mart BRSR, PDF 4/19, printed 163/178.

Cochin Shipyard’s standalone FY2023–24 BRSR includes a contractor note beside the FY2022–23 comparative. Archived checks found no equivalent displayed note in the original selected table. That wording difference proves no change in historical membership. Sources: original BRSR, PDF 16/printed 131; later BRSR, PDF/printed 16. Original retrieval returned 502 on 15 October; archived evidence is used.

Ask whether the coverage statement can travel

Start with the proposed destination claim. Identify the measure, entity or sites, people included, time basis and source edition it would describe. Then record the statement actually supported by the source. The gap between these two descriptions is the clarification request; it should not be filled with an assumed meaning of contractor, permanent or worker.

Use four relationship states: confirmed same population, confirmed overlap, confirmed difference, or unresolved. Attach the evidence and decision owner. Do not treat unresolved as excluded. An explicit restriction supports a definite boundary; silence about a group leaves a different question. Keep that distinction visible in a coverage map.

A cross-table mapping needs more than matching labels or figures. It should explain which people or groups each table covers and whether the time bases agree. Where the responsible custodian cannot establish membership, preserve the two disclosures independently. A valid mapping for one metric does not automatically approve transfer into another metric.

Original test: three files, three limited permissions

The following evidence packet is wholly fictional. Rowan Works is preparing an internal reporting note for 1 April 2025 to 31 March 2026. File P1 is an approved year-end company-payroll snapshot of 80 people. File A1 is a register of 120 site-access credentials issued during the year. Some people can hold more than one credential; no person-level matching result is supplied.

File S1 is an authorised annual safety schedule. Its scope note includes the company’s employees and specified on-site contractor groups Alpha and Beta. It supplies no contractor headcount, membership crosswalk or hours-worked total. The exercise accepts that stated scope for this fictional schedule only; it supplies no applicable statutory definition or safety-rate calculation.

Three proposed claims must therefore be considered separately. “P1 records 80 people on company payroll at year-end” is supported by P1. “S1’s annual scope includes Alpha and Beta” is supported by S1. “There were 40 contractors” is unsupported: subtracting 80 people from 120 credentials mixes counting units, time bases and unidentified membership.

The arithmetic 120 − 80 = 40 is reproducible, but its proposed contractor interpretation is not. A credential can be replaced, cancelled or held alongside another credential; the fictional packet supplies no mapping. The appropriate outcome is an unresolved contractor count, with the custodian asked for an authorised aggregate matching the intended population and time basis.

Two hypothetical credential allocations fit the same totals. In one, all 80 payroll people receive one credential and 40 receive a replacement; no contractor credential is needed. In another, those 80 receive one each and 40 different contractors receive one each. Neither allocation is established by A1. Their coexistence shows why the subtraction cannot choose a contractor count.

Now add a second note, S2, stating that the same safety scope also includes contractor group Gamma. This changes the available scope description. It does not prove whether Gamma was absent from the earlier data or whether S2 merely explains earlier coverage more clearly. Retain S1 and S2 and ask which source version and periods the explanation applies to.

In a final variation, the custodian confirms that S1 already included Gamma, but supplies no hours. The historical scope clarification can be recorded with that confirmation. The contractor headcount and any frequency-rate calculation remain unresolved. Resolving one question should not silently mark unrelated evidence gaps as complete.

Original fictional coverage test. Each source permits a limited statement. A shared workplace does not establish matching membership, timing or counting units.

Build a metric-by-source coverage map

Create one entry for each metric and source edition. Preserve the original label and the explicit included or excluded groups. Record the entity/site boundary and exact time basis. Link the source location, the proposed reuse and the relationship state. Finish with the evidence needed for a broader claim and the person authorised to decide it.

For the fictional packet, P1 supports only the selected payroll snapshot; A1 supports the credential count on its stated issue-period basis; S1 supports its described annual safety perimeter. The map should show three separate permissions. It should not offer a combined workforce total, contractor share or rate because none follows from the supplied evidence.

Check the map in both directions. From an output claim, an authorised reviewer should reach the exact source and qualification that supports it. From each source, the reviewer should see which outputs reuse its coverage statement. If a source version changes, this reverse view identifies the claims that need reconsideration without rewriting unaffected records.

Keep interpretation and data movement separate. A new narrative note may change how a figure should be described while leaving its numeric value unchanged. A different extract may instead add or remove people. Record the actual change, its stated reason where available, and any unresolved consequence. Identical numbers are not evidence that two population definitions are identical.

Access decisions remain separate from population decisions. Use authorised aggregates and controlled references for reporting; keep person-level rosters, agreements and sensitive records with their custodians. The coverage map does not grant access, establish a legal relationship or authorise public disclosure of real workforce information.

Limits, reproduction and next step

The four cases are purposive and historical. We did not inspect every issuer publication or determine the actual membership behind each disclosure. The case register retains six observation IDs and five report editions, including the original/later pairing. Reproduction means checking the specified source statements and their limitations, without converting unresolved coverage into zero or assuming a population change.

Four exact report PDFs were available for this check; the remaining original report is supported by recovered, hash-verified review records from October 2026. Its current delivery failure is disclosed beside the paired case. That evidence is not presented as a fresh rereading of an unavailable report. No absence statement is extended beyond the selected table.

Use the occupational-safety data guide for the separate task of matching an agreed metric, population and hours basis. This study stops at whether coverage evidence can be transferred across tables or editions. It does not compare injuries, infer exposure hours, diagnose working conditions or establish legal reporting compliance.

Nitium’s reporting support can help organise source qualifications, coverage maps and owner questions within an agreed scope. Discuss your intended reporting output through Contact, using aggregate context in the initial enquiry.

Prepared 15 October 2026 with AI-assisted research and internal checks. The source facts, original fictional test and proposed review workflow are distinct. No independent assurance, underlying measurement verification, issuer endorsement or client relationship is claimed. The reports retain their own rights; source PDFs, images and third-party assurance material are not redistributed.