Reconcile textile worker rosters without counting agency transfers twice
A worker can change factory and labour agency in the same move. If two teams each subtract that transfer from the combined roster, they remove the same extra entry twice. Start with the people represented by the assignment rows, then explain the reasons those rows repeat.
This guide is for textile HR, agency coordinators, production planners and reporting teams. Its original worked case reconciles factory assignments with home-finishing records. It shows an overlapping-transfer error, a packet-recipient mismatch and an unresolved participant question. The result is a bounded internal record check, with no employment-status or legal conclusion.
Fix the question before opening the files
The fictional reporting owner requests the number of distinct people who performed the selected factory assignments or selected home-finishing work during June 2026. The exercise covers two factories, A and B, agencies P and Q, and four home-finishing packets. Direct payroll and all other production work are outside this deliberately limited exercise. This is not a company-wide workforce disclosure.
All people, identifiers, relationships and dates below are invented. The case assumes that authorised owners have already approved access, confirmed the selected work population, and checked that every accepted assignment included work during June. It supplies no test for deciding whether a real person belongs in a statutory or framework-specific category. Obtain that decision separately for the actual request.
Keep an assignment row, a person reference and a change explanation as separate objects. The row records where and through which agency an assignment occurred. The person reference connects that row to the individual established by the authorised custodian. The explanation can contain several reasons for the change. Two reasons do not necessarily mean two extra rows.
In this case, the custodian supplies a confirmed person reference for every accepted row. Different agency codes can therefore link to P02 without exposing the underlying identity documents. A matching name alone would not establish that link. If a proposed link is disputed, retain the unresolved pair and ask the custodian to decide; do not let a spreadsheet merge it automatically.
Read the twelve assignment rows
Each date range below includes both named dates, using the factories’ local calendar. A replacement assignment starts the day after the preceding one ends. The example has no gaps, simultaneous assignments, departures or new arrivals within these eight people. Those assumptions make the transfer arithmetic visible without introducing an attendance or hours calculation.
P01 has R01 at factory A through agency P, 1–30 June. P02 has R02 at A through P, 1–15 June, followed by R03 at B through Q, 16–30 June. P02 changes both site and agency, but the records describe one continuing person.
P03 has R04 at A through P, 1–10 June, followed by R05 at B through P, 11–30 June. P04 has R06 at A through Q, 1–12 June, followed by R07 at A through P, 13–30 June. P03 changes site; P04 changes agency.
P05 has R08 at B through Q, 1–7 June, followed by R09 at A through P, 8–30 June. P05 changes both site and agency. P06 has R10 at B through P, P07 has R11 at A through Q, and P08 has R12 at B through Q, each for 1–30 June.
The person-to-row groups are P01: R01; P02: R02 and R03; P03: R04 and R05; P04: R06 and R07; P05: R08 and R09; P06: R10; P07: R11; P08: R12. There are four one-row groups and four two-row groups: 4 × 1 + 4 × 2 = 12 rows representing eight people.
Find the overlap between the transfer explanations
The site-change list contains P02, P03 and P05: three people. The agency-change list contains P02, P04 and P05: also three. Adding those list lengths gives six entries, but P02 and P05 appear on both lists. Only four people have replacement rows.
The tempting calculation is 12 assignment rows − 3 site transfers − 3 agency transfers = 6 people. It is wrong by two. Each of P02 and P05 has only one extra row, yet that extra row is removed under both explanations. A review that checks the two subtotals independently can miss this error.
For this particular case, the overlap check is 3 + 3 − 2 = 4 people with an extra row. The corrected bridge is 12 − 4 = 8 people. Confirm it against the eight person groups, rather than treating the corrected subtraction as sufficient evidence on its own.
A stronger check works directly from each confirmed person group: count its rows, subtract one, then add those excess-row counts. Here the excess is 0 + 1 + 1 + 1 + 1 + 0 + 0 + 0 = 4. If one person had three assignments, that group would contribute two excess rows. Simply counting people who transferred would then be insufficient.
Retain all twelve assignment rows. They describe different service intervals and may still be needed to review the agencies’ invoices. The person-count calculation uses each person group once; it does not cancel a valid assignment, alter a payment record or decide whether either agency billed correctly. Record-matching questions and commercial disputes need their own owners.
Use the closing roster as a separate cross-check
At the end of 30 June, the stated rows place P01, P04, P05 and P07 at A, and P02, P03, P06 and P08 at B. Each site has four people. Agency P has P01, P03, P04, P05 and P06, while Q has P02, P07 and P08: five plus three.
Both views reconcile to eight because this case explicitly has no departures, arrivals or simultaneous assignments. That agreement does not turn a closing roster into a record of everyone who worked earlier in a different month. Keep the June person groups and the closing assignment view available so that the result can be reproduced without guessing which date each file describes.
Check what the homeworking recipient list represents
OECD’s 2018 garment and footwear guidance suggests intermediary records cover everyone receiving work, including all working family members. A packet-recipient list can therefore need clarification before it is treated as a worker roster. This source informs that question; it does not prescribe the counting method used in this fictional case.
The selected finishing records contain packet H01, issued to contact C1 on 5 June; H02, issued to C2 on 20 June; and H03, issued to C1 again on 27 June. These are three packet transactions and two recipient codes. For the exercise, the authorised production owner has supplied separate, checked maker links for these three packets.
H01 links to makers P02 and P09. H02 links to P10 and P11. H03 links to P09 again. The supplied maker links therefore contain five appearances but four distinct people: P02, P09, P10 and P11. C1 and C2 are contact codes only; neither code is assumed to be an additional maker.
The factory schedule already contains P02. Joining the confirmed factory and finishing groups therefore gives 8 + 4 − 1 = 11 distinct people across these selected, resolved records. Counting all three packets as people would also happen to give 8 + 3 = 11. That numerical coincidence conceals the actual relationships and would fail as soon as the packet pattern changed.
Test the join with H03. Removing its duplicated P09 appearance changes the number of packet-to-maker links from five to four, but removes no person from the maker group because P09 also made H01. Adding a new packet for those same makers would increase transactions without increasing the confirmed person count. These are properties of the invented links, not production-efficiency measures.
Leave the fourth packet visibly unresolved
H04, dated 29 June, names contact C3 and says “four people helped”. No maker references or explanation of that phrase have been supplied. The owner has not established whether C3 worked, whether four means four additional people, or whether any participant is already P01–P11. The statement is retained as an unresolved report, not silently accepted as four new workers.
Do not report 15 by adding four to eleven, or 12 by adding the contact. Do not call eleven the complete result for all four packets. The defensible statement is that the selected resolved records establish eleven distinct people, while H04’s participants and overlap remain unconfirmed. This case provides no defensible numerical upper bound for the unresolved population.
The next request goes to the authorised owner of H04: explain what the reported four counts, confirm which people actually performed the selected work, and resolve any links to the existing person groups through the approved matching route. Ask for a decision reference and corrected aggregate bridge. The reporting coordinator does not need household biographies to resolve this record question.
Close the reconciliation with a reproducible explanation
The handover should let another authorised reviewer answer three concrete questions. Which rows support each of the eight factory person groups? Why do the site-change and agency-change explanations overlap for two people? Which of the four confirmed finishing makers is already in the factory set? The unresolved H04 record remains beside those answers, with an owner and a response date.
Keep the restricted matching evidence with the existing authorised custodian. The wider schedule can carry the selected scope, twelve-row bridge, checked aggregate counts, record-version references and the outstanding H04 question. The fictional person codes in this article illustrate relationships; they are not a proposal to publish coded real-worker lists. Apply the organisation’s approved access and release process to actual records.
Before closing, ask a reviewer to recompute the person groups without seeing the headline total. Then compare their result with both transfer lists and the packet-to-maker links. If their answer differs, locate the specific relationship responsible. An unexplained balancing deduction does not resolve whether a row represents a repeat assignment, another person or information that is still missing.
For help organising a textile reporting request and its unresolved evidence questions, discuss the scope with Nitium Advisory. Share a non-confidential description of the files and the decision needed; keep personal records out of the initial enquiry. Employment-law judgments, worker-protection assessments and independent assurance require their own appropriately authorised specialist work.
Source checked 14 October 2026. All worked records and calculations are fictional. The exercise establishes internal consistency only for its stated relationships and assumptions.
Comments