Research libraryInventory as a Decision
Chapter 02 of 14 · Part I · Draft web edition
An inventory policy fails when it orders against quantities that are wrongly identified, reserved twice or unavailable for use.
The cloth is woven from this chapter's 5 exhibits: each weft is one exhibit's own data, normalised; each warp group is a section. The probe reads the exhibit's own value where it crosses. Procedural visualisation, not a measurement.
Read-out
Illustrative ledger schema proposed by GFT. Separates master specification and version control, batch and shade-lot attributes, handling-unit location and legal ownership, stock usability state, and event-linked transaction logs. Informed by MRP bill-of-material control (S07) and ship-receive event reconciliation (S31); no installed GFT factory database is claimed.
Unit: proposed ledger schema layers. Status: illustrative. Sources: S07, S31. Illustrative figures are teaching examples, not measured factory results.
An inventory policy can only protect a customer promise when the stock record behind it identifies the exact material that production can substitute, converts physical units without distortion, separates quality-held or reserved goods from free supply, and reconciles physical movements before new commitments are made. A reorder point, a material requirements plan or an order confirmation is an arithmetic operation applied to a ledger balance. If that balance includes rolls from an incompatible dye bath, fabric awaiting incoming inspection, or quantities already promised to another work order, the formula will suppress replenishment while the sewing line starves.[S01][S05][S07] Green Future Tech Limited (GFT), Hong Kong SAR, therefore treats usable-stock truth as the first operational gate in any inventory system: before choosing a statistical buffer or an optimisation rule, the factory must establish an event-linked ledger in which identity, unit conversion, stock state, ownership and discrepancy resolution are governed by explicit engineering and accounting controls.[S27][S31][S0203]
This chapter specifies that ledger architecture for an emerging-market garment maker within the design context of GFT's three-part Cameroon textile programme (feasibility study, performance-management system design and digital strategy, prepared 31 January 2024 with source-refresh notes retained in 2026). Whereas the national baseline records about 95 percent of cotton fibre exported raw and about 5 percent processed locally alongside a policy target of 50 percent local processing by 2030, a garment factory operating at the cut, make, trim and finishing stage cannot manage stock as undifferentiated fibre tonnage.[R06][R07] Every numerical case, roll conversion and stock balance in this chapter is illustrative and constructed solely to demonstrate ledger mechanics. No live factory ledger, warehouse audit or deployed GFT software performance is claimed, and nothing here constitutes investment, legal or tax advice.[S09][S27]
2.1
An item code in a garment factory is an engineering equivalence rule, not a purchasing shorthand. Two materials belong under the same stock-keeping unit (SKU) only if the cutting table and sewing line can substitute one for the other without altering the customer specification, the garment fit, the visual appearance or the approved bill of materials.[S01][S07] When a warehouse groups distinct materials under a broad description such as "navy cotton twill" or "white core-spun thread", the ledger creates an illusion of pooled availability. A planner who sees an illustrative book balance of 2,000 metres of navy fabric and releases a 1,200-garment institutional uniform order will cause a line stop if those 2,000 metres consist of four incompatible shade lots, two different cuttable widths and an unapproved weave construction.[S07]
…an illustrative book balance of 2,000 metres of navy fabric and releases a 1,200-garment…
S07 · illustrative
For finished and semi-finished garments, item identity requires style code, garment revision, colourway, size and pack configuration. For fabrics, item identity requires fibre composition, yarn count, weave or knit construction, nominal areal weight in grams per square metre (g/m2), nominal cuttable width between selvedges, finish treatment, colour reference and the list of qualified suppliers.[S07] For trims and packaging, identity must capture functional dimensions and compliance markings: a zip requires teeth material, gauge, exact length, slider lock type, tape colour and approved vendor; a care or size label requires language version, artwork revision and customer brand authorisation.[S07][S20] If any of these attributes changes in a way that prevents free mixing on the production line, the material requires either a distinct item code or a mandatory lot-level restriction that the planning engine respects during netting.[S07]
Even when two rolls share an identical item code and supplier specification, textile dyeing and finishing introduce batch-to-batch variation that prevents arbitrary mixing within a single garment or across matching pieces of a uniform set. A ledger that tracks only item code and aggregate quantity cannot prevent a cutter from laying rolls of noticeably different shade cast into the same bundle. GFT therefore separates five ledger dimensions that must never be collapsed into a single text field:
SKU) together with its active specification version.[S07]A1, A2, B1, for example, in an illustrative shade-sorting scheme) so that cutting markers draw from compatible shade groups.[S07][R29]CMT or assembly contracts) that cannot legally be diverted to another buyer's order.[S07][S09]Exhibit 5 illustrates how these dimensions sit in distinct architectural layers linked by immutable transaction events. The layered design is a proposed GFT control schema informed by material requirements planning (S07) and supply-chain event reconciliation (S31), not a depiction of an installed warehouse system.
Separating these layers protects the factory when specifications evolve mid-season. Suppose a buyer updates a uniform shirt specification from revision v1 to revision v2 by changing the pocket button diameter and the care-label composition text. If the warehouse ledger ties reservations only to the parent style name, the planning run will net the new v2 order against obsolete v1 labels and buttons sitting in the bin. Retaining both the item specification version and the bill-of-material (BOM) version on every purchase order, quality release and production reservation forces an explicit engineering decision: either the technical lead signs a written concession allowing the v1 trim to be consumed on the v2 order, or the v1 trim remains excluded from the v2 net requirement calculation.[S07][S0203] Substitution is therefore recorded as a dated, auditable engineering approval with a named role, never as an informal bin swap on the shop floor.[S01][S07]
2.2
Exhibit 068 paragraphs step it
Read-out
| Nominal invoice conversion | 500 kg at 200 g/m2, 1.50 m width, 1.50 m/pc | 1,667 | 1,111 | 31 |
|---|---|---|---|---|
| Measured gross roll length | 500 kg at 205 g/m2, 1.48 m actual width | 1,648 | 1,084 | 4 |
| Less defect splice allowance | 12 logged defects at 1.25 m splice loss each | 1,633 | 1,074 | -6 |
| Less unusable roll-end remnants | 17 roll tails shorter than a 9.12 m 6-pc lay | 1,596 | 1,050 | -30 |
Illustrative reconciliation of a 500 kg fabric receipt across 17 rolls against a 1,080-garment order. Nominal conversion uses 200 g/m2, 1.50 m width and 1.50 m/pc (1,666.67 m; 1,111 pcs). Inspection measures 205 g/m2 and 1.48 m width (1,648.00 m gross), deducts 15.00 m for 12 splices (1.25 m each) and 29.04 m in roll remnants, and applies 1.52 m/pc in 9.12 m 6-pc lays to yield 1,596.00 m (175 lays; 1,050 pcs). Context: S07, R29. All figures are teaching assumptions.
Unit: illustrative fabric metres and trouser pieces. Status: illustrative. Sources: S07, R29. Illustrative figures are teaching examples, not measured factory results.
A stock ledger fails silently when purchasing, warehousing and production measure the same material in incompatible units and convert between them using unverified averages. In apparel manufacturing, unit changes occur at almost every physical boundary. Knitted cotton jersey is frequently purchased from spinning and knitting mills by conditioned mass in kilograms; woven shirting and trouser twill are purchased by linear metre or roll; sewing thread is purchased by cone with a nominal length in metres; buttons and zips are purchased by gross or carton; and the cutting and sewing rooms schedule bundles in garment pieces.[S07] If the conversion factors linking these units omit width variations, areal-weight drift, splice allowances or roll-end remnants, the ledger will report positive book metres after the physical racks are empty.[S07][S0203]
The physical relationship between fabric mass and linear length shows why a fixed conversion factor is hazardous. For a roll of fabric with conditioned mass $M_{\text{kg}}$ (in kilograms), actual areal weight $w$ (in grams per square metre, $\text{g/m}^2$) and actual cuttable width $b$ (in metres), the physical linear length $L_{\text{m}}$ (in metres) is:[S07]
$$L_{\text{m}} = \frac{1{,}000 \times M_{\text{kg}}}{w \times b}$$[S07]
If a planner converts an invoiced delivery of knitted or woven fabric from kilograms to linear metres using only the nominal catalogue values $w_0$ and $b_0$, two physical errors enter the stock record simultaneously.[S07] First, textile finishing tolerances allow areal weight $w$ to vary by several percent around the nominal specification; a heavier-than-nominal fabric yields fewer linear metres per kilogram. Second, if the cuttable width $b$ between pinhole selvedges is narrower than the nominal marker width $b_0$, the cutting room must either re-lay a narrower marker that consumes more linear metres per garment or reject the narrow roll. Furthermore, aggregate linear metres across multiple rolls cannot be divided directly by average garment consumption if individual roll lengths leave unusable end remnants shorter than a single marker lay.[S07]
Just as a fabric roll cannot be converted by a single unverified multiplier, upstream agricultural figures must never be converted directly into garment stock plans. The inherited evidence register records Cameroon seed-cotton production at 330,000 tonnes in 2022, 314,455 tonnes in 2023 and about 340,000 tonnes in 2024, with a 2025 projection above 350,000 tonnes.[R01][R02] Those figures describe harvested seed cotton entering ginning, not spun yarn, finished fabric or cuttable garment kits. Ginning turnout from seed cotton to lint fibre, carding and combing waste in spinning, knitting or weaving loss, wet-processing shrinkage in dyeing and finishing, and marker utilisation in cutting each occur at separate physical boundaries with distinct yield distributions. In the GFT white-paper library, WP06 owns the measurement of process yields and conversion losses across those stages; within WP12, the inventory ledger records only the measured physical input and accepted output at each stocking boundary, never a universal cotton-to-garment conversion constant.[R01][S07]
Exhibit 6 summarises the illustrative reconciliation from invoiced weight to cuttable garment capacity. The table shows why the ledger must store both the purchase unit and the verified production unit on every handling unit.
To prevent this failure mode, GFT proposes a strict dual-unit ledger rule: every roll, cone or carton is received in its commercial invoice unit (kilograms, cones or cartons) for accounts-payable matching, but cannot enter available production stock until its actual production unit (inspected usable linear metres, cuttable width and shade lot) is recorded at the handling-unit level.[S07][S31] Any discrepancy between ticketed roll length and measured roll length outside the supplier's contractual tolerance remains attached to the supplier lot record for commercial claim or replenishment adjustment rather than vanishing into cutting-floor scrap.[S27][S31]
2.3
Book on-hand
Book on-hand: total 1,000. Running total 1,000 illustrative component or garment pieces · illustrative
Even when item identity and physical units are exact, a stock ledger will mislead planners if it treats every physical unit inside the warehouse fence as available to promise.
Quarantine hold
Quarantine hold: removes 100. Running total 900 illustrative component or garment pieces · illustrative
Accepted usable
Accepted usable: total 900. Running total 900 illustrative component or garment pieces · illustrative
Valid reservations
Valid reservations: removes 200. Running total 700 illustrative component or garment pieces · illustrative
Unreserved usable
Unreserved usable: total 700. Running total 700 illustrative component or garment pieces · illustrative
The method behind it: Reconciling book stock, accepted usable stock and net requirements.
Illustrative arithmetic waterfall. Starts with 1,000 physical pieces on hand in the book ledger, subtracts 100 quarantined pieces awaiting quality disposition to reach 900 accepted usable pieces, and subtracts 200 validly reserved pieces committed to active orders to leave 700 unreserved usable pieces available for new promises. Reservations must not be subtracted a second time if demand has already been netted. S07, S15 and S31 inform the separation of physical, inspected and committed states; all quantities are teaching assumptions.
Unit: illustrative component or garment pieces. Status: illustrative. Sources: S07, S15, S31. Illustrative figures are teaching examples, not measured factory results.
2.3 · the bodyExhibit 07 above; its figures are illustrative where marked.
Even when item identity and physical units are exact, a stock ledger will mislead planners if it treats every physical unit inside the warehouse fence as available to promise. At any moment, a factory warehouse contains materials in fundamentally different operational and legal states: cartons just unloaded on the receiving apron, rolls awaiting laboratory shrinkage or shade testing, accepted materials already allocated to a confirmed work order, defective lots awaiting return to a vendor, and customer-returned garments awaiting inspection.[S07][S09][S26] Collapsing those quantities into a single on-hand figure causes the planning system to double-promise scarce components and delay reorders until the cutting room physically discovers that the bin holds only quarantined stock.[S05][S07]
GFT's ledger specification enforces six mutually exclusive status states for physical inventory on site, alongside a separate pipeline state for goods in transit:
Q_trans): Material dispatched by the supplier against an open purchase order and supported by shipment documentation, but not yet physically received at the factory gate.[S13][S31] In-transit stock is part of pipeline supply and cash exposure, never part of on-hand stock.[S09][S31]Q_quar): Material physically received at the dock or returned from the line that is blocked from production commitment pending count verification, incoming quality inspection, shade-band classification or technical disposition.[S07][S31]Q_avail): Material that has passed incoming inspection, carries a verified unit quantity and shade lot, and is free to be allocated to a production order or customer promise.[S07]Q_res): Material that has passed inspection and is bound by an active, dated reservation to a specific confirmed customer order or released work order.[S07]Q_rej): Material that has failed specification or quality tolerances and is physically segregated pending supplier return, authorised rework or accounting write-down under IAS 2 net realisable value rules.[S09][S27]Q_ret): Customer returns or post-packing rejects held in a segregated inspection zone until grading determines whether they can be restocked as first-quality goods, repaired, downgraded to seconds or lawfully disposed of.[S09][S26]Exhibit 7 shows how an illustrative physical book balance of 1,000 pieces steps down to 700 unreserved usable pieces once quarantined holds and valid reservations are excluded. Every figure in the chart is an independent teaching assumption used to make the state transitions inspectable.[S07][S15][S31]
…illustrative physical book balance of 1,000 pieces steps down to 700 unreserved usable pieces…
illustrative
The waterfall also highlights why reservations require an expiry and reconciliation rule. In many factories, planners place manual reservations on fabric or trims for tentative customer enquiries that are later cancelled or modified, leaving "ghost reservations" (Q_res) locked in the ledger. Those phantom reservations depress U_free, trigger unnecessary replenishment orders and block other feasible work orders from releasing.[S07][S0203] GFT would attach a mandatory parent-order reference and an expiry timestamp to every reservation record: if a tentative customer quote is not converted into a confirmed order by its expiry date, or if a work order is closed short, the ledger automatically releases the residual reservation back to Q_avail and logs the state transition.[S07][S27]
2.4
The moment external supply crosses the factory gate is where most ledger inaccuracies originate. When a busy receiving dock posts a goods receipt by simply clicking "receive in full" against the open purchase order screen, any supplier short-shipment, carton substitution, transit damage or unapproved shade split is silently absorbed into the factory's book stock.[S15][S31] Months later, that unrecorded receiving discrepancy resurfaces as an unexplained stockout on the sewing line or an inventory write-off at year-end.[S09][S27] Receiving must therefore operate as a four-way chain of evidence that links commercial authorisation, supplier dispatch, physical dock tally and technical quality release.[S07][S31]
Evidence from retail and apparel supply chains confirms that order, shipment and receipt records diverge frequently even among established trading partners. In Project Zipper, a one-year study conducted from June 2017 to July 2018 by GS1 US and the Auburn University RFID Lab, researchers examined EPC/RFID and transaction data exchanged between eight brand owners and five retailers.[S31] While the numerical error percentages reported in that North American study are specific to its retail sample and must never be cited as guaranteed operating savings or a baseline for a Cameroon factory, the study establishes a vital structural principle: purchase orders, advance shipping notices (ASN), physical receipts and inventory records are distinct claims about reality that require explicit event-by-event reconciliation.[S31]
PO): Records the authorised item code, specification and BOM version, ordered quantity, purchase unit, agreed price, Incoterms rule and named place, and promised delivery window.[S07][S29]ASN or packing list): Records the quantities, roll or carton serial identifiers, dye-bath lots and carrier references that the supplier claims to have dispatched on t_ship.[S31] Until the physical goods cross the factory dock, this quantity sits in in-transit (Q_trans) rather than warehouse stock.GRN): Recorded at the dock gate on t_dock via a blind or unit-verified physical count of cartons, rolls or cones before the operator sees the supplier's packing-list subtotals.[S15][S27] The GRN moves the physically verified quantity from in-transit into quarantine (Q_quar), not directly into available stock. Any gap between the ASN quantity and the GRN tally immediately opens a shipment discrepancy record (short receipt, over-receipt, damaged container or wrong item label).[S27][S31]IQR): Recorded by the quality role on t_release after physical width and weight checks, shade-band continuity grouping and defect inspection.[R29][S07] The IQR transitions accepted handling units from quarantine (Q_quar) to available (Q_avail) (or directly to reserved, Q_res, if pegged to a waiting order) and moves failed units to rejected (Q_rej) with a mandatory exception reason code.[S07][S27]Separating t_event from t_rec across these four steps serves two diagnostic purposes. First, inspired by the World Customs Organization's Time Release Study methodology (S30), which uses process mapping and segmented timestamps to isolate clearance bottlenecks and inspect outliers, recording both dock arrival (t_dock) and quality release (t_release) reveals how much replenishment time is consumed inside the factory's own receiving and inspection queue before material becomes usable.[S30] Second, comparing t_event with t_rec exposes delayed clerical posting: if rolls are physically unloaded on Monday morning (t_dock) but keyed into the ledger on Wednesday afternoon (t_rec), planners making decisions on Tuesday are looking at stale supply data.[S27][S31]
Crucially, a short delivery or a rejected roll must also be preserved in the supplier performance and replenishment dataset. As research on inventory systems with uncertain supplies notes, both demand and supply data can be censored when only truncated or completed quantities are recorded.[S22] If a storekeeper overwrites the original ordered quantity of 1,000 metres down to the 850 metres actually delivered so that the purchase order can be administratively closed without a variance flag, future lead-time and supplier-fill calculations will see a 100 percent complete delivery of 850 metres instead of a 150-metre supply shortfall.[S22][S31] The ledger must retain Ordered (1,000), Shipped (900), Received (880) and Accepted (850) as separate immutable fields on the receipt line.[S31]
Interlude · the cloth half-woven
Woven: sections 2.0 to 2.4Still to weave: 2.5 to 2.7
2.5
Define count scope and tolerances
Define count scope and tolerances. Starting point. Leads to Execute blind physical bin count (hide book qty). · illustrative
Even with disciplined receiving, physical stock and book records drift apart during daily operations.
Execute blind physical bin count
Execute blind physical bin count. From Define count scope and tolerances (hide book qty). Leads to Compare count to book at cut-off (submit result). · illustrative
Compare count to book at cut-off
Compare count to book at cut-off. From Execute blind physical bin count (submit result). Leads to Recount and check open movements (outside tolerance), Approver posts ledger adjustment (within tolerance). · illustrative
Recount and check open movements
Recount and check open movements. From Compare count to book at cut-off (outside tolerance). Leads to Assign root-cause reason code (variance confirmed). · illustrative
Assign root-cause reason code
Assign root-cause reason code. From Recount and check open movements (variance confirmed). Leads to Approver posts ledger adjustment (remediation logged). · illustrative
Approver posts ledger adjustment
Approver posts ledger adjustment. From Compare count to book at cut-off (within tolerance), Assign root-cause reason code (remediation logged). End point. · illustrative
Illustrative discrepancy-resolution workflow proposed by GFT, adapted from public-sector cycle-counting control procedures (S27, S0201, S0202). Separates count definition, blind physical entry, variance comparison, independent recount and transaction cut-off check, root-cause classification and approver sign-off before ledger adjustment.
Unit: proposed workflow stages. Status: illustrative. Sources: S27, S0201, S0202. Illustrative figures are teaching examples, not measured factory results.
2.5 · the bodyExhibit 08 above; its figures are illustrative where marked.
Even with disciplined receiving, physical stock and book records drift apart during daily operations. Operators issue an extra roll of interlining without logging a pick ticket, return a partial cone of thread to the wrong bin, misread a size label during bundle kitting or fail to record cutting-table scrap.[S15][S27] Waiting for an annual financial stocktake to discover those errors leaves the replenishment engine flying blind for fifty-one weeks of the year. Cycle counting replaces the once-a-year shutdown with a continuous, stratified audit process whose primary purpose is not merely to adjust the ledger balance, but to diagnose and eliminate the transaction failures that caused the drift.[S01][S27][S0203]
As documented in the North Carolina Office of the State Controller's cycle-counting procedures (S27, INV-44), the cycle-counting process assesses stock levels within an inventory organisation, links count frequency to ABC classifications and enforces structured discrepancy correction.[S27] For an apparel manufacturer, however, stratifying count frequency by annual monetary usage (ABC value) alone leaves a dangerous operational gap: a custom woven size label, a specific Pantone sewing thread or a uniform collar zip may represent less than 1 percent of annual purchasing spend (C value) while possessing 100 percent line-stop criticality if its book balance is overstated.[S01][S07] GFT therefore recommends a two-dimensional count-priority rule that assigns the highest cycle-count frequency to items that are either high in annual usage value (Class A) or flagged as single-source, long-lead-time or line-stopping components in the bill of materials.[S01][S07][S27]
To prevent a cycle count from becoming a rubber-stamp exercise where warehouse staff simply copy the computer balance onto the count sheet, the workflow must enforce strict role separation and blind counting, drawing on the procedural controls detailed in S27, S0201 (INV-15 Enter Cycle Count Results) and S0202 (INV-28 Approve Cycle Counts):[S27][S0201][S0202]
Warehouse Manager role): Defines the count schedule, ABC assignment groups, item-location scope and acceptable quantity and value variance tolerances before generating count sequences.[S27]Warehouse Counter role): The counter receives a count sequence listing the warehouse organisation, zone, bin, item code, lot and unit of measure, but omitting the current book quantity. The counter physically measures or counts the handling units in the bin, enters the count quantity and submits the sequence.[S0201]Cycle Count Approver role): Before any ledger adjustment posts, the approver must review the cycle-count variance report, compare the quantities, assign a mandatory root-cause reason code (such as unrecorded issue, misplaced bin location, unit-of-measure conversion error, unlogged scrap or receiving tally error) and formally approve, reject or return the sequence for recount.[S0202]Reporting cycle-count results also requires mathematical care. A single "99 percent net financial inventory accuracy" headline can conceal catastrophic operational errors if positive and negative bin discrepancies cancel each other out. Empirical studies of inventory record accuracy in fashion and apparel (S15, S16) emphasise evaluating physical count deviations at the granular record level rather than relying on netted monetary totals.[S15][S16] GFT therefore requires every cycle-count report to publish three separate metrics across the $N$ counted item-lot-bin records:
GAUD): Defined for each unit-of-measure group as $\sum_{i=1}^{N} |Q_{\text{phys},i} - Q_{\text{book},i}| / \sum_{i=1}^{N} Q_{\text{book},i}$. Because the numerator uses absolute values, a $+50$ piece error in Size M and a $-50$ piece error in Size L adds to a $100$-piece absolute error rather than netting to zero.[S15][S16]When a bin count reveals a missing roll (-1 in Bin A) and a simultaneous surplus of the exact same roll serial ID in an adjacent rack (+1 in Bin B), the approver should classify the event as a location transfer error rather than a stock loss, correct the bin location and retrain the put-away team. Treating location errors and physical losses separately prevents false shrinkage alarms while still fixing the bin records that pickers rely on.[S27][S0202]
2.6
When managers confront recurring scan omissions or slow cycle counts, vendors frequently propose replacing barcode labels with radio-frequency identification (RFID) as an automatic cure for inventory inaccuracy. A disciplined engineering evaluation must resist that shortcut. Both barcode scanning and ultra-high-frequency (UHF) passive RFID are physical signal-capture mechanisms subject to distinct error modes, environmental interference, unit economics and reconciliation overhead.[S15][S16][S31] Neither technology guarantees stock truth by itself, and published studies from retail stores or brand-to-retailer distribution centres answer different questions that must not be conflated into a factory purchase justification.[S15][S16][S31]
A printed 1D or 2D barcode on a fabric roll ticket, thread carton or garment bundle is an optical line-of-sight token. Its strengths are low unit media cost, unambiguous one-to-one targeting (the operator knows exactly which roll or carton the laser or imager just read), and immunity to electromagnetic reflection from metal racking or adjacent bins. Its operational weaknesses are human and mechanical: every roll or carton must be physically oriented so the label is visible, high-speed counting requires manual handling of each unit, thermal labels can scuff or peel in humid dyehouse or cutting environments, and operators under time pressure may scan one carton label and manually type "$\times 20$" on the keypad rather than scanning all twenty cartons.[S15][S27]
Passive UHF RFID eliminates line-of-sight presentation by reading tag antennas in bulk via radio waves, reducing the physical time needed to sweep a rack of tagged garments or pass a carton pallet through a portal.[S15][S16] However, deploying UHF RFID inside a garment factory or fabric store introduces four engineering challenges that require empirical measurement:
false negatives): Dense stacks of folded garments, high-moisture cotton bales, metallic buttons or zips, foil packaging and metal storage racks absorb, detune or reflect UHF signals. A wand sweep that moves too rapidly past a dense rack will miss tags buried in the centre of a stack, causing the cycle count to understate physical stock and create false negative variances.[S15][S16]false positives): Because radio waves propagate through cardboard and across aisle boundaries, a handheld reader or dock portal tuned at high power can inadvertently read tags sitting in an adjacent bin, a quarantine cage two metres away or a forklift passing behind the operator. In a ledger that separates quarantine, available and reserved bins, stray reads corrupt location and status boundaries.[S15]EPCs) within a single scan window, printing a duplicate tag or attaching a replacement tag without decommissioning the damaged original causes one physical roll or garment to be counted twice.[S15][S31]Three published studies in the paper's evidence register illustrate how carefully external RFID findings must be interpreted:
S15, 2013): This fashion-retail study explicitly compares the quality of an RFID cycle count against a physically conducted count as its starting point.[S15] Its methodological lesson is that an RFID scan is a noisy measurement instrument whose read completeness and error rate must be benchmarked against a verified physical baseline, never assumed to be 100 percent accurate.[S15]S16, 2015): This apparel-retail field study empirically derives a trend of inventory accuracy as a function of the time required to conduct RFID inventory counts.[S16] It demonstrates an operational trade-off: shortening the time spent sweeping a zone reduces read completeness, so count speed and resulting inventory accuracy must be evaluated jointly rather than treated as independent benefits.[S16]S31, 2018): The GS1 US and Auburn University RFID Lab Project Zipper study examined data obtained from eight brand owners and five retailers over one year to reconcile brand shipments against retail receipts and purchase orders.[S31] That study addresses multi-party supply-chain data exchange and order fulfilment discrepancies at the shipping/receiving boundary; its findings do not measure internal factory raw-material bin counting or fabric-roll length tracking.[S31]Exhibit 9 places these three sourced studies on a qualitative two-dimensional scope matrix: the horizontal axis distinguishes single-site internal stock counting from multi-party shipment-to-receipt exchange, while the vertical axis distinguishes count-speed curve analysis from discrepancy-audit and reconciliation design. The coordinates (0 to 10) are qualitative layout positions that map each study's analytical boundary, not numerical accuracy percentages.[S15][S16][S31]
The coordinates (0 to 10) are qualitative layout positions that map each study’s analytical boundary
S15, S16, S31
Read-out
Qualitative comparison of analytical scope across three opened sources: S15 (2013) compares RFID cycle-count quality to a physical count in fashion retail; S16 (2015) derives inventory accuracy as a function of count duration in apparel retail; S31 (2018) reconciles ship and receive data across eight brand owners and five retailers. Coordinates are qualitative layout indices mapping operational boundary (x) against primary evaluation focus (y), not measured accuracy scores.
Unit: qualitative study scope index (0 to 10). Status: sourced. Sources: S15, S16, S31.
For GFT's Cameroon textile programme, the practical implication of Exhibit 9 is clear: barcode scanning on handling units (fabric rolls, thread cartons, cut-bundle tickets and finished-goods cartons) is the mandatory baseline capture method for Phase 1, while UHF RFID is a candidate option that requires a bounded local trial before any capital commitment.[S15][S16][S31] A local trial must test actual fabric roll density, warehouse racking geometry, operator sweep speed, false-positive stray reads across adjacent bins, tag failure rates in humid conditions and total cost per thousand reads against a controlled physical baseline.[S15][S16]
2.7
A stock ledger designed for an emerging-market factory must assume that electrical power or local wireless connectivity will occasionally drop in the middle of a working shift. If the warehouse management interface requires a live, synchronous round-trip to a central server before a storekeeper can receive a truck or issue fabric rolls to the cutting room, a two-hour network outage forces an unacceptable choice: either halt physical material movement and starve the sewing lines, or wave the rolls through on informal paper notes that get lost, misread or keyed in the wrong sequence hours later.[S27][S31][S0203] Conversely, poorly designed offline caching can corrupt the ledger even faster than paper notes if reconnected mobile terminals overwrite current bin balances or post duplicate deductions.[S31]
While WP07 owns site power and edge hardware selection and WP18 owns network and software transport architecture across the GFT white-paper library, WP12 specifies the inventory-specific data invariants required to keep the stock ledger consistent during and after an connectivity interruption:[S07][S27][S31]
event_id) constructed from the registered device ID, a strictly monotonic local sequence counter, the operator role ID and the local hardware clock timestamp (t_event).[S31]SET Bin_A12 = 14 rolls). Instead, every offline action is recorded in a local non-volatile append-only log as a signed movement delta tied to a specific handling-unit serial ID or lot (ISSUE Roll_ID R-0492, Lot L-08, Qty 88.5 m from Bin_A12 to WorkOrder WO-204 at t_event).[S07][S31] Because signed movements of distinct handling units commute, two operators working in the same zone during an outage can synchronise their logs later without one terminal overwriting the other's deductions.t_event versus t_sync): When connectivity resumes and the terminal uploads its queued log, the ledger server records the ingestion timestamp (t_sync) alongside the original physical movement timestamp (t_event).[S30][S31] Ordering the historical ledger by t_event reconstructs the true physical sequence of dock receipts, quality releases and line issues, while the gap t_sync - t_event measures offline lag.event_id: if a retried batch contains an event_id already present in the committed log, the server acknowledges receipt and suppresses the duplicate without deducting stock a second time.[S31]sync_pending = true), any item-bin in that zone is flagged as reconciliation-pending: cycle-count variance approvals (S0202) are held in queue until the terminal synchronises and the cut-off comparison (t_count versus t_event) is re-evaluated, and unconfirmed offline picks are deducted conservatively before new available-to-promise commitments are approved.[S07][S27][S0202]GFT note
Never allow an operator or automated script to post an unexplained "miscellaneous stock adjustment" to force a book balance to match a physical bin during or after a connectivity outage. Require every offline terminal in the affected warehouse zone to complete synchronisation, deduplicate by event_id, sort movements by physical t_event and re-run the cut-off check before the Cycle Count Approver reviews any residual discrepancy.[S27][S31][S0202] If an unallocated balance remains uncertain because a terminal cannot be recovered before an urgent production release, verify the specific roll or carton serial IDs physically at the bin and record a provisional reservation against those verified serials rather than overriding the book ledger.[S07][S15]
This stock-truth contract forms the foundation for the rest of WP12 and its interfaces across the GFT research library and tool suite. For WP05 (traceability and provenance), the handling-unit serial ID, dye-bath lot and append-only event_id log provide the internal chain of custody linking incoming fibre or fabric lots to finished garment packs, feeding the trace-a-garment tool without asserting that end-to-end field traceability is already qualified.[S07][S31] For WP14 (data and model governance) and WP18 (digital architecture), the separation of t_event and t_sync, idempotent event ingestion and role-segregated discrepancy approval define the data-hygiene gates that any replenishment algorithm must check before generating an order.[S27][S31] Most directly, the safety-stock tool and the segmentation and replenishment rules in Chapter 3 through Chapter 8 consume unreserved usable stock (U_free) and accepted usable stock (U_acc) rather than raw book balances.[S01][S05][S07] Once the factory can prove what stock it actually holds, in what unit, in what quality state and under whose reservation, it is ready to segment commercial promises and choose differentiated stock policies in Chapter 3.[S01][S05]
Check the record
Every knot is a source this chapter cites; select one to read its record, or filter the register by grade. Publication dates and inherited-source limits stay explicit. Read the paper's full evidence summary.
22 sources · 229 citation marks on this pageMost cited: S07, 47 times
Grade AGrade BGrade CKnot size: citations on this page
All 22 records shown
R01 · Grade C · inherited record
Inherited evidence R01
Cameroon seed-cotton production 2022-2024: 330,000 t (2022), 314,455 t (2023), about 340,000 t (2024).
Publication date unrecorded. Inherited source-text record. The original publication has not been independently reopened for this web edition.
R02 · Grade C · inherited record
Inherited evidence R02
2025 seed-cotton projection above 350,000 t.
Publication date unrecorded. Inherited source-text record. The original publication has not been independently reopened for this web edition.
R06 · Grade C · inherited record
Inherited evidence R06
About 95 percent of cotton fibre exported raw against 5 percent processed locally; industrialisation target of 50 percent local processing.
Publication date unrecorded. Inherited source-text record. The original publication has not been independently reopened for this web edition.
R07 · Grade C · inherited record
Inherited evidence R07
SND30 restates the 50 percent local-processing target by 2030.
Publication date unrecorded. Inherited source-text record. The original publication has not been independently reopened for this web edition.
R29 · Grade C · inherited record
Inherited evidence R29
Manual defect inspection 60 to 70 percent accurate, missing 20 to 30 percent of defects; computer vision above 90 percent across more than 40 defect types; line-scan inspection up to 60 metres a minute.
Publication date unrecorded. Inherited source-text record. The original publication has not been independently reopened for this web edition.
c02-S01 · Grade A
Supply chain fundamentals and segmentation analysis
This resource contains lecture slides on supply chain fundamentals and segmentation analysis.
2006. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c02-S05 · Grade A
Inventory management IV: Probabilistic demand and safety stock
This resource contains lecture slides on inventory management, probabilistic demand, and safety stock.
2006. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c02-S07 · Grade A
Inventory management VII: MRP and DRP systems
This resource contains lecture slides on inventory management and MRP and DRP systems.
2006. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c02-S09 · Grade A
IAS 2 Inventories
Inventories are measured at the lower of cost and net realisable value.
n.d.. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c02-S13 · Grade A
Trade facilitation
The TFA contains provisions for expediting the movement, release and clearance of goods, including goods in transit.
n.d.. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c02-S15 · Grade B
The Performance and Reliability of a RFID Cycle-Count: A Quantitative Approach from Fashion Retail
As a starting point, we compare in the present study the quality of a RFID cycle-count to a physically conducted count.
2013. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c02-S16 · Grade B
The impact of RFID technologies on inventory accuracy in the apparel retailing: Evidence from the field
From those findings, a trend of the inventory accuracy as a function of the time required for RFID inventory counts is empirically derived.
2015. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c02-S20 · Grade A
OECD Due Diligence Guidance for Responsible Supply Chains in the Garment and Footwear Sector
This Guidance will also support enterprises to implement the due diligence recommendations contained in the UN Guiding Principles on Business and Human Rights.
2018-03-07. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c02-S22 · Grade B
Learning to Order for Inventory Systems with Lost Sales and Uncertain Supplies
A key challenge in this learning problem is that both demand and supply data can be censored; hence only truncated values are observable.
2023-10-31. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c02-S26 · Grade A
The destruction of returned and unsold textiles in Europe’s circular economy
Returned and unsold textiles are a challenge and expensive to handle for the industry.
2024. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c02-S27 · Grade A
INV-44 Cycle Counting Process
The cycle counting process is used to assess stock levels within an inventory organization and correct any discrepancies.
2023-10-24. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c02-S29 · Grade A
Incoterms 2020
It recognises that not all commercial trade transactions from the seller to the buyer are conducted by a third-party carrier.
2020. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c02-S30 · Grade A
Time Release Study Guide: Version 4
Improvements have been made to questionnaire design and sampling techniques to ensure data accuracy.
2025. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c02-S31 · Grade B
EPC/RFID Data Exchange Study: Project Zipper Executive Summary
During the one-year study, Auburn University’s RFID Lab team examined the data obtained from eight brand owners and five retailers.
2018. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c02-S0201 · Grade A
INV-15 Enter Cycle Count Results
Once submitted, the Cycle Count Approver will be notified to approve, reject, or request a recount of this cycle count.
2023-07-26. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c02-S0202 · Grade A
INV-28 Approve Cycle Counts
The approver must review the NC Cycle Count variance report and compare the quantities before choosing an action.
2023-07-27. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c02-S0203 · Grade A
Inventory management VIII: Inventory management and optimization in practice
This resource contains lecture slides on inventory management and optimization in practice.
2006. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗S01 MIT OpenCourseWare. 2006.
S05 MIT OpenCourseWare. 2006.
S07 MIT OpenCourseWare. 2006.
S27 North Carolina Office of the State Controller. 2023-10-24.
S31 GS1 US and Auburn University RFID Lab. 2018.
S0203 MIT OpenCourseWare. 2006.
R06 Cameroon Trade Hub. Publication date unrecorded; inherited source-text check.
R07 Ecofin Agency. Publication date unrecorded; inherited source-text check.
S09 IFRS Foundation. n.d..
S20 OECD. 2018-03-07.
R29 MDPI; industry inspection literature. Publication date unrecorded; inherited source-text check.
S15 Association for Information Systems Electronic Library. 2013.
R01 Business in Cameroon. Publication date unrecorded; inherited source-text check.
R02 Fibre2Fashion; Business in Cameroon. Publication date unrecorded; inherited source-text check.
S26 European Environment Agency. 2024.
S13 World Trade Organization. n.d..
S30 World Customs Organization. 2025.
S29 International Chamber of Commerce. 2020.
S22 arXiv, Chen, Jiang, Zhang and Zhou. 2023-10-31.
S0201 North Carolina Office of the State Controller. 2023-07-26.
S0202 North Carolina Office of the State Controller. 2023-07-27.
S16 University of Modena and Reggio Emilia institutional repository. 2015.
Keep the thread · chapter 02 of 14
Move the shuttle along the rack to preview any chapter: its sections, exhibits and reading time. Draft web edition. This research is not investment, legal or tax advice; illustrative models are not measured factory results.
The paper's rack · 14 boltsBolt length: reading time
01Part I: Define the stock decision before buying the stock
GFT argues for measured, cash-constrained inventory decisions that protect usable material and customer commitments without manufacturing surplus.
02Part I: Define the stock decision before buying the stock
An inventory policy fails when it orders against quantities that are wrongly identified, reserved twice or unavailable for use.
03Part I: Define the stock decision before buying the stock
A high-value fashion line, a low-cost line-stopping trim and a repeat uniform fabric need different decision rules.
04Part II: Match replenishment rules to material and demand
The relevant lead time ends when approved material is usable, not when a supplier announces that it has shipped.
05Part II: Match replenishment rules to material and demand
Calculate protection against uncertainty with consistent time units, explicit service definitions and checks for non-normal or dependent demand.
06Part II: Match replenishment rules to material and demand
Order quantity must reflect holding costs, pack sizes, supplier minimums, time-phased needs and available cash.
07Part II: Match replenishment rules to material and demand
Limited selling windows make overage, underage and replenishment opportunities central to the order decision.
08Part III: Resolve constraints across the supply chain
Inventory optimisation must respect dependent material requirements, quality release and finite conversion capacity.
09Part III: Resolve constraints across the supply chain
Centralisation and postponement can reduce duplicated buffers while increasing transit time, concentration and conversion risk.
10Part III: Resolve constraints across the supply chain
Procurement should price usable delivery and shared risk rather than reward low quotes that create expensive disruption.
11Part III: Resolve constraints across the supply chain
A decision that reduces expected shortages may still be unaffordable once deposits, transit stock and receivables are included.
12Part IV: Prove the policy before widening its authority
A small transparent policy engine is more useful than an opaque optimiser that cannot explain missing data or infeasible orders.
13Part IV: Prove the policy before widening its authority
Backtests, simulations and a controlled operating pilot answer different questions and must be reported separately.
14Part IV: Prove the policy before widening its authority
GFT would implement a phased inventory decision system whose authority grows only with measured reliability, service and cash evidence.