Research libraryMeasure Before You Model
Chapter 03 of 14 · Part I · Draft web edition
Strategic intent becomes an operating system when each line measure carries a decision, an owner, a feasible record and a guardrail.
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 GFT design. NIST supports selection by organisational purpose; ISO supports operational definition. Nodes and links are proposed responsibilities, not an observed implementation.
Unit: illustrative decision sequence. Status: illustrative. Sources: S0301, S01. Illustrative figures are teaching examples, not measured factory results.
Overture · 1 min
A factory should cascade decisions and responsibilities before it cascades targets. The line needs to know which result matters, which observed condition requires a response, what action it may take and when the problem exceeds its authority. GFT proposes a short scorecard covering production, quality, delivery, cost, people and sustainability, with explicit connections between strategic outcomes and controllable work. The proposed architecture below is a design for a greenfield garment factory, not evidence that a GFT installation has achieved these conditions. Its purpose is to make the operating promise inspectable before that promise becomes a performance judgement.
3.1
Begin with the factory's operating promise: supply conforming garments against an agreed requirement while controlling cost, protecting people and accounting for resources. Translate each part of that promise into a question management must answer. Can the accepted order be completed with demonstrated capability? Is the current method producing conforming work? Which constraint needs intervention? A measure earns a place when it helps answer a question and changes the next decision. A count that attracts interest but carries no consequence belongs in diagnostic analysis until a useful decision has been identified.
NIST's manufacturing performance-assurance report connects measurement selection to organisational objectives and policies. The ISO manufacturing KPI abstract also describes indicators through their calculation elements, time behaviour and units. These are complementary starting points: purpose determines admission, while operational definition determines whether the resulting signal can be interpreted. Neither source establishes that every available measure should appear on a line board.[S0301][S01]
GFT's proposed admission test asks the sponsor of each measure to identify its strategic parent, decision owner, source event and available response. Consider a proposed efficiency tile. The sponsor must explain whether it supports capacity planning, method improvement or immediate line balancing. If nobody can distinguish these uses, the tile is premature. A proposed absenteeism measure must similarly distinguish support planning from individual performance assessment. The same record can support different questions, but those questions require separate permissions, interpretation and routes for challenge.
Collection burden belongs in the admission decision. GFT would ask an operator and a supervisor to demonstrate how a proposed record is completed during ordinary work, including an interruption and a correction. The design should preserve necessary production context without requiring workers to reconstruct the shift from memory. A measure whose capture interrupts the process it is meant to improve needs redesign, a different collection point or a lower reporting frequency. This is a practical design test, not a claim that manual capture is always superior to sensing.
The illustrative cascade in Exhibit 11 preserves the connection from the operating promise to a permitted response. Information also travels upwards: a station observation can invalidate the assumptions behind the line plan. The arrows describe proposed responsibilities rather than a causal model with verified coefficients.
The cascade needs a retirement rule as well as an admission rule. GFT would remove a routine tile when its decision disappears, its source cannot be maintained or another measure answers the same question more directly. Retain historical records under the applicable retention policy rather than deleting inconvenient evidence. A retired indicator may remain useful for investigation, but its removal from a board should be explicit. This prevents a growing archive of former priorities from becoming an ever larger demand on current production.
An unresolved strategic choice should remain visible instead of being disguised as a missing KPI. If management has not decided which order has priority when capacity is constrained, another output chart cannot resolve the conflict. GFT would ask the accountable manager to approve the priority rule and its exception route. The line can then interpret a local shortfall against an agreed plan. Until that decision exists, performance reporting should describe the constraint and competing requirements without implying that a supervisor has been given a coherent instruction. This separates a measurement problem from an unresolved business decision.
3.2
Spend commitment
Spend commitment: Timing: in-process (0) to end (1) 0, Finance 0, behaviour 1, process 2 0 · illustrative
The ILO’s ready-made garment guidance recommends balance across financial, behavioural and core-process metrics, alongside a mixture of in-process and end-of-process measures.
Closed unit cost
Closed unit cost: Timing: in-process (0) to end (1) 1, Finance 0, behaviour 1, process 2 0 · illustrative
Help response
Help response: Timing: in-process (0) to end (1) 0, Finance 0, behaviour 1, process 2 1 · illustrative
Action follow-up
Action follow-up: Timing: in-process (0) to end (1) 1, Finance 0, behaviour 1, process 2 1 · illustrative
Blocked work
Blocked work: Timing: in-process (0) to end (1) 0, Finance 0, behaviour 1, process 2 2 · illustrative
Accepted output
Accepted output: Timing: in-process (0) to end (1) 1, Finance 0, behaviour 1, process 2 2 · illustrative
Illustrative GFT classification using ILO distinctions. Horizontal coordinate 0 means in-process and 1 end-of-process. Vertical coordinate 0 means financial, 1 behavioural and 2 core-process. Coordinates are category codes, not weights, scores or measured distances.
Unit: illustrative category positions. Status: illustrative. Sources: S02. Illustrative figures are teaching examples, not measured factory results.
3.2 · the bodyExhibit 12 above; its figures are illustrative where marked.
The ILO's ready-made garment guidance recommends balance across financial, behavioural and core-process metrics, alongside a mixture of in-process and end-of-process measures. It also warns that excessive or insufficient numbers of metrics can impair improvement. GFT uses that principle to organise the proposed scorecard into production, quality, delivery, cost, people and sustainability. These families are the chapter's organising design, not a claim that the ILO prescribes this exact classification.[S02]
Production asks whether available work is becoming accepted output. Quality asks whether the method is producing conformity without correction. Delivery asks whether the remaining work can meet the agreed promise. Cost asks what the conversion consumes relative to the accepted result. People asks whether the work has suitable capability, support and safeguards. Sustainability asks what physical resources and waste cross the declared boundary. Each question has a different decision consequence. Combining them into a single weighted mark would conceal which consequence requires attention.
GFT would choose a small active set within these families for each management level. A station may need a clear acceptance condition, a visible abnormality signal and a route for assistance. It need not display the entire plant scorecard. The plant director, however, needs visibility across all the families before approving a recovery plan. Coverage means that a relevant consequence has an accountable review somewhere in the system. It does not mean that every person must calculate every indicator.
Treat relationships between families as hypotheses to test. A method change intended to raise throughput might reduce rework, leave it unchanged or create a new defect. A consolidation of orders might improve machine utilisation while extending the wait for another customer. GFT would record the expected mechanism and watch the affected measures over a comparable production cohort. An improvement claim should identify the boundary at which the gain appears and check whether a queue, rejected unit or resource demand has merely moved elsewhere.
The illustrative classification in Exhibit 12 distinguishes the timing of a signal from the kind of consequence it describes. The positions are category coordinates. Their distances and areas carry no analytical meaning. This deliberately avoids assigning an invented financial value to behavioural evidence or averaging unlike units into an apparent factory score.
A family can contain both kinds of timing. Resource use while a line is operating can support immediate investigation; the reconciled resource total supports period reporting. A skill observation during an operation can prompt assistance; a completed competence review supports future assignment. GFT would document these different uses instead of calling a metric inherently leading or lagging. Whether it warns in time depends on the decision deadline, not simply on how frequently a dashboard refreshes.
GFT would also test the completeness of a proposed recovery proposal across families before comparing its headline benefit. A production proposal should state its quality check; a delivery proposal should state the cost and work conditions it assumes; a resource proposal should preserve the required product specification. This does not require assigning an artificial monetary value to every effect. It requires naming an affected consequence and identifying who can judge it. Where the consequence cannot yet be measured reliably, record the uncertainty and choose a bounded trial rather than presenting the gain as settled.
3.3
Exhibit 137 paragraphs step it
Read-out
Illustrative arithmetic, not a factory forecast. Promise assumed 720. At 800 gross completions and acceptance 0.90, accepted output is 800 x 0.90 = 720. At the same gross count and acceptance 0.80, accepted output is 640. Assumptions exclude rework recovery and hold style and boundary constant.
Unit: illustrative accepted garments per window. Status: illustrative. Illustrative figures are teaching examples, not measured factory results.
The ILO's Factory Improvement Toolset report describes line supervisors balancing work before production through planning and during daily production. This supports assigning balancing decisions close to the work. It does not justify assigning every plant outcome to a supervisor or treating all missed output as operator underperformance. The line-level cascade must distinguish what can be changed locally from what requires resources or authority elsewhere.[S04]
GFT proposes a controllability map for each important gap. The supervisor can investigate blocked work, confirm the approved method, call technical support and rearrange work within verified competence and authorised practice. A production manager may change the allocation between lines. Purchasing owns the supplier commitment; maintenance owns the technical repair response; commercial management owns renegotiation of a customer promise. The local board should show the dependency and escalation owner even where the local team cannot remove the cause.
Output must remain connected to the relevant product boundary. A completed operation is not necessarily a completed garment, and an inspected garment may still be awaiting a release decision. GFT would require the line planner to state which boundary the current plan uses before comparing actual performance. The detailed counting contract belongs in the metric-definition chapter. The cascade's task is to ensure that the customer outcome and the operating driver refer to compatible work, rather than assuming a common label makes them equivalent.
The common denominator in Exhibit 13 is first-pass accepted garments over the same illustrative planning window. The chart compares the promise with alternative outcomes at an unchanged gross completion count. It does not compare gross output and accepted output as if they were the same measure.
This example also separates responsibility from blame. A supervisor should be responsible for detecting the accepted-output gap and initiating the authorised response. That responsibility does not establish the cause of a quality change. GFT would keep the observation, suspected mechanism and confirmed cause separate. Until the evidence supports attribution, the escalation should describe the condition and assistance required, preserving both accountability for action and fairness in interpretation.
The dependency map should remain useful when the physical line performs exactly as planned but the order still cannot ship. For example, a hypothetical batch may have completed sewing while awaiting an externally supplied trim. GFT would preserve the valid sewing result, show the unfulfilled release condition and route the shortage to its owner. Increasing the sewing target would not resolve that dependency. Conversely, a materials owner should not close the shortage merely because a purchase instruction exists. The completion condition must follow the physical event needed by the next responsible role.
3.4
The Federal Data Strategy playbook describes governance through priorities, authority, roles, policies and accountability. It addresses public administration rather than garment production. GFT adapts its accountability principle here; the factory roles below are proposed operating arrangements, not a regulatory requirement or a claim of compliance with that playbook.[S22]
Every KPI should have a business owner who answers for its purpose and permitted use. The data steward maintains the record rules, checks completeness and investigates inconsistent submissions. The review chair ensures that a decision is made or escalated. These responsibilities can be held by the same person in a small organisation, but the charter should keep them distinguishable. Otherwise a correction can be treated as a target revision, or a discussion can be mistaken for a completed operational response.
Ownership should also identify who may change the definition. GFT would let the steward propose a correction to a calculation or source mapping, but require the business owner to approve a material change in meaning. Planning and affected users should confirm how the change alters comparisons. Historical records need their original interpretation preserved, with any restatement made visible. This chapter establishes the decision right; the neighbouring metric-contract chapter specifies the detailed versioning and measurement checks.
The illustrative charter in Exhibit 14 shows how to attach accountability without creating an unmanageable table. Each row represents a field on the proposed accepted-output charter. The cadence is a proposed management choice. It should be adapted to the time available for a useful response and to the reliability of the source record.
Read-out
| Business owner | Production manager |
|---|---|
| Data steward | Production-record custodian |
| Review chair | Shift supervisor |
| Source and rule | Unique first-pass acceptances in shift log |
| Decision | Rebalance, assist or escalate |
| Cadence | Within shift, with shift-end reconciliation |
| Guardrail | Quality, competence and safe work protected |
| Target basis | Approved order plan and tested capability |
Illustrative GFT charter, adapted from accountability and operational-definition principles. Roles, cadence and guardrails are proposals. Accepted output counts unique first-pass acceptances at the declared end-of-line inspection boundary, with subsequent corrections retained.
Unit: illustrative charter fields. Status: illustrative. Sources: S22, S01. Illustrative figures are teaching examples, not measured factory results.
A production manager remains accountable for the business interpretation even when software computes the value. GFT would not make the software supplier the default owner of the factory's production promise. Equally, the steward should be able to flag an unreliable signal without being instructed to supply a convenient number. A missing inspection batch should produce an explicit incomplete status, a bounded explanation and a correction route. It should not silently become accepted output or a fabricated zero.
A deputy and an escalation route are part of ownership. Before a shift begins, the line should know who can authorise assistance when the usual owner is absent. A record correction should preserve the reason and the person approving it. A target dispute should reach someone with authority over the original assumption. These arrangements make the cascade usable under ordinary absences and disagreements, rather than depending on a particular manager being present whenever the data becomes inconvenient.
Approval rights should distinguish ordinary assistance from changes that create new obligations. A supervisor may request help under an agreed staffing arrangement, while the authorisation of additional expenditure belongs to the designated budget owner. A quality release follows the authorised acceptance route even when production is behind plan. GFT would put these distinctions in the charter as practical boundaries, then test them with a simulated exception. If the proposed response depends on authority that nobody actually holds, the measure is not ready for operational use regardless of how accurately its value can be calculated.
Interlude · the cloth half-woven
Woven: sections 3.0 to 3.4Still to weave: 3.5 to 3.7
Read-out
Illustrative worked example. 64 OTIF orders + 16 orders not OTIF = 80 orders due. OTIF rate = 64 / 80 x 100 = 80 percent. These are assumed values used to demonstrate the calculation.
Unit: illustrative orders and percent. Status: illustrative. Illustrative figures are teaching examples, not measured factory results.
Read this exhibit in its chapter ↗3.5
Related · chapter 078 paragraphs step it
Read-out
Illustrative planning sequence. S42 supports at-least-monthly management review at product-family level. The four weekly windows are assumed to show a possible sequence, not required deadlines or realised milestones.
Unit: weeks in monthly cycle. Status: illustrative. Sources: S42. Illustrative figures are teaching examples, not measured factory results.
Read this exhibit in its chapter ↗GFT proposes that a target be measurable, understandable, verifiable, equitable and achievable under its stated conditions. These are recommended acceptance criteria for this design. They do not establish an employment rule, a legal standard or a guarantee of factory capability. A target should state the required result, the relevant work population, the decision window and the evidence supporting feasibility. Without those conditions, apparent precision is a management preference written as a number.
Separate the customer requirement from demonstrated capability and the management improvement objective. The customer requirement states what has been agreed. Capability describes what comparable work has actually achieved under recorded conditions. The improvement objective proposes a change and the intervention expected to support it. A gap between requirement and capability needs a planning decision. Relabelling the requirement as a capability target does not supply missing materials, competence, maintenance or time.
GFT would classify each target's basis explicitly. External evidence provides context whose scope and transferability must be checked. Internal arithmetic converts agreed inputs into a requirement, as in the worked example. A planning assumption supports a scenario pending observation. An internal baseline supports comparison only for its declared population. These categories should remain visible in the target register. A sourced ambition is still an ambition; an illustrative scenario remains illustrative even when a plausible formula produces it.
A target should expire when its stated basis no longer holds. A style change, a revised method, a different inspection boundary or a changed staffing condition should trigger a review of comparability. The owner may retain the customer commitment while changing the recovery plan. Alternatively, management may need to renegotiate the commitment. Neither response permits rewriting the old record. The scorecard should show the original requirement, the changed assumption and the authorised revision separately.
The review rule should avoid both automatic relaxation and automatic tightening. An adverse result may expose a weak method, an unreliable record or a missed dependency; it does not automatically prove that the target was wrong. A favourable result may reflect an easier mix or a temporary support arrangement; it does not automatically justify a permanent increase. GFT would require an explanation of changed conditions before either decision, preserving the distinction between learning from evidence and rewarding a convenient interpretation.
During early production, the target register should also expose uncertainty. A provisional planning band can be useful if its basis is stated, but its edges are not statistical control limits unless a valid analysis establishes them. The cascade should record what observation would narrow the uncertainty and who owns obtaining it. This keeps target setting tied to the development of evidence, while leaving the detailed study of variation to the measurement and improvement chapters.
Target review should preserve the original decision deadline. Revising a plan after the relevant production window has closed cannot improve that completed window; it changes only how the result is presented. GFT would keep any retrospective explanation separate from the prospective instruction for the next comparable work. The owner should record when the changed information became available and what response was still possible at that time. This gives management a fairer basis for judging decisions under uncertainty and prevents later knowledge from being treated as something the line should already have possessed.
3.6
Read-out · illustrative
Illustrative: GFT hour-by-hour line-board layout for one 8-hour sewing shift (40 operators, 24.0 SMV woven trouser, ramp cohort week 4 at 65 percent target efficiency). Values are illustrative design entries showing how balanced in-process and end-of-process measures (S02), daily line balancing (S04) and unmerged downtime causes (S07) appear with an explicit owner and response window. Not an observed GFT factory result.
Unit: illustrative shift-board control blocks. Status: illustrative. Sources: S02, S04, S07. Illustrative figures are teaching examples, not measured factory results.
Read this exhibit in chapter 11 ↗A review of the Balanced Scorecard identifies criticism of its presumed rational, top-down strategy execution, including its fit with decentralised or adaptive organisations. That is a reported criticism of a framework, not proof that every cascade fails. GFT's design response is to let the people doing and supervising the work test the operational assumptions before measures become fixed performance judgements.[S46]
A station test should ask the worker to explain what event changes the displayed number. The supervisor should explain what can be done when the number is adverse. The steward should explain how a correction reaches the record. If these explanations conflict, the design has not yet reached a shared operational meaning. GFT would resolve that disagreement at the work location, using the actual record and method, before adding another layer of reporting or a more elaborate visual display.
The feedback route should distinguish a temporary exception from a proposed definition change. A broken machine is an operating event. A counting rule that treats a restarted garment as a new completion is a definition problem. A target that assumes unavailable competence is a planning problem. Routing each issue to the appropriate owner prevents workers from having to invent a local workaround simply to make the scorecard match their experience.
GFT would retain the original observation while a challenge is assessed. The challenge record should describe the work condition, the contested interpretation, supporting evidence and requested decision. The responsible owner acknowledges it, investigates with affected users and publishes the outcome. If the definition changes, the explanation should reach every affected shift. If the definition remains, the reason should be equally visible. A route that merely receives comments without returning decisions cannot correct the cascade.
A hypothetical pressing station illustrates the distinction. Suppose the completion mark is recorded before the garment has cooled enough for the specified inspection. Moving the mark might improve the apparent timing of output while leaving the release delay unchanged. GFT would ask production and quality to agree the counting boundary together, then keep any earlier operation-completion event as a separate diagnostic record. This is a constructed design case, not an observed GFT event or a universal prescription for pressing methods.
Worker voice also requires restraint in how records are displayed. GFT recommends shared line-level problem solving without public individual rankings whose basis workers cannot inspect or challenge. Access to person-level information should follow its necessary use and the applicable local requirements. This chapter offers operating-design recommendations, not legal advice. The practical question for the cascade is whether the proposed display helps authorised assistance and improvement, or merely transfers uncertainty into individual blame.
Local knowledge should test the strategic model as well as the metric. If a recurrent observation shows that an assumed driver does not explain the outcome, GFT would revisit the link. The response may be a better measure, a changed intervention or retirement of an unsupported assumption. The scorecard remains aligned to strategy by correcting its theory of how work produces results. Alignment should not require the line to defend a relationship that the operating evidence repeatedly contradicts.
GFT would test the challenge route during commissioning with an intentionally ambiguous record. Ask the worker to identify the ambiguity, the supervisor to route it and the steward to show how the eventual decision reaches other users. The test passes only when the record can be corrected without destroying its history and the agreed meaning reaches the next shift. Record any failure as an operating-system defect with an owner. This rehearsal does not prove that all future disputes will be resolved, but it establishes whether the proposed route exists beyond the written policy.
3.7
A decision that changes production should bring its quality, delivery, cost, people and resource consequences into the same review. GFT proposes paired measures and explicit guardrails, rather than a weighted total that lets a favourable output result cancel a breach elsewhere. The organising balance follows the ILO guidance; the particular approval rule and pairings here are GFT's proposed design choices.[S02]
The illustrative responsibility layers in Exhibit 15 identify where an unresolved trade-off should go. They do not require a new committee at every layer. The relevant distinction is the scope of authority: assistance within an approved method, a cross-functional change to the operating plan, or a change to the business commitment and its resources.
Read-out
Illustrative GFT design, not an implemented reporting structure. Accountability follows governance principles; line balancing remains close to production. A guardrail issue follows the relevant authorised protection procedure at any layer.
Unit: illustrative responsibility layers. Status: illustrative. Sources: S22, S04. Illustrative figures are teaching examples, not measured factory results.
Before authorising a recovery action, GFT would require a short statement of its expected benefit, affected families, owner and stopping condition. Reallocating a competent operator may help a constraint but weaken another operation. Increasing release to a line may build work in progress without improving completion. Deferring maintenance may preserve current running time while leaving an unresolved technical risk. These are proposed questions for decision review, not quantified predictions of what any particular factory will experience.
A guardrail must name the response to a breach. If the approved method cannot be maintained, the responsible role follows the authorised protection and escalation procedure. If quality evidence is incomplete, affected output remains in its declared status pending the required decision. If a resource measure deteriorates, the owner investigates the relevant boundary and absolute use alongside intensity. Each condition needs an accountable response; a red tile on its own does not protect the process or resolve the problem.
The review should also distinguish a decision from its result. Authorising assistance is a recorded action. Observing a subsequent improvement is a separate event. Attributing the improvement to the action requires further evidence, including consideration of product mix and other changed conditions. GFT would keep these stages visible in the action record. The later improvement chapter develops verification in depth; here the cascade must ensure that the follow-up measure is chosen before the intervention loses its original context.
The resulting scorecard is ready to become a metric contract when its users can state the outcome, the controllable driver, the counting boundary and the decision right without inventing missing assumptions. It is ready for a management routine when the relevant owners can act within the required window. A later digital tool may assemble that draft and expose incomplete fields, but the owner still approves its meaning. The final test is practical: can the line take a defensible action, and can management see the consequences across the operating promise?
At handover, GFT would ask management to inspect a complete decision path rather than approve a slide of metric names. Select a real order requirement, trace the planned work and locate the underlying source event. Confirm who checks it, who acts and how an adverse consequence reaches the appropriate decision owner. Then test an incomplete record and a disputed assumption. The cascade should preserve uncertainty in both cases while still giving the line a usable next step. That is the acceptance condition for this chapter's design, before more detailed metric contracts and management routines are installed.
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.
6 sources · 13 citation marks on this pageMost cited: S02, 3 times
Grade AGrade BGrade CKnot size: citations on this page
All 6 records shown
c03-S01 · Grade A
ISO 22400-2:2014 Automation systems and integration: Key performance indicators (KPIs) for manufacturing operations management, Part 2: Definitions and descriptions
ISO 22400-2:2014 specifies a selected number of KPIs in current practice.
2014-01. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c03-S02 · Grade A
Lean Manufacturing Techniques for Ready Made Garments Industry
Use a standard definition form for your metrics.
2017. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c03-S04 · Grade A
Implementation of the Factory Improvement Toolset in the Asian garment sector: Final report
Line balancing is done by line supervisors systematically, before production through planning, and during production, daily.
2023. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c03-S22 · Grade A
Data Governance Playbook
Data governance is the process of setting and enforcing priorities for managing and using data as a strategic asset.
2020-07. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c03-S46 · Grade B
Balanced Scorecard: History, Implementation, and Impact
Addressing these issues requires a nuanced understanding of the framework’s limitations and the context in which it is applied.
2025-03-19. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c03-S0301 · Grade A
Methods and Tools for Performance Assurance of Smart Manufacturing Systems
The decision on what to measure stems from the objectives and policies of the organization.
2015-12. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗Keep the thread · chapter 03 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: The measurement contract
A factory without an operating baseline cannot rely on hindsight, so GFT treats measurement as part of production from the outset.
02Part I: The measurement contract
Early targets should describe a controlled learning trajectory, with yield, rate and capability observed together rather than assumed away.
03Part I: The measurement contract
Strategic intent becomes an operating system when each line measure carries a decision, an owner, a feasible record and a guardrail.
04Part I: The measurement contract
Operational definitions, data-quality controls and measurement-system analysis must precede target comparisons and automated analytics.
05Part II: Measure the operating system
Good output is the result, while the production system must retain the component losses that explain why the result occurred.
06Part II: Measure the operating system
The quality system should connect first-pass yield, defect causes and variation so that inspection leads to prevention rather than sorting alone.
07Part II: Measure the operating system
Standard work, balanced capacity and order-level flow records must support the delivery promise before expediting becomes routine.
08Part II: Measure the operating system
A low nominal wage does not establish a low cost per good garment, so the people scorecard must include skill, supervision, attendance and working-time effects.
09Part II: Measure the operating system
Energy, water, waste and emissions become manageable only when each quantity has a declared boundary, source record and production denominator.
10Part III: Turn records into decisions
Source capture, master data and explicit stewardship turn daily logs into an operational asset without requiring a heavy platform on day one.
11Part III: Turn records into decisions
Daily correction, weekly problem solving and monthly reconciliation give each signal a forum, an authority level and a recorded response.
12Part III: Turn records into decisions
A corrective-action system should distinguish abnormal variation, test causes, verify effects and update standard work rather than count completed actions.
13Part IV: Scale with control
The factory should earn each digital step through reliable records, active management and verified improvement rather than buy a platform to compensate for their absence.
14Part IV: Scale with control
GFT recommends four gated stages: clean capture, active cadence, verified improvement and a data-driven floor, with value measured in the same system throughout.