Our approach / Warp and Signal
Measure,
then model.
A model is only useful when its inputs belong to the factory, and its output changes a decision.
Reading pick 01 of 04
Decision crossingWhat is a unit?
A good garment, a length of fabric, or a mass of fibre.
What is the time boundary?
Planned production time, running time and excluded time.
Who owns the event?
The person who records it and the person who can resolve it.
How trustworthy is it?
Calibration, missing events, rejected readings and revision history.
Pick 01 of 06 / Start at the crossing
Production makes the material. Measurement makes it legible.
The warp carries material through the factory. The weft carries the evidence needed to improve it. We design their crossings before specifying the technology.
4 of 204crossings kept
Production. The warp carries material through the factory. No evidence is collected yet.
Measurement. Each pick of weft crosses every warp thread. Crossings multiply faster than anyone can use them.
Decision. Each kept crossing was chosen because a decision depends on it.
Crossing 01 of 04
What is a unit?
A good garment, a length of fabric, or a mass of fibre.
Decision it supports. Which process is being compared?
What is the time boundary?
Planned production time, running time and excluded time.
Decision it supports. Where was productive capacity lost?
Who owns the event?
The person who records it and the person who can resolve it.
Decision it supports. Who checks and acts on the exception?
How trustworthy is it?
Calibration, missing events, rejected readings and revision history.
Decision it supports. Can the signal support an intervention?
Conceptual drawing: a measurement point is chosen for a decision. The counts are a drawing, not a measurement.
Pick 02 of 06 / Make the denominator explicit
A useful number has a boundary.
A figure without a time window, material basis and accountable owner is difficult to act on. Define the measurement contract before building a dashboard. The four crossings from the loom become four questions.
Example contract 01 (illustrative wording, not a GFT standard)
One garment that passes final inspection counts once. A garment sent back for rework counts when it passes, not before.
Missing: No rule for rework, so the same garment is counted twice, or not at all.
Example contract 02 (illustrative wording, not a GFT standard)
The window is the planned production time of the shift. Each excluded stop is listed with its reason.
Missing: Stops are left out without a reason, so lost capacity cannot be traced.
Example contract 03 (illustrative wording, not a GFT standard)
The line supervisor records the event. The maintenance lead resolves it. Both are named by role.
Missing: No one is accountable for the exception, so it is noted and left.
Example contract 04 (illustrative wording, not a GFT standard)
Every reading carries a calibration date, a missing-event flag and a revision note.
Missing: Gaps are filled silently, so a poor signal looks as good as a sound one.
Complete. The figure can support an intervention.
Incomplete. Treat the figure as commentary, not as a signal.
The ring marks the constraint. The dashed line is the agreed baseline.
Three threads woven together: the routine, the data contract and the test of success.
One bounded line inside the dashed box, with its fallback drawn back to the start.
Hover or focus a mark on the drawing to read it.
First failure: a stop outside the planned window. It stays in the record and is not averaged away.
Second failure: a rework loop. It stays in the record, so the comparison stays honest.
The gap between the pilot and the baseline is what has to be explained, under like conditions, before anything expands.
Pick 03 of 06 / Follow the signal
Diagnose. Design. Deploy. Prove.
Each phase earns the next. A pilot that fails its operating test changes the design before a wider rollout. Scroll, and the twill changes with every phase.
Phase 01, Diagnose
Observe the constraint.
Map the process, define the customer requirement and agree a baseline. Distinguish an observation from an estimate and a policy target.
Exit test: a baseline the team accepts.
Phase 02, Design
Choose the smallest useful system.
Specify the operating routine, the data contract and the test of success together. The architecture follows the decision and the conditions on the floor.
Exit test: the test of success is written before the build.
Phase 03, Deploy
Test inside the operating boundary.
Introduce the process on a bounded line or use case. Keep a fallback, a named owner and a record of what changed.
Exit test: a named owner and a working fallback.
Phase 04, Prove
Compare like with like.
Review the result against the agreed baseline and operating conditions. Retain the failures. Expand only when the evidence supports the next decision.
Exit test: failures retained, expansion decided on evidence.
Pick 04 of 06 / A model earns its place
Compute the loss before prescribing the remedy.
Overall equipment effectiveness separates the time a process was available, the rate at which it ran and the share of output that was good. The model is useful because each factor leads to a different question.
Product Effective length
Limiting factor: Time
Ask this first
Why was the line not running?
Why did it run below its ideal rate?
Why was output rejected?
87% × 92% × 96% = 76.8%
Illustrative inputs, not a GFT result. The factors multiply, so one low factor lowers the whole product.
Pick 05 of 06 / The method joins the programme
Feasibility and digital design share an operating model.
GFT's Cameroon textile programme connects the commercial question to the factory routine and the data architecture. Scroll or pick a thread to trace what each part hands to the next.
What the feasibility study hands onHands the operating assumptions to the performance system, so the indicators test what the plan relies on.
What the performance-management system hands onHands the indicators and review routines to the digital design, so the architecture serves a routine that exists.
What the digital strategy hands onHands the power, connectivity and team limits back to the feasibility study, so the plan stays inside what can be run.
Pick 06 of 06 / Keep the evidence attached
An ambition is not an operating result.
Prepared, proposed, piloted, measured and operating describe different states. A target stays a target until evidence supports the change in status. Drag the status or choose a rung.
The programme is prepared.
Written, reviewed and sourced. Nothing has been built or run.
The programme is proposed.
Put to a decision-maker with its conditions and a baseline. Not yet accepted.
The programme is piloted.
Run on a bounded line with an owner and a fallback. Early, and not a result.
The programme is measured.
Compared with the baseline under like conditions. The failures are kept.
The programme is operating.
Running in routine production, with its measure reviewed by its owner.
Where GFT's Cameroon textile programme stands: prepared work, not an operating result.
Next evidence: a decision-maker accepts the conditions and the baseline.
Read the measurement contractNext evidence: a bounded line, a named owner and a fallback.
Follow the four phasesNext evidence: a comparison with the agreed baseline.
Move the OEE factorsNext evidence: the measure is reviewed in routine production.
See how we workNext evidence: the review continues to change decisions.
Return to the first threadMeasure, then model. The strongest signal can be the one that changes the plan.