The first shift must make evidence as well as garments

The inherited evidence register records a substantial difference between Cameroon's cotton production and local transformation: about 95 percent of cotton fibre is exported raw and about 5 percent is processed locally. It also records a national ambition for 50 percent local processing by 2030, restated in the national development strategy.[R06][R07] These are approximate sector observations and a policy target, with the original publication dates not recorded in the inherited register. They are not current GFT operating measures.

Exhibit 2 places the baseline and the target beside one another without treating them as equivalent observations.

The exhibit shows the scale of the industrial proposition, not its commercial inevitability. Moving from raw export towards local processing requires more than additional equipment. Each added stage must convert inputs into accepted output under conditions that can be repeated, costed and improved. Weak operating control makes the gap harder to close because the organisation cannot locate where material, time and quality are lost.

The figures do not establish demand for any proposed garment factory. They say nothing by themselves about product specification, order size, customer eligibility, delivered cost or the route from national ambition to realised factory utilisation. This paper therefore does not convert the policy target into a sales assumption. It addresses a different question: if a greenfield operation is commissioned, what evidence must exist for its sponsors and managers to understand and govern production?

That distinction changes the design priority. A feasibility model can estimate a future operating condition from assumptions, but only actual records can test whether the condition has occurred. The first shift is where assumed work content meets observed work, where expected quality meets inspection outcomes and where planned resource use meets meter readings. A performance-management system should preserve those comparisons rather than overwrite the plan with an unexplained actual.

The strategic consequence is practical. Local value addition depends on learning that remains inside the operating organisation. If each problem is resolved through informal intervention and no durable record survives, the factory repeats the same uncertainty with each style and shift. If events, causes and responses are recorded consistently, the organisation can turn isolated corrections into reusable method knowledge.

A factory remembers through its records or through the recollection of individuals. The second route is fragile during ramp-up. People change roles, methods are revised and the explanation for an unusual shift becomes difficult to reconstruct. A contemporaneous record preserves what was known at the time, including uncertainty. It allows a later team to distinguish an observed event from a story assembled after the result was already known.

Garment-floor research shows that source capture is technically feasible. One study describes RFID systems recording front-line production data every day in varied, small-batch garment production.[S19] Another uses power monitoring attached to sewing machines to collect consumption data remotely and extract working times for sewing tasks.[S20] These studies demonstrate possible sources of operational evidence. They do not establish that either technology is the correct starting point for every new factory, nor that a sensor stream automatically carries the right operational meaning.

The record must precede the capture device. Before selecting RFID, power monitoring or another input, management needs to state which event matters and how it will be used. A machine becoming active is not necessarily a garment becoming good. A tag crossing a reader is not necessarily a unit completing an operation. A power pattern may support an inferred work interval, but the inference still needs validation against the actual method, product and quality outcome. Source capture reduces reconstruction only when the event definition is sound.

GFT would therefore begin with a disciplined manual record and automate fields where doing so improves timeliness or reduces demonstrated capture error. Manual does not mean informal. The sheet or lightweight application needs controlled identifiers, required fields, explicit cut-offs and a correction history. A blank entry should remain distinguishable from zero. A late entry should retain both event time and entry time. A corrected value should preserve its predecessor and reason rather than silently rewriting the shift.

Exhibit 3 defines a practical minimum register. Its purpose is to connect each measure family to a source event and an immediate decision, not to prescribe an exhaustive data model.

The table is deliberately organised around events. It prevents a common failure in which each function builds a separate report with a different meaning for output. Production, quality, cost and sustainability should be able to reconcile to the same accepted quantity at the same boundary. Where their scopes legitimately differ, the difference should be explicit. A sustainability calculation may use a monthly meter close while a production huddle uses a shift close; neither should borrow the other's denominator without carrying that time basis.

Institutional memory also depends on vocabulary. Defect names, interruption causes and method identifiers should be stable enough to compare, while remaining open to controlled revision. The aim is not a perfect taxonomy on the first day. It is a governed starting vocabulary whose changes are dated and whose old records remain interpretable. Otherwise a definitional improvement can look like a change in factory performance.

Finally, memory must include unsuccessful responses. If a supervisor changes a work allocation and the problem persists, that test is useful evidence. Recording only apparent successes creates an archive of unjustified confidence. The first-shift system should retain the observed problem, the response and the subsequent result so later improvement work begins with what has already been tried.