A controlled pilot can reject an attractive equipment story

The counterfactual is the service that would be used if the proposed digital route were not adopted. It may be conventional physical sampling followed by an outsourced print run, decoration of stocked blanks through an existing supplier, or another route that can genuinely fulfil the intended order. It is not a deliberately weak assembly of unnecessary steps. The comparison should name the garment, approval obligations, accepted quantity, delivery region and commercial terms that both routes must satisfy. Only then can a difference in time, cost or resource use inform adoption.

Equivalence starts with function rather than appearance. The historical VTT printing study makes batch size, quality, durability and product lifetime important to its comparison, rather than treating printed area as a sufficient description of the service.[S26] A pilot should apply the same discipline to a garment. A digital sample that avoids a physical fitting but leaves fit unresolved is not equivalent to an approved physical sample. A printed article with inferior care performance is not equivalent to one that meets the buyer's agreed use conditions. The protocol must either close those differences through testing or retain them as explicit limitations.

Residual physical work therefore belongs in both route definitions. Research comparing virtual and physical fitting concludes that simulation cannot fully replace a real physical fitting because technical issues remain.[S07] The proposed digital route might still reduce some iterations, but the pilot must count the fitting, strike-off, care test or sewn approval that remains necessary. It should not compare a complete baseline service with an incomplete digital preview. The unit of comparison is an approved style or retained order, with all work required to reach that state.

A useful design pairs routes at the style or order level. The same technical brief, material requirement, size coverage, artwork and acceptance evidence go to the baseline and proposed workflows where practical. Pairing reduces the chance that an easy style flatters one route while a difficult style burdens the other. It also lets the reviewer inspect differences within each pair instead of relying only on cohort averages. When simultaneous execution is impossible, the record should identify changes in staff, supplier, material lot and buyer requirements between periods.

Some factors cannot be held constant without distorting the decision. A digital route may require fabric digitisation that the baseline does not need. An outsourced baseline may include courier movement that an internal route avoids. Those differences are part of the routes and should remain visible. The analyst should not force identical activity lists. The requirement is an equivalent delivered outcome, not identical work. Each route receives the activities and costs that it actually needs to reach that outcome.

The baseline also needs an availability statement. If no stable internal sampling route exists, the protocol should compare documented supplier offers and observed outsourced work rather than inventing current factory performance. If a reliable baseline cannot yet be measured, the first pilot stage is baseline reconstruction. That work may delay an equipment decision, but it prevents a false saving from becoming the foundation of one.

A gate written after the results tends to fit the preferred answer. Before execution, the buyer and operations leads should agree what must pass, which evidence is acceptable, who decides and what happens after failure. Thresholds should reflect the garment's intended use, the buyer's specification, process capability and the decision's consequence. A supplier demonstration or an unrelated publication cannot supply a universal acceptance limit.

Physical tests need complete protocols rather than method names alone. ISO's laundering scope warns that industrial and hospital articles may require special procedures that are more severe in some aspects.[S10] Its rubbing scope distinguishes dry and wet rubbing cloths.[S11] Its domestic washing and drying scope states that a complete test consists of both washing and drying.[S12] These public catalogue entries help define the questions; they do not provide the pilot's pass grades. A qualified laboratory or competent reviewer must agree the full procedure, specimens, conditions and evaluation before testing.

Fit gates require the same specificity. The small real-versus-virtual garment study provides a warning against transferring study-specific discrepancies into a new operation.[S06] The pilot should instead state the locations, poses, intended wearer coverage, measurement uncertainty and buyer-approved tolerances that matter for this garment. It should record both evaluator observations and relevant wearer feedback. A mean difference cannot compensate for a local failure that prevents use.

Operational gates should cover the original delivery promise, complete retained-order contribution and recovery work. Delivery is measured against the promise made when the order was accepted, with later amendments shown separately. Contribution includes failed work, refunds and remakes rather than only clean output. Recovery includes reprints, repeated tests, additional courier movements and the labour required to correct them. A favourable development result cannot compensate for a service that loses money or arrives late under the agreed criteria.

Exhibit 54 provides the gate sheet's minimum structure. Its cells deliberately do not invent numerical thresholds. Each organisation must complete them from its own buyer, process and risk requirements before running the pilot.

Risk assessment should determine which failure can be conditional and which must stop the route. IFC guidance supports using a risk assessment form to prioritise which risks should be addressed first.[S18] Prioritisation does not mean averaging a severe safety or physical failure against commercial success. The gate sheet should identify non-compensable failures explicitly. A lower-priority documentation gap may receive a bounded correction period; an unclosed physical release failure keeps the affected service out of adoption.

Working conditions belong inside the protocol rather than in a later social review. OECD guidance identifies inefficient production planning and poor purchasing practices among common factors associated with excessive working hours.[S28] The pilot should therefore predefine how it records work outside normal shifts, late buyer changes and urgent recovery. Meeting a delivery gate through unrecorded or unacceptable hours is not a pass. Applicable employment obligations require qualified local review rather than inference from international guidance.

Finally, the decision rule should handle missing evidence. An unmeasured gate is not a passed gate, and an incomplete return window is not evidence of retention. The review can postpone a decision, narrow the claim or commission another test. It should not convert absent data into a favourable assumption. This rule protects the pilot from ending with a precise score whose most important fields were never observed.