A buyer and a mill can agree on everything important and still end up in disagreement, and the reason is usually that they were never working from the same document. The buyer believed the ribbon should match a sample; the mill believed the buyer had approved a Pantone reference with a tolerance the buyer had never seen written down; the mill produced a good ribbon to a specification the buyer had not intended. Neither party made a mistake in execution. The failure was upstream, in the specification itself, and it is the most common cause of a sampling dispute in trim sourcing. Three characteristics make a ribbon specification dangerous. The first is that it is usually incomplete, because the buyer specifies what they can see and name, and leaves the rest to the mill's judgement. What they can see and name is typically the material family, the width, the colour and maybe a finish. What they leave out is the majority of what determines whether the ribbon performs: the construction, the colour tolerance and measurement method, the shrinkage behaviour, the print rub and fastness characteristics, the edge treatment, the core and winding specification, the labelling and packing requirements, and the acceptance criteria. The second is that it is usually written in a language that permits more than one reading. ‘Good quality satin’ and ‘matching the sample’ and ‘similar to approved’ are all phrases that a reader will interpret through their own frame, and in a different frame they mean different physical specifications. The third is that it is usually written once and never versioned, so after two rounds of samples and three email exchanges there are four versions of the requirement in circulation and no record of which one is current. A specification is not a document; it is an agreement about what will be measured. The purpose of this guide is to describe what that agreement should contain for ribbon and decorative trim, and how to make it survive the distance between a brand's design room and a factory's production floor.
Fourteen fields are sufficient to make a ribbon specification quotable, enforceable and capable of surviving a dispute. The first is the product identity: a unique code, the product name, the buyer and the factory reference, so that every subsequent document and every carton label refers unambiguously to one thing. The second is the construction: base material with its composition, weave or knit type, the yarn count or denier, the number of ends and picks per centimetre, the lamination or backing if any, and whether the construction is woven, knitted, non-woven or a composite of ribbon plus wire. The third is the dimensional specification: nominal width in millimetres with the permitted deviation, the finished length per roll, and the cross-section profile where the ribbon is not flat. The fourth is the material and finish: fibre type and, where relevant, recycled content with the certification basis, the surface finish, the sheen or lustre level, the hand or stiffness rating, and the softness or drape characteristic where the application depends on it. The fifth is the colour specification: the Pantone or colour reference, the measurement method and instrument, the target Delta E tolerance, and the illuminant and observer angle under which the tolerance is judged. The sixth is the decoration specification: the print method, the number of colours, the repeat length, the registration tolerance, the printing side or sides, and the substrate onto which the print is applied. The seventh is the edge and finish treatment: selvedge, hemming, wired or unwired edge, the wire specification and gauge where present, cut ends, and the edge quality standard. The eighth is the performance specification: the tests that must pass, the test method and standard for each, the pass or fail threshold, and the conditioning state in which the test is performed. The ninth is the labelling specification: the roll label content, the carton label content, the country of origin statement, the fibre content statement, the care label requirement where the product is sold as a consumer item, and the language and format required. The tenth is the packing specification: the packing method, the roll length, the carton dimensions and quantities, the maximum units per carton, and the master carton quantity. The eleventh is the inspection and acceptance specification: the sampling method, the AQL level by criticality class, the inspection points, the definition of a production lot, and the documentation to be provided with the shipment. The twelfth is the tolerances table, in which each measurable characteristic is listed with its target, its tolerance, its measurement method, and the party who measures it. The thirteenth is the artwork and colour approval reference: the file names, the version numbers, the approval dates and the names of the approvers. The fourteenth is the version control block: the specification version, the revision date, the list of what changed, and the signature or approval record. Every one of these fields can be empty, but each must be present and explicitly marked either as specified or as not applicable, because an empty field and an absent field mean different things. An absent field is an assumption. An empty field is a decision.
A description states what the ribbon should be. An acceptance criterion states how anyone, including a person who has never met you, would decide whether it is acceptable. The difference is worth a great deal in a dispute, because it removes the need for argument about intent. Seven criteria do most of the work. The first is dimensional: width is measured at three points along a relaxed ribbon laid without tension, and the result must fall within the stated tolerance. Stating where and how the measurement is taken matters more than the tolerance itself, because a ribbon measured under tension reads narrower than the same ribbon measured relaxed. The second is colour: colour is measured with a spectrophotometer against the agreed reference, and the Delta E value must not exceed the stated threshold under the agreed illuminant and observer angle. Both the instrument and the geometry must be stated, because the same ribbon can legitimately produce different readings under different measurement conditions. The third is construction mass: weight per square metre or per linear metre within tolerance, verified by weighing a defined cut length rather than by inspecting a roll label, because the label records the theoretical value and the produced value drifts. The fourth is width consistency across the roll, measured at a stated frequency along the length, because a ribbon that is correct at the start of a roll and off-tolerance at the end indicates a tension or feed problem that will not be visible in a sample taken from the head. The fifth is decoration quality: print rub, crocking, fastness and registration assessed against the stated method, with a defined number of specimens and a defined rating scale. The sixth is absence of defect: a stated limit on defects per unit by type, distinguishing critical from major from minor, with an illustrative definition of each category rather than a reference to a standard the reader must locate. The seventh is the dimensional and count check on the delivered goods: roll length, number of rolls per carton, carton count, and net weight, verified against the packing specification at receipt. Each criterion needs a threshold and a method. A criterion with a threshold and no method is not testable, and a criterion with a method and no threshold produces data rather than a decision.
Six phrases account for a large share of the sampling disputes in ribbon sourcing, and each has a specific, practical remedy. The first is ‘match the sample’. It is ambiguous because a physical sample is a poor specification: it carries the colour of one dye batch, the finish of one finishing run and the edge quality of one cutting operation, and it degrades in transit and in storage. The remedy is to specify the measurable characteristics against which the sample is to be judged — the Pantone reference and Delta E tolerance, the width and tolerance, the mass and tolerance, the finish description — and to state that the sample is a visual reference for construction and hand only. The second is ‘good quality’. It is ambiguous because it has no threshold and because it reads differently to a buyer who wants appearance preserved and to a factory who wants to avoid rejection. The remedy is to replace it with the acceptance criteria above, and to state that quality is assessed against those criteria rather than subjectively. The third is ‘similar to the approved’. It is ambiguous because it is used at exactly the moment when two parties have already disagreed, and it invites a new round of comparison without a decision rule. The remedy is to use a defined tolerance with a stated measurement method, and to specify the escalation route when a measurement falls outside it. The fourth is an unqualified colour reference. ‘Pantone 186 C’ without a tolerance and a measurement geometry is not a colour specification, because the reference is a standard swatch under a defined light and a defined substrate, and production ribbon will not reproduce it exactly. The remedy is to add the Delta E tolerance and the measurement conditions. The fifth is a width without a measurement condition. The remedy is to state relaxed or tensioned, the number of measurement points, and the measurement instrument. The sixth is an unqualified material term. ‘Polyester satin’ spans a range of constructions with meaningfully different performance, and so do ‘velvet’, ‘grosgrain’ and ‘organza’. The remedy is to add the construction fields — yarn count, weave, backing, pile or layer structure, and the finishing route — so that the material term narrows to a specific article. None of these remedies is difficult. Each of them removes an argument that would otherwise consume a sampling round.
A specification will be read by more people than the buyer who wrote it, and the handoffs between them are where meaning is lost. In a typical ribbon programme the document passes from a designer to a developer, from a developer to a merchandiser, from a merchandiser to a factory sales engineer, from a factory sales engineer to a production planner, from a planner to a colourist or a printer, from there to a quality inspector, and from the inspector to a warehouse that packs the cartons. Seven readers, each with a different frame, each pressed to act on what they have been given. The specification has to survive that chain, and it does so if three properties are present. The first is that every requirement is measurable. If a sentence cannot be measured, it cannot be enforced and it cannot be defended; it can only be negotiated, and it will be negotiated under time pressure. The second is that every requirement has a single unambiguous referent: one width, one colour reference, one artwork file version, one carton quantity, one inspection method. Where a range is genuinely intended, the range is stated with the rule for deciding within it. The third is that the document is versioned and the version is named on every artwork file, every sample label, every carton and every purchase order. This last point seems trivial and is the single most common failure. When a sample carries no version reference, the factory cannot know whether it corresponds to the current requirement, and both parties proceed on the assumption that it does. Version marking on the physical sample and on the artwork file costs nothing and prevents a whole class of dispute. Two further practices are worth adding. The first is a written requirement summary, of no more than one page, that lists the critical-to-quality characteristics in priority order, so that the planner and the inspector know which characteristics matter most when the schedule is under pressure. The second is a change log with dates and reasons, so that when a requirement changes mid-programme — and one always does — the file shows what changed, when, and why, and the goods already produced against the earlier version can be accounted for correctly.
Once a specification goes out, the responses arrive, and comparing them is where a great deal of money is quietly lost. The instinct is to compare unit prices, and unit price is the least informative number on the page because it is the one that varies least between suppliers who are otherwise equivalent. A comparable response answers all fourteen fields, so that two responses can be placed side by side line by line. Where a field is answered differently, the difference is not automatically a better or worse answer; it frequently reflects a different product construction that the mill is quoting rather than the one that was asked for. The comparison should therefore proceed in three passes. The first pass is a specification conformance pass, in which each response is checked field by field against the specification and any deviation is listed. A response that silently substitutes a different construction or a different material term is the most expensive kind of non-conformance, because it is not visible in the unit price. The second pass is a like-for-like normalisation, in which each quote is restated onto a common basis: the same material construction, the same width, the same print configuration, the same packing specification and the same Incoterm and payment basis. Only after this pass are the prices meaningful, and a quote that looks expensive before normalisation often proves neutral afterwards, because it was priced on a different basis. The third pass is a capability check, which asks whether the mill can actually make the article at the volume and to the tolerance required — confirmed by reference to comparable work rather than by assertion. This last point is the one most often skipped, and it is the point at which a plausible quote from a supplier who cannot run the specification becomes visible. A disciplined comparison therefore produces not a price but a decision: a preferred mill, the deviations you are willing to accept, the ones you are not, and the evidence for both.
The spec and the sample are two representations of the same requirement, and they are useful only if they are explicitly linked. The link is established in the sample approval step, and it should be recorded in four ways. First, the approved physical sample is labelled with the specification version, the artwork version and the approval date, and it is retained by both parties as the reference article. Second, the colour and construction of that article are recorded numerically rather than photographically, because a photograph cannot be measured and cannot be compared against a spectrophotometer reading; the Delta E value of the approved sample against the target is the number that governs subsequent production. Third, the acceptance criteria are restated as a measurable checklist on the approval record, so that the mill and the buyer are looking at the same set of numbers when they compare the next lot against the previous one. Fourth, the change-control rule is stated on the record: any change to the material, the construction, the print method, the colour reference or the tolerance requires a new approval, and production against the superseded version is not authorised. This last rule matters more than it appears, because the most common cause of an unauthorised change is not a supplier acting in bad faith but a benign substitution made for good reason — a yarn lot unavailable, a print base substituted for availability, an adhesive system changed to meet a delivery date — and a programme without an explicit rule will simply absorb it. The article-level approval is also the right point to attach the remaining programme documentation, because it is the last moment at which a requirement can be changed cheaply: the technical pack, the artwork and print files, the colour reference, the material safety data where applicable, the certificate scope and expiry, the packing instruction, the inspection plan and the label artwork. A programme that assembles these documents at approval and files them together has a development record that satisfies a retailer's vendor-compliance review, an auditor's quality review and an insurer's document request without a separate exercise. A programme that assembles them retrospectively spends weeks reconstructing what was agreed.
Fourteen fields. Product identity (unique code, buyer and factory references). Construction (base material and composition, weave type, yarn count, ends and picks, lamination or backing). Dimensions (nominal width and tolerance, roll length, cross-section profile). Material and finish (fibre type, recycled content with certification basis, surface finish, sheen, hand and drape). Colour (reference, measurement method and instrument, Delta E tolerance, illuminant and observer angle). Decoration (print method, colour count, repeat length, registration tolerance, printing sides, substrate). Edge and finish treatment (selvedge, hemming, wired edge, wire gauge, cut ends). Performance (required tests, standard or method, pass or fail threshold, conditioning state). Labelling (roll and carton label content, country of origin, fibre content, care label where applicable). Packing (packing method, roll length, carton dimensions, units per carton, master carton quantity). Inspection and acceptance (sampling method, AQL by criticality class, inspection points, lot definition, documents supplied with shipment). A tolerances table listing each measurable characteristic with target, tolerance, method and measuring party. Artwork and colour approval references with file names, versions, dates and approvers. And a version control block with version, revision date, change list and approval record. A field may be empty, but it should be present and explicitly marked as specified or not applicable, because an absent field is an assumption and an empty field is a decision.
Three properties make a specification survive handoff. First, every requirement must be measurable: a sentence that cannot be measured cannot be enforced and can only be negotiated, and it will be negotiated under time pressure. Second, every requirement must have a single unambiguous referent — one width, one colour reference, one artwork version, one carton quantity, one inspection method. Where a range is genuinely intended, state the rule for deciding within it. Third, the document must be versioned, with the version named on every artwork file, every sample label, every carton and every purchase order. This last point is the most common failure and the cheapest to fix: when a sample carries no version reference, neither party can confirm it corresponds to the current requirement, and both proceed assuming it does. Two supporting practices are worth adding: a one-page critical-to-quality summary listing requirements in priority order, so planners and inspectors know what matters most when the schedule is tight, and a change log with dates and reasons, so that when a requirement changes mid-programme the file shows what changed and when.
Both, with different roles, and the common mistake is to treat them as alternatives. A Pantone reference is a colour standard defined under a specific illuminant and on a specific substrate, so it cannot be reproduced exactly by production ribbon and therefore cannot stand alone as a colour specification. A physical sample is a poor specification for the same reason from the opposite direction: it carries the colour of one dye batch, the finish of one finishing run and the edge quality of one cutting operation, and it degrades in transit and in storage. The correct structure is to use the reference and the sample together. Specify the Pantone or colour reference as the target, add a Delta E tolerance with the measurement instrument and the illuminant and observer angle under which it is judged, and define the approved sample as the visual reference for construction and hand. Then record the Delta E value of the approved sample against the target numerically, because that measured value, not the physical swatch, is what governs subsequent production comparison.
Compare in three passes rather than by unit price, which is the least informative number on the page because it varies least between equivalent suppliers. The first pass is specification conformance: check each response field by field against the specification and list every deviation. A response that silently substitutes a different construction or a different material term is the most expensive non-conformance because it is invisible in the unit price. The second pass is like-for-like normalisation: restate each quote onto a common basis — the same material construction, width, print configuration, packing specification, Incoterm and payment basis. Only then are prices meaningful, and a quote that looked expensive before normalisation frequently proves neutral afterwards because it was priced on a different basis. The third pass is a capability check: confirm by reference to comparable work that the mill can produce the article at the volume and to the tolerance required. This is the pass most often skipped, and it is where a plausible quote from a supplier who cannot run the specification becomes visible.
They are three related documents with different jobs, and conflating them causes problems. The brief states intent: what the ribbon is for, the design direction, the application, the aesthetic reference points and the commercial context. It is deliberately non-binding and is where the brand explores options with the supplier. The tech pack is the technical realisation of that intent: construction, materials, dimensions, colour, decoration, packaging and the artwork files that go with them. It is a working document that changes during development. The specification is the binding version: the fields, the tolerances, the acceptance criteria, the measurement methods and the version-control block, approved by both parties. The practical rule is that a brief may be ambiguous, a tech pack may be provisional, and a specification may not. Smith Ribbon issues the approved specification and tech pack together at the article-approval stage, with the artwork version, colour measurement values and acceptance criteria on the approval record, so that development and production are working from the same document rather than from two versions of it.
Attach an explicit change-control rule to the approval record: any change to the material, the construction, the print method, the colour reference or the tolerance requires a new approval, and production against the superseded version is not authorised. This matters more than it appears, because the most common cause of an unauthorised change is not a supplier acting in bad faith but a benign substitution made for good reason — a yarn lot unavailable, a print base substituted for availability, an adhesive system changed to meet a delivery date — and a programme without an explicit rule will simply absorb it and discover the change at inspection. Closely related is a closed substitution list in the specification itself, naming which components may be substituted without a new approval and which may not. Smith Ribbon maintains version-controlled material and colour standards across programmes, and issues the approved specification with the retained sample and the in-process inspection record, so a substitution can be identified against a specific reference rather than reconstructed from memory.
The approved specification, the tech pack with the artwork and print files at their approved versions, the colour reference with the measured Delta E value of the approved sample, the bill of materials, the packing instruction, the inspection plan and the label artwork. Filed together at the article-approval stage, this set becomes the development record for the programme, and it is the set that satisfies a retailer's vendor-compliance review, an auditor's quality review and an insurer's document request without a separate exercise. A programme that assembles the same documents retrospectively spends weeks reconstructing what was agreed, and the reconstruction is usually incomplete. Xiamen Smith Ribbon & Bow Co., Ltd. has manufactured OEM and ODM custom woven and printed ribbon, satin, grosgrain, velvet, organza, RPET and decorative trims since 2004, and issues the approved specification, tech pack, bill of materials and inspection plan with each programme development file.
Smith Ribbon has supplied OEM and ODM custom woven and printed ribbon, satin, grosgrain, velvet, organza, RPET and decorative trims since 2004. Our 15,000 m² plant with 200+ employees, 100,000 metres of daily capacity, and certifications spanning OEKO-TEX®, FSC®, BSCI, SEDEX, SMETA and ISO 9001, develops every programme against a written specification with measured acceptance criteria, version-controlled artwork and colour references, and a closed substitution list.
If you are preparing a ribbon specification or comparing supplier responses, send us your tech pack, your target specification and the widths, colours and print configurations you are evaluating. We will return a field-by-field specification conformance comment, a like-for-like normalised quotation, a colour measurement proposal with the instrument and illuminant stated, and the documentation set we file at article approval so you can answer your own retailer vendor-compliance review without a second request cycle.