For most programmes, the piece of the process that gets compressed is supplier development, and it is compressed for an understandable reason: nothing is visible about it. Development produces no saleable goods, generates no revenue, and produces a set of samples that look much like the samples the previous round produced. Meanwhile the launch date is fixed and the retailer is waiting. So development gets treated as a delay to be managed rather than as an engineering activity whose output determines whether the first production order succeeds. The economics of that decision are straightforward and consistently unfavourable. A programme that completes development properly will typically spend a modest amount of additional time and a modest amount of additional money on samples and documentation before production release. A programme that completes it hastily will typically discover the problem during the first production run, at which point the cost is not the sample but the rework, the delayed delivery, the retailer's chargeback for late or non-conforming goods, and in the worst case the discontinuation of the article, because a launch that fails its first cycle rarely gets a second. The three most expensive outcomes in this category - a colour that does not match, a construction that does not perform in the end use, and a fit that fails during assembly - are all discoverable during development, and all of them are expensive to discover later. Development is the cheapest place to be wrong, and the discipline of the nine gates is simply a way of making sure that you are wrong there rather than in the container.
The first gate is the one most often bypassed, usually because a sample can be made from a verbal brief and a specification takes two weeks to write properly. A sample made from an incomplete brief produces a result that cannot be interpreted, because the buyer and the mill will be looking at different things. The specification needs to be complete enough to answer the questions that determine whether the sample means anything: the material construction and the specific product, the width and the tolerance and how it is measured, the colour reference and the method, the print or decoration specification including the process, the number of colours and the repeat length, the edge treatment, the selvage condition, the finish, the length tolerance, the minimum order quantity and the price basis, the labelling and packaging, the inspection level and the acceptance criteria, and the compliance documentation required for the destination market. Two of these deserve comment because they are commonly under-specified. The colour reference should be a physical reference rather than a Pantone number alone, because a printed Pantone chip on paper behaves differently from the same colour on a dyed or printed ribbon, and a physical reference makes the conversation about a real object instead of about an abstraction. The inspection level should be stated in the specification rather than assumed, because the inspection level determines the mill's own setup and therefore its price, and discovering a change in inspection level after the quotation has been accepted creates a dispute about who bears the difference. A specification that is complete at Gate 1 is what makes every later gate meaningful; a specification that is incomplete means that every sample round is really a negotiation about what to make.
Gates 2 and 3 establish the two things that most often go wrong downstream: the colour and the print. The lab dip, or the dyed sample, answers a narrow question: can this construction carry this colour at the required depth and fastness, and what does the dye look like on this substrate? This is not a formality, because the interaction between a dye chemistry and a fibre content is genuinely unpredictable until it is run, and the colour that comes out of the lab can differ from the target in hue while appearing identical on a screen. The approval should record the measured colour difference against the target under a stated illuminant and observer condition, and it should be an approval of a physical sample that is retained on both sides, because the retained sample is the only reliable reference available when a dispute arises six months later. Gate 3, the print proof, answers the same question for decoration: does the artwork reproduce at the specified colour, does the repeat align with the specified repeat length, is the registration within tolerance, and does the print adhere adequately. The print proof should be run on the actual substrate rather than on a proofing stock, because ink behaviour on polyester satin and on a paper proof are different enough that a proof can be approved and then be unmanufacturable. These two gates are where the buyer's colour discipline is either established or lost. A buyer who approves a lab dip on a screen, or who accepts a print proof on the wrong substrate, will find out during the first production run, and by that point the colour is on thousands of metres of ribbon and the money is spent.
Gates 4 and 5 ask whether the ribbon actually works, and they are the gates most often skipped because they require the buyer to do something rather than simply approve something. Gate 4 is the functional sample, where the ribbon is used in the actual assembly in the actual end application. If the trim is destined for a gift box, it goes into a gift box; if it is a hair accessory, it is built into the accessory; if it is a pet collar trim, it is assembled into a collar. This is where a great many apparently perfect specifications fall apart. The ribbon may be exactly the right width and exactly the right colour and still fail in the assembly, because the cut length tolerance interacts with the fold geometry, because the selvedge is too heavy for the fold, because the stiffness is wrong for the loop, because the adhesive or the stitching interacts badly with the fibre, because the trim is too slippery to hold position or too stiff to conform. None of these are visible on a length of ribbon and all of them are immediately obvious when the ribbon is built into the product. Gate 5 extends this to an end-use test, and its purpose is to establish the performance the programme requires before production commits to it. The specific tests depend on the application, and the principles for selecting them are well established: the test should predict the failure that the end use actually produces, and the pass criterion should be a number rather than a description. A ribbon that must sit against skin, that must survive a hot car in August, that must not stretch and stay stretched under tension, that must not fray at the visible cut end, that must hold a bow's shape through a warehouse journey, or that must not mark the tissue it is wrapped around, has a specific and identifiable failure mode in each case, and a specific test that predicts it. The discipline of stating the method, the threshold, the criticality classification and the measurement point at Gate 5, rather than at the end of the programme, is what converts a general expectation into a criterion the mill can actually run.
Gates 6 to 8 convert an approved sample into an approved production, and each of them addresses a specific way in which a first order can go wrong. Gate 6 is the pre-production approval, and it is the formal point at which the specification, the artwork, the colour reference, the bill of materials and the inspection criteria are all confirmed together in one version. Its purpose is to eliminate the situation where the mill is working to a colour approved in March, artwork updated in May and an inspection plan agreed in April, with nobody being at fault for the mismatch. The pre-production approval should be a signed, version-controlled document set with a stated effective date, and every downstream document - the tech pack, the bill of materials, the inspection plan, the packing instruction, the label artwork - should reference that version. Gate 7 is the golden sample, which is the physical artefact that everything afterwards is measured against. It has four required properties and it is worth stating them separately because each one is a source of dispute. It must be the actual production construction, not a hand-made or specially-finished sample, because a sample produced outside the normal route will not match production. It must be from a production run, or at minimum from a run using production materials, production settings and the production machine, because the settings are part of the definition. It must be retained and identified, with a reference number that appears on the inspection plan and the packing documents, so that any roll can be traced back to it. And it must be signed by both parties, with a stated number of metres retained by each. A programme without a properly constituted golden sample will resolve every quality question by argument, because there is no physical reference against which to argue. Gate 8 is capacity confirmation, and it is the gate most often treated as a formality in a supplier's favour. Confirmation should cover not only whether the mill can make the quantity but whether it can make it in the window required, which means checking the actual loading against the actual schedule rather than against a general statement of capacity. The questions to ask are specific: which lines will run this article, what is the changeover time, what is the run rate on this width and construction, what other commitments does the same line carry in the same period, and what is the plan if the peak season compresses the schedule. A buyer who confirms capacity in the abstract and discovers the mill is running the article on a line shared with a higher-margin order has a schedule problem that no amount of goodwill resolves. Smith Ribbon confirms capacity against the actual loading and schedule for every programme, with the assigned lines, the run rates and the changeover windows stated in the programme file, because a capacity claim that is not tied to a line and a window is not a capacity confirmation at all.
The final gate is the production release, and it should be a decision rather than a formality. The readiness criteria behind it should be explicit and, crucially, should be written down before the release decision is made rather than assembled afterwards to justify it. Six items belong on the readiness check. First, a complete and version-controlled specification with all acceptance criteria and tolerances stated, including the criticality classification. Second, approved artwork at the version that will be used, with the colour reference attached and the measured colour difference recorded. Third, an approved golden sample with a reference number, retained by both parties. Fourth, an agreed bill of materials showing every component and confirming which steps are in-house at the mill and which are subcontracted. Fifth, an inspection plan with the sampling method, the lot definition, the AQL levels by criticality class and illustrated defect categories. Sixth, the compliance documentation set for the destination market, with the certification scope and expiry confirmed. Two further items are conditional rather than universal and should be confirmed where they apply: where recycled or sustainably sourced content is claimed, the substantiation for that claim; and where the programme has a retailer's own vendor-compliance requirements, confirmation that the mill's documentation satisfies them. The value of writing the readiness criteria down is that it converts a judgement call into a checklist, which means that the decision to release can be defended later to a retailer, to an auditor or to a colleague who was not in the room. It also makes the release a genuine gate rather than a formality, because a gate that can be passed by assertion is not a gate. And it gives the mill a clear definition of done, which matters more than buyers sometimes expect: a mill that knows precisely what the release requires can organise its own planning around it, whereas a mill working to an unstated expectation will meet a different standard than the buyer had in mind.
Sample rounds fail in a characteristic way: the parties meet, look at the sample, agree it is close, and agree to look again after the next round, and the rounds continue without becoming more informative. The reason is usually that each round is treated as another attempt at the same object rather than as the answer to a question raised by the previous round. The more productive structure is to define, at the end of every round, the specific question the next round must answer. A first sample establishes whether the construction and the colour direction are right at all. If it is off in hue, the next round must answer only the hue question and nothing else, because a round that changes construction, colour and decoration simultaneously produces a result that cannot be attributed to any of the three changes. If the colour is right and the print registration is off, the next round must isolate the registration variable. If the functional sample failed because the ribbon was too stiff, the next round must change only the stiffness, and the mill needs to be told which property is being adjusted so it does not compensate elsewhere. This approach is usually faster in wall-clock terms than running open-ended rounds, because it removes the discussion of what to change and replaces it with a single change per round, and it is cheaper because the number of samples per round can be reduced once the variable is isolated. Three supporting practices make it work. The first is to keep a written sample log recording, for every round, the version of the specification that was worked to, the change that was requested, the result and the next open question. The second is to have the mill state what it expects the sample to show, before it is made, which surfaces disagreements about the specification before the money is spent rather than after. The third is to decide in advance how many rounds are needed and what happens when they are exhausted. Without that last point, a programme that is not converging continues indefinitely, and the pressure to release eventually wins over the discipline, which is the same outcome as never having had the gates.
The output of a completed development programme is not just a release approval; it is a file that will be read years later by people who were not in the room, including a retailer's vendor-compliance reviewer, an auditor, an insurer and a future procurement lead. It should be assembled as one coordinated set rather than as a folder of separate documents that each reference a different version. Nine items belong in it. The approved specification, with its acceptance criteria, tolerances and criticality classification. The tech pack, with the artwork and the print files at the approved versions and the print configuration recorded. The colour reference, with the approved physical reference and the measured colour difference against the target under the stated illuminant and observer condition. The bill of materials, showing every component and identifying which steps are in-house and which are subcontracted. The golden sample record, with the reference number and the retained quantities on both sides. The inspection plan, with the sampling method, the lot definition, the AQL levels by criticality class and illustrated defect categories. The functional and performance test results, with the method, threshold and classification for each. The compliance documentation set, with the certification scope and expiry for anything claimed. And the release approval itself, dated and version-controlled, listing what was confirmed. Filed together and maintained through the programme, this set answers the questions that get asked at the worst possible moment: when a claim arrives, when a retailer audits, when an insurance claim is made, when a new buyer takes over the category and needs to understand what they have inherited, and when a second supplier is qualified and has to be brought to the same standard. Smith Ribbon issues the specification, tech pack, bill of materials, inspection plan, colour reference and the in-process inspection record with measured values for every programme file, so the buyer receives a coordinated documentation set rather than a set of documents assembled from separate requests.
Nine, and each one exists because of a specific failure it prevents. Gate 1 is specification completeness, before any sample is made, because a sample made from an incomplete brief produces a result nobody can interpret. Gates 2 and 3 are the lab dip and the print proof, which establish colour and print on the real substrate. Gates 4 and 5 are the functional sample and the end-use test, which establish whether the ribbon works in the actual assembly and performs in the actual application. Gates 6, 7 and 8 are the pre-production approval, the golden sample and capacity confirmation, which convert an approved sample into an approved production. Gate 9 is the production release, which is a decision made against written readiness criteria rather than an assertion. The sequence is not arbitrary: each gate tests a different class of risk, and skipping one leaves that class uncovered no matter how carefully the others were handled.
The useful answer is not a number of rounds but a number of variables. Each round should isolate one change, because a round that alters construction, colour and decoration together produces a result that cannot be attributed to any of the three. A programme that follows this discipline typically converges in two or three rounds: one to establish whether the direction is right at all, one to correct the specific property that was off, and one to confirm the corrected version. Programmes that take five or six rounds are usually not converging because the rounds are not isolating variables, which is a process problem rather than a material one. The discipline that makes the count predictable is deciding in advance how many rounds are allowed and what happens when they are exhausted, because without that point the pressure to release eventually overrides the gates, which produces the same outcome as never having had them.
The golden sample is the physical reference artefact against which every subsequent production run is judged. It needs four properties. It must be the actual production construction, not a hand-made or specially finished sample. It must come from a production run, or at least from a run using production materials, production settings and the production machine, because the settings are part of the definition. It must be retained and identified by a reference number that appears on the inspection plan and the packing documents, so any roll can be traced back to it. And it must be signed by both parties with a stated number of metres retained by each. Its function is to end the recurring argument. Without a properly constituted golden sample, every colour question, every print registration question and every shade complaint becomes a dispute between two descriptions rather than a comparison against a shared object, and that comparison is the only kind that resolves quickly.
It should cover not only whether the mill can make the quantity but whether it can make it in the window required, which means checking the actual loading against the actual schedule rather than a general statement of capability. The specific questions are which lines will run the article, what the changeover time is, what the run rate is on that width and construction, what other commitments the same line carries in the same period, and what the plan is if the peak season compresses the schedule. Confirming capacity in the abstract is a common and expensive error, because a mill running the article on a line shared with a higher-margin order has a schedule problem that goodwill does not resolve. Where genuine priority in a constrained period matters, it should be stated in terms of the quantity the mill must hold available rather than in general language about priority.
Six items are unconditional, and two are conditional. The unconditional set is a complete version-controlled specification with acceptance criteria, tolerances and criticality classification; approved artwork at the version that will be used with the colour reference attached and the measured colour difference recorded; an approved golden sample with a reference number retained by both parties; an agreed bill of materials identifying which steps are in-house and which are subcontracted; an inspection plan with the sampling method, lot definition, AQL levels by criticality class and illustrated defect categories; and the compliance documentation set with certification scope and expiry confirmed. The conditional pair is substantiation for any recycled or sustainably sourced content claim, and confirmation against any retailer-specific vendor-compliance requirements. Writing these down before the release decision is made is what converts the release from a judgement call into a checklist that can be defended later to a retailer, an auditor or a colleague who was not in the room.
Three practices do most of the work. The first is to define, at the end of every round, the specific question the next round must answer, and to require that only one variable changes. The second is to keep a written sample log recording the specification version worked to, the change requested, the result and the next open question, because the log is what prevents a round from re-opening a question that was already closed. The third is to decide in advance how many rounds the programme has and what happens when they are exhausted, and to agree that outcome with the mill at the start rather than under pressure at the end. The underlying cause of runaway rounds is usually that each round is treated as another attempt at the same object rather than as the answer to a question the previous round raised, so the rounds never become more informative and the programme drifts.
Nine items, assembled as one coordinated set rather than as a folder of documents that each reference a different version: the approved specification with acceptance criteria and criticality classification; the tech pack with artwork and print files at the approved versions; the colour reference with the approved physical standard and the measured colour difference recorded; the bill of materials identifying in-house and subcontracted steps; the golden sample record with reference number and retained quantities; the inspection plan with sampling method, lot definition, AQL levels and illustrated defect categories; the functional and performance test results with method, threshold and classification; the compliance documentation set with certification scope and expiry; and the dated, version-controlled release approval. 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 specification, tech pack, bill of materials, inspection plan, colour reference and in-process inspection record with measured values for every programme 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 and 100,000 metres of daily capacity runs a structured development programme with retained golden samples, version-controlled documentation and confirmed capacity for every article.
If you are qualifying a ribbon or decorative trim supplier and want to see what a documented, gate-based development programme looks like in practice, send us your brief and your target market to xmmsd@126.com or call +86 13779951780, and we will walk you through how we would take it from specification to production release.