Home › Field notes › How to Choose a Modular Bag Manufacturer: A Vetting Checklist

Modular Bag Pro · 2026-10-10 · 2,718 words · 12 min read

A modular bag manufacturer is best assessed as a production partner whose interface discipline can be demonstrated rather than as a vendor whose quotation can be compared. Modularity multiplies the number of things that have to agree with each other: chassis geometry, attachment pitch, module envelope and reinforcement all have to be right on the same drawing before a single unit is worth making. Price differences between candidates are usually smaller than the cost of one interface error discovered after mass production begins. This checklist covers the evidence to request, how to read it, and which commercial terms actually allocate risk. A first prototype is normally available within 6-10 working days, extending to 12-15 working days when a magnetic-guided fitting has to be sourced, and using that round deliberately rather than skipping it to save a week is the cheapest vetting available.

Why Modular Programmes Fail on Interface Discipline

Most sourcing failures on modular work are not quality failures in the usual sense. The stitching is fine, the fabric meets specification, and the product still does not work because a module sits two millimetres off its intended anchor or because the pouch from season one will not mount on the chassis from season two.

That places interface discipline at the top of the vetting list, and it is a documentation question before it is a manufacturing one. A candidate who can produce a drawing with toleranced callouts for webbing width, row pitch and horizontal repeat is demonstrating a capability that no photograph of a showroom can demonstrate. A candidate who responds with a rendering is not necessarily worse at sewing, but is telling you that the geometry is being decided on the floor.

The second failure mode is revision discipline. Modular programmes invite small changes: one more module, a moved pocket, a different fitting. Each change has to be logged against a revision number and checked against the interface drawing, because the cost of an unlogged change is a batch of modules that mount inconsistently. Ask how revisions are recorded and who signs them.

A third failure mode is the load case that nobody specified. A module mounted and loaded by the user generates shear and peel at the anchor, not the straight tension that most hardware datasheets describe. A candidate who asks what mass each anchor row must carry is demonstrating that they have met this problem before; one who does not ask will build to whatever feels sufficient on the floor.

Selection rule: Rank candidates on whether an anchor-row mass or interface tolerance is volunteered during the first conversation, then confirm execution across 7 production lines and 149 machines rather than accepting a headline throughput figure on paper.

Capability Evidence Worth Requesting

Requests for information are usually answered with a brochure. The table below replaces the brochure with seven artefacts, each of which is difficult to fake and cheap to produce if the capability is real.

Two of these deserve emphasis. The counter-sample is the single most informative artefact on the list: it is produced against your drawing, so any ambiguity in your specification shows up as a physical deviation you can measure. The revision log is the one most candidates cannot produce, and its absence predicts the interface drift that appears in season two.

Our own production team answers the same seven requests and issues the revision log as a standard deliverable, because a programme without one cannot be audited by the brand after the fact. If a candidate treats a revision log as an unusual request, that is the finding.

Verdict: Request interface drawings, revision logs and method-named test reports instead of a brochure, because seven concrete artefacts can be checked in a day while unreferenced claims cannot be reconciled once the 6-10 working day sampling round opens.

The Sampling Round as a Vetting Instrument

Sampling is normally treated as a product step. It should also be treated as a capability test, and read that way it produces more information than any document request.

Three observations matter. The first is whether the sample matches the drawing: measure the attachment pitch on the delivered unit against the callout and record the deviation. A candidate whose sample is close but not on-spec will produce the same behaviour at volume. The second is whether the sample was built with the specified hardware, or with whatever was on the shelf. The third is how the deviation is explained when you raise it - a measured answer with a corrective action is a different signal from a reassurance.

Turnaround is part of the signal. Most configurations come back inside 6-10 working days; anything needing a plated or moulded part moves out to 12-15 working days. A candidate quoting a much longer window is usually telling you that materials are bought against your order rather than held in advance, which becomes a scheduling risk once volume starts.

Use the round to test the interface rather than the appearance. Mount every module you intend to sell, load each one to its stated mass, and check for sag, release and stitch movement. Appearance problems are cheap to fix; interface problems found after mass production are not.

Ask for the sample to be built on the production line rather than in a development room. A sample made by a specialist operator proves that the product can be made; a sample made on the line proves that the line can make it. Where the two differ, the difference is usually in reinforcement and stitch consistency at the anchor rows, which is precisely where modular products fail.

Send one ambiguous requirement deliberately. A deliberately loose line in the specification is a useful test: a strong partner queries it before cutting, and a weak one builds the most favourable interpretation and invoices the correction later. The response to ambiguity predicts the response to every change that follows.

Bottom line: Score how a sampling partner handles an ambiguous clause in the brief, since one that queries before cutting protects the 35-50 day mass production window and one that builds the favourable reading invoices the correction afterwards.

Quality System: What Certificates Do and Do Not Prove

Certificates are often treated as a verdict. They are better read as a scope statement, and the useful question is always what the certificate covers and who issued it. The table sets out the common documents and the boundary of each.

Verification is a two-minute task that is rarely done: check the certificate number against the issuing body's register, confirm the scope covers the activity you are buying, and confirm the date has not lapsed. A certificate for a different scope or an expired one is the most common form of misrepresentation in this category, and it is detected by anyone who looks. Ask for the accreditation mark on the certificate as well, since a document issued by an unaccredited body carries no independent assurance however it is presented.

What no certificate replaces is the inspection protocol. Agree the acceptance quality limit in writing - AQL 2.5 is conventional for this class of goods - and classify defects before production rather than during the final inspection. A certificate will never resolve an argument about whether a coating mark is major or minor; a signed classification will.

Capacity, Lines and Reading a Throughput Figure

A published capacity figure is a statement about a product, and the first question is which product. A headline of 200,000 units per month refers to single-body output spread over 7 production lines; the floor carries 149 machines and employs 137 people within 4,950 m² of SGS-verified space. Modular work loads those same lines unevenly, because every pouch brings an extra handling pass, a mounting verification and an individual pack-out that a plain body does not require.

Request a figure stated as completed assemblies per line-week for the mix you are actually buying. A chassis shipping with four module types is not one assembly; it is five handling units before packing. Candidates who cannot convert their headline figure into your configuration are quoting from a brochure rather than from a schedule, and the difference shows up as a missed date rather than as a bad quotation.

Line allocation is the second question. A modular programme benefits from staying on one line for the whole run, because interface settings and inspection fixtures stay set up. Ask whether the programme will hold a line or be moved between lines to fill gaps; the second arrangement is how interface drift enters a production run that started correctly.

The third question is what happens at the top of the range. Our vetted partner facilities can produce 200,000 units per month in single-body terms, but a modular release competes for the same lines as every other programme in the building. Ask how line time is allocated when two programmes want the same window, and whether your window is reserved by the deposit or merely noted in a schedule.

Finally, ask what the candidate does when a run overruns. The answer that matters is whether rework is separated from new production. Rework mixed into a live line is the most common origin of mixed-revision goods in one shipment, where some units carry the corrected interface and some do not.

Takeaway: Convert any stated capacity of 200,000 units per month into assemblies per line-week before comparing offers, because 500 chassis each carrying four modules load 7 production lines differently from 500 finished single-body packs.

Commercial Terms and the Risk They Allocate

Terms are not administrative detail; they are the risk allocation of the relationship. Four items carry most of it.

Minimum quantity sets which configurations are economically possible. MOQ 500 per configuration reflects the material call-off and sewing setup behind each one, and it is the reason a module mix has to be declared at quotation stage rather than after sampling. Quoting a chassis quantity without the module mix produces a number that is correct for a different product.

Payment structure sets where leverage sits. Under T/T 30/70 the opening instalment funds material commitments and holds line time, while the remainder falls due at inspection release, which is the last moment at which a defect can still be corrected. Structures that push most of the money to a point after shipment remove that leverage from both sides and generally return as slower service rather than as better goods.

Incoterm sets where risk transfers. Quotations on FOB Xiamen place freight, insurance and post-loading risk with the buyer, which makes the packing specification a buyer-side responsibility. If mounted hardware is being shipped, the packing standard protecting it has to be written before loading, not argued about after discharge.

Lead time sets the calendar, and the numbers to hold a candidate to are sampling at 6-10 working days or 12-15 working days when complex, and mass production at 35-50 days from material availability. A candidate who will not commit to the second figure in writing is not committing to a date at all.

Transit is the fourth item and the one most often excluded from the comparison. Sea freight occupies 25-35 days, air 5-8 days and courier 3-5 days, and a carton of modular goods packs worse than a single-body carton because mounted fittings prevent nesting. A candidate who quotes production dates but will not discuss how the packed carton measures is handing the buyer a schedule with the last leg missing.

Intellectual property is worth one paragraph in the agreement even where both parties are acting in good faith. Modular geometry is the asset: the interface drawing, the module envelopes and the reinforcement scheme. State who owns the drawing, and state whether tooling paid for at USD 300-2,500 is exclusive to the programme or available to the range. Where tooling is shared, say so in writing at the outset, because a moulded interface part that later appears on another brand's chassis is difficult to contest once it has been running for a season. Exclusivity is far cheaper to agree before tooling is cut than to argue about afterwards.

Scorecard for Ranking Production Partners

Once the evidence is in, ranking should be explicit rather than intuitive. The scorecard below weights the criteria that predict modular success, and it deliberately weights interface discipline above unit price.

Weighting price last is deliberate and worth defending. In modular work the price spread between competent candidates is usually narrow, while the cost of one interface error discovered after mass production is not narrow at all. A scorecard that ranks on price produces a decision that looks rigorous and is not.

Run the scorecard after the sampling round rather than before it. Documents can be prepared for a bid; a counter-sample cannot, and the difference between what a candidate promises and what a candidate builds is exactly the thing being measured.

Judgement: Weight the scorecard towards evidence produced during sampling at MOQ 500 rather than towards documents assembled for the tender, since a counter-sample cannot be written overnight and a certificate can.

What to Lock Before the First Purchase Order

Nine items should be settled in writing before money moves: the geometry callouts and their tolerance bands; the module mix and the mass each anchor row must carry; the retained counter-sample and its reference number; the inspection protocol including AQL 2.5 and how defects are graded; the packing carton and its measured volume; the delivery term as FOB Xiamen; the payment split as T/T 30/70; the assembly window of 35-50 days; and who owns the interface drawing.

Two further items are worth adding because they are cheap now and expensive later. The first is a substitute-part rule for hardware, covering what makes an alternative fitting acceptable without re-validation. The second is a spare-part commitment: how long interface hardware and replacement modules will remain available after a chassis leaves the range.

Retention of the approved reference is the control that makes all of the above enforceable. A signed counter-sample held by both parties converts every later disagreement into a comparison against a physical object rather than a recollection. Programmes that skip this step spend the disagreement reconstructing what was agreed.

To run this checklist against a live brief, the services page sets out what our production team needs in order to quote, and the about page gives the background on the production base behind the figures quoted here.

Spec rule: Settle nine items in writing before the T/T 30/70 deposit moves: geometry tolerances, module mix, anchor-row masses, revision rules, defect classes set against AQL 2.5, incoterms as FOB Xiamen and the tooling ownership question among them.