Home › Field notes › Starting a Bag Brand: An Operating Plan for a Modular Launch

Modular Bag Pro · 2026-10-10 · 2,831 words · 13 min read

Starting a bag brand is an operating exercise more than a design one, and the order of operations is what decides whether the first year is survivable. Specify the chassis before the module catalogue, because the interface constrains everything that follows. Sequence the cash, because a first order consumes money in five separate stages spread across roughly one hundred days. Then handle compliance before listing rather than after. The numbers that frame it do not move: MOQ 500 per reference, 6-10 working days to sample, and USD 300-2,500 across tooling and screens, a build of 35-50 days, and release at AQL 2.5 by ISO 2859-1. The plan follows that order: architecture first, then cash, then compliance, then channel, and finally the cadence of the first ninety days after the goods land. It is written for a founder specifying a modular range for the first time, where the chassis and its attachment rule constrain everything that follows.

What you are actually buying when you launch a brand

A first-time brand founder usually believes they are buying a product. What they are actually buying is a repeatable process: a specification that can be rebuilt, a sample that can be approved, an inspection rule that can be enforced, and a supply position that can be reordered. The product is the output of that process, and it is the part that matters least to whether the brand survives its second year.

That reframing has a practical consequence. It means the deliverables worth paying for early are documents rather than goods. A dimensioned drawing with tolerances. A material specification naming fabric class, coating system and hardware by brand and size. An interface drawing with a version number. A test method behind every claim the brand intends to make. Those four documents are what allow a reorder to match the first order, and a reorder that does not match is a brand problem rather than a production one.

The second consequence is about scope. A launch range should be narrow enough to specify properly. One chassis and two or three modules, specified to the level where a dispute could be settled by measurement, outperforms five products specified loosely. Narrow scope also concentrates the opening buy, which matters when the working minimum is 500 units per reference and the cash has to cover all of them.

The third consequence is that the brand owns the decisions that production cannot make for it. Duty cycle, destination market, price position and channel are all brand decisions, and each one has a specification consequence. A production partner can advise on feasibility; it cannot decide who the customer is.

Selection rule: Selection rule: Decide what the launch actually buys, namely a repeatable interface, a documented compliance file and a supply arrangement that survives the second season, before spending anything on artwork or on an opening order of 500 units.

Range architecture: the chassis before the catalogue

The most common architectural error in a modular launch is designing the module catalogue first. It feels logical, because the modules are where the ideas are, and it produces a catalogue whose mounting requirements do not agree with each other. One pouch wants a wide face, another needs a rigid anchor, a third expects to hang from a strap, and the chassis ends up as a compromise that serves none of them well.

The correct order is the reverse. Fix the chassis geometry first: the mounting faces, their dimensions, the reinforcement behind them, and the interface rule. Then design modules against a known chassis, which is a bounded problem rather than an open one. The interface used across this site is 25 mm webbing with 38 mm row spacing and a 50 mm column pitch, and building to it means every module inherits a known, published mounting condition from the first sketch.

The second architectural decision is how many scenarios to cover at launch. Two or three named scenarios is the right number. a padded sleeve for the office day, a wash pouch for overnight travel, and a tool or utility organiser for everything else covers most of what a first customer will do, and each one has a name a customer recognises. A generic pouch has no occasion attached to it and consistently underperforms in attach rate.

The third decision is what to hold back. Reserving one or two module concepts for the second season gives the brand a reason to communicate with its installed base and a reason for that base to buy again. Launching with everything available removes both, and it front-loads development cost into the period when the brand has the least information.

Bottom line: Bottom line: Build the chassis before the catalogue and keep the module set to two or three references, because a range wider than the 6-10 working day sampling budget allows produces several untested articles in one launch.

Where the cash actually goes on a first order

Cash sequencing is where first launches fail, and the failure is rarely about total cost. It is about timing: money leaves in five stages across roughly one hundred days, and a founder who budgets the total but not the sequence runs out of cash in the middle with goods on the water.

Two of those five lines are controllable in ways founders often miss. Setup is controlled by reusing an existing chassis pattern and limiting artwork positions, which holds the number near the bottom of its band. Freight is controlled by mode and by consolidation, where sea freight at 25-35 days is materially cheaper than air at 5-8 and the choice is really a decision about how much inventory float the brand can carry.

The balance line deserves a note because it is the largest single outflow and the only one with leverage attached. It is paid against release after inspection passes, which means quality problems are still the production side's problem at that point. Protecting that sequencing, rather than accepting a request to pay before inspection, is worth more than any discount offered in exchange.

Sampling: getting the reference right before scaling it

Sampling is the cheapest place to be wrong and the most expensive place to be casual. A reference approved on appearance rather than on measurement carries its errors into the production run, where correcting them costs the whole order rather than one unit.

The window is 6-10 working days for a conventional build and 12-15 where a moulded part, a custom pull or a welded pouch enters the specification. Plan for two rounds. The first round reveals what the specification did not say; the second confirms the corrections. A third round usually means the brief was ambiguous rather than the execution poor, and the useful response is to stop and re-read it.

Review by measurement. Check the dimension callouts against their tolerances. Gauge the interface repeat over the entire panel width, not at a single point, because drift accumulates toward the edges where fabric tension is highest during sewing. Verify the reinforcement area at every anchor against the figure on the drawing. Cycle every closure and note resistance. Write each finding against the specification line it covers, so the second round is a comparison rather than a fresh opinion.

Approve in writing and retain the unit. The approved sample is the reference against which the production run is inspected, and a signed sample sitting in a folder resolves disputes that a description never will. Pair it with a revision note recording what changed, and update the specification document to match, because the two drift apart quickly otherwise.

Verdict: Verdict: Sample the chassis alone before sampling the modules, and sign off geometry, harness and capacity first, because a module corrected after the shell is approved costs a second round and a second sampling fee in the USD 50-150 band.

Sizing and shaping the opening order

MOQ 500 per reference is the working floor for a modular programme, and the number is structural rather than negotiable in spirit. Below it, the cutting marker does not amortise its setup, trim orders drop beneath supplier minimums, while the sample an AQL 2.5 plan demands takes too large a share of output.

How to shape 500 is the real decision. Concentrate on the chassis and keep the module buys light, since each base unit sold becomes a prospective pouch buyer and the attach data has not arrived yet. A reasonable opening shape is 400 base units in two colours alongside 100 apiece of two named pouches, leaving the reorder to follow whatever the pouch data shows.

Resist the temptation to spend the opening buy on breadth. Five colourways at 100 units each produces five runs too small to reveal preference, five trim sets to manage, and no clear signal at the end. Two colourways at 200 each produces one clean comparison and a defensible basis for the reorder.

Plan the reorder while the first order is still in production. Modular economics depend on the module line staying continuously available, and a two-season stockout removes a module from the customer's mental catalogue permanently. Booking material for the follow-on run early is the control that prevents it.

Compliance and labelling, before the listing goes live

Compliance is the item most likely to be deferred and the one least able to be deferred. A product listed without the correct labelling can be delisted, and relabelling goods that have already arrived is expensive and slow. Settle it ahead of artwork sign-off, since it sets label content.

Two practices make this manageable. Reuse one qualified material set across all colourways, so that a single test report covers the whole opening range. And request the reports before the goods ship rather than after, because a report obtained later cannot prevent a shipment that has already been held.

Takeaway: Takeaway: Assemble labelling, fibre content, country of origin and the restricted-substance file before the listing goes live, and treat a missing children's-product assessment as a reason to delay the launch rather than to relabel during the 35-50 day production run.

Channel economics and how to set the first price

Price setting for a modular range differs from a conventional one because the margin events are split. The chassis is a considered purchase and carries a normal margin. The modules are repeat purchases and carry a higher one. Pricing either side on the same logic leaves money on the table.

Work backwards from the channel. If the route is direct, the landed cost plus the desired contribution sets the price, and the brand retains the accessory margin on every module. If the route is wholesale, the retailer's required margin has to come out first, and the brand should check whether the module price still sits below the threshold at which a consumer deliberates, because above it the accessory does not convert.

Landed cost is the number most often guessed. Build it from the FOB Xiamen unit price at 500 units, add freight by mode, add duty at the correct classification, add terminal handling and onward delivery, and add a share of setup. A founder who skips this step discovers the true margin at the end of the first quarter, which is the worst time to learn it.

Finally, price the system rather than the unit. A chassis priced to make the modules feel inexpensive is a defensible strategy, because the attach revenue is where the compounding sits. That is the same logic that makes a platform purchase attractive to the customer, and it works from the brand's side as well.

The first ninety days of operation

The period after arrival is when a launch becomes a business, and it needs a cadence rather than a hope. Three activities carry it, and each produces something concrete.

Measurement comes first. Record attach rate, meaning the proportion of base-unit buyers who then add a pouch. Record return reasons coded by failure mode rather than by sentiment. Record interface complaints apart from appearance complaints. Those three figures show whether the grid, the assembly and the surface finish are each correct, with enough precision to draft the next brief directly.

Supply continuity comes second. Confirm the material position for the follow-on run before the opening stock is half sold, because a stockout during the first season costs more than the carrying cost of reordering early. Confirm the interface is unchanged, so the installed base stays compatible.

Specification tightening comes third. Hold a structured review against the original brief and identify the two or three lines that drifted or were never measured. Tighten them before the second buy rather than after, because the second buy is larger and the same ambiguity will cost proportionally more.

The second buy is where the model pays. Put the volume behind the colour and pouch that the figures support, hold the interface constant, and allow the bigger volume to absorb the setup already spent. That leaves the brand in a far stronger place than repeating a whole-range buy, and it is only available to brands that treated the first one as a trial.

Mistakes that cost a launch its first season

Four mistakes account for most first-season failures, and all four are avoidable at the planning stage rather than at the production stage.

The first is specifying the module catalogue before the chassis. It feels logical and it produces a set of mounting requirements that do not agree with each other, so the chassis becomes a compromise serving none of them well. Fix the mounting faces, the reinforcement behind them and the interface rule first, then design modules against a known condition.

The second is approving a sample on appearance. A sample that looks right and measures wrong carries its error into every unit of the run. Measuring the dimension callouts, the interface repeat, the bar-tack coverage and the closure cycles takes an hour and is the highest-return hour in the whole programme.

The third is treating compliance as a post-launch task. A listing without correct fibre content, origin marking or care instructions can be delisted, and rectifying goods that have already landed is slow and expensive. Settle the file before artwork is finalised, because labelling content depends on it.

The fourth is spreading the opening buy across too many variants. Breadth feels like risk reduction and is the opposite: five small runs produce no usable signal, while two produce one clean comparison that the reorder can act on. The discipline is to buy information with the first order and volume with the second.

A fifth mistake deserves mention because it is quieter and more common: failing to retain the approved sample. The signed unit is the reference against which the production run is inspected, and without it a disagreement about whether a change was requested becomes two accounts of a conversation rather than a comparison against an object. Retain the sample, retain the revision note that accompanies it, and update the specification document to match. Those three artefacts, held together, resolve disputes that no email thread ever will.

There is a pattern connecting all five mistakes. Each one trades a small amount of discipline now for a large amount of ambiguity later, and each one is cheapest to avoid at the point where it is available to avoid. Specification before catalogue, measurement before approval, compliance before listing, concentration before breadth, and retention before production. None of them requires additional budget. All of them require doing the unglamorous work first, which is the actual content of launching a range.

The two documents to prepare first are the interface rule and a shortlist drawn from the existing modular range.

Spec rule: Spec rule: Protect the first season by keeping the opening order at 500 units per reference, allowing 6-10 working days for sampling and 35-50 days for production, and holding release at AQL 2.5 inspection.