SKU Rationalization and Product Line Planning for Waterproof Bag Ranges

What SKU proliferation actually costs

SKU rationalization is usually presented as a tidy-up, which is why it loses every argument against the excitement of a new launch. It is not a tidy-up. It is the removal of a compounding cost that product line planning almost never prices, and the cost has five distinct components that behave differently as the range grows.

The first is tooling and setup. Every structurally distinct style carries cutting dies, welding fixtures or electrodes, patterns and process documentation. A style that is merely a colourway of an existing shell carries none of that; a style with a different panel geometry carries all of it. This is the cheapest component to measure and the one people remember.

The second is material. Each additional colourway is a mill minimum — commonly one hundred to three hundred metres per colour — and each distinct laminate specification is a separate procurement event with its own lead time. Hardware follows the same pattern: a different buckle finish or zipper length is a separate pack minimum. Material minimums mean that variety is paid for in inventory, not in unit price, which is exactly why it does not appear where anyone is looking.

The third is forecast error, and it is the largest. Forecast accuracy falls as the number of items being forecast rises, because each item has less history and more noise. Splitting the same total demand across ten SKUs instead of four does not keep total error constant — it increases it substantially, because each of the ten is harder to predict than each of the four. That increased error has to be absorbed somewhere, and it is absorbed as either stock-outs or safety stock, both of which cost money.

The fourth is dead stock, which is forecast error that has already happened. Long-tail SKUs are where dead stock lives: the bottom twenty per cent of a range typically carries fifty per cent or more of the excess inventory while contributing under five per cent of sales. The fifth is complexity itself — the planning time, the sample rounds, the photography, the listing maintenance, the packaging variants, and the number of things that can go wrong in a given week. Complexity does not scale linearly with SKU count; it scales with the number of interactions between SKUs, materials, suppliers and channels.

The true annual cost of one more SKU

Until this is written down, every SKU discussion is a matter of opinion and the opinion always favours adding. The exercise below takes twenty minutes per candidate SKU and changes the conversation immediately.

Add the low ends and the high ends and one added SKU costs somewhere between roughly thirteen hundred and eight thousand dollars a year before a single unit is sold. Against that, the median tail SKU in a bag range generates a few thousand dollars of gross margin in a good year. That is the arithmetic that makes rationalisation obvious, and it is also why the decision feels so different in the meeting: the cost is spread across five budgets and three years, while the revenue appears in one place immediately.

Two things make this number even less favourable in practice. The first is that the costs are mostly fixed and the revenue is probabilistic — a SKU that underperforms still carries all seven cost lines. The second is that added SKUs rarely replace existing ones; they are added alongside, so the range grows and each incremental SKU competes with SKUs that already exist. Cannibalisation is the hidden line: a new colourway that takes sixty per cent of its volume from an existing one has produced no net revenue and a full set of new costs. Our cost breakdown guide shows where these same lines sit on the unit-cost side.

The contribution-versus-complexity matrix

Once each SKU has a cost and a contribution, the decision is a two-axis plot rather than a debate. Contribution is what the SKU earns after variable cost and after an allocation of its own complexity cost. Complexity is how much organisational and supply chain burden it carries. Four quadrants fall out, and each has a clear action.

The counterintuitive part is the fourth quadrant. Low-contribution, low-complexity SKUs should mostly be kept, because a colourway on a shared platform in a running material costs very little to hold and does real work: it fills a shelf, gives a retailer a reason to list the range, and provides a customer a reason to choose you over a competitor with four options. The instinct to cut everything with low contribution destroys range credibility for a small saving.

The quadrant to attack is the third, and in a neglected range it is usually the largest. These are the SKUs that were built for one customer, for one season, or for one enthusiastic conversation, that carry custom tooling or an unusual laminate, and that nobody wants to defend because nobody remembers why they exist. Cutting them frees working capital, removes material minimums and eliminates forecasting noise with almost no revenue impact.

Score complexity honestly rather than by feel. A workable scoring rubric: tooling required (none, shared, dedicated), material status (running, occasional, bespoke), colour count per style, number of distinct components, and whether the SKU is made on the same line as the core range. Five yes-or-no questions are enough to rank the range and produce a defensible list. Market context for where the range should be heading sits in our market and range analysis.

Which SKUs to cut first: a ranked rule set

A matrix tells you which quadrant; a rule set tells you the order within it. Apply these in sequence and the list stops being a negotiation.

• Cut anything with no reorder in the last two cycles and more than six months of cover. It has already told you the answer.
• Cut anything whose custom tooling generates less annual contribution than the tooling cost to replace or maintain. Tooling that does not pay for itself is a subscription nobody cancelled.
• Cut bespoke laminates and bespoke hardware finishes before cutting anything else, because they carry procurement minimums and lead time that affect the whole range.
• Cut SKUs that duplicate a function already served by a better-selling item, unless they serve a genuinely different channel or price point.
• Cut sizes outside the two or three that carry eighty per cent of volume, unless a specific channel requires the extremes.
• Cut colourways below a defined contribution threshold, and keep the number of colourways per style capped rather than open.
• Never cut the item that anchors the range story, even if its own numbers are mediocre — it is doing marketing work that does not show in its contribution line.

Two cautions. First, cut in waves rather than all at once, because a single large deletion hides mistakes — if you remove twenty items and revenue holds, you will not know which two should have been kept. Two waves of ten with a quarter between them gives you a read on each. Second, check for attachments before deleting: a SKU that is a component of a bundle, a spare part for another product, or a contractual obligation to a retailer cannot be cut on contribution alone.

And run one specific test before finalising: the substitution test. For each candidate, ask what a customer who wanted it would buy instead. If the answer is a product in your range, the deletion is low-risk. If the answer is a competitor, you are not cutting a SKU, you are exiting a demand segment — which may still be right, but should be a deliberate decision rather than a spreadsheet outcome.

Platform and modular design: shared chassis, changed exterior

The alternative to deleting variety is to make variety cheaper, and that is what platform design does. A platform is a common structural core — panel geometry, weld fixtures, base laminate, closure system, harness attachment — across which several visually distinct products are built. The customer sees a range; the factory sees one construction.

• One weld fixture set serves four exterior designs, which removes three tooling projects from the range.
• One base laminate is bought in volume across the platform, which removes three mill minimums and improves price through concentration.
• One closure subsystem is validated once, so the waterproof claim is tested per platform rather than per SKU.
• Operators build the same core repeatedly, so line efficiency rises and defect rates fall, because the learning curve is not reset every time a new style starts.
• Safety stock can be held at the material and sub-assembly layer for the whole platform rather than per finished SKU, which is the cheapest place to hold it.

Modularity goes a step further: the range becomes a chassis plus a set of interchangeable modules. Strap systems, pocket configurations, closure options and branding panels attach to a common body. Done well this gives genuine customisation at low complexity, because the modules are the variety and the chassis is the volume. Done badly it produces a product that is more expensive to assemble than the monolithic version it replaced, and the test is simple: if a modular design costs more per unit and takes longer to build, it is not modularity, it is just complication.

The construction decision underneath all of this matters more than most brands realise. Choosing one primary joining method across the platform — welded, taped or bonded — determines which fixtures, which skills and which validation regime the whole range shares. Mixing methods without a reason is one of the most common sources of unnecessary complexity in bag ranges, and the trade-offs are set out in our construction methods guide. The hardware selection guide applies the same logic to components: standardise the interface, vary the visible part.

Where variety is cheap and where it is expensive

Not all variety costs the same, and brands that understand the difference can offer a broader range at lower cost than competitors who do not. The dividing line is whether the variation touches the tooled, validated structure or only the surface.

Read the table and the strategy writes itself: give customers variety in colour, print and size, and ruthlessly standardise closure, laminate and geometry. That is how a range can look broad to a shopper while being narrow to a factory — and it is the difference between a brand with forty SKUs that is profitable and one with forty SKUs that is not.

Two housekeeping rules keep this discipline alive. First, cap colourways per style — three to five for a mid-range product — and require a new one to displace an existing one rather than be added. Second, treat every bespoke laminate as a platform decision rather than a SKU decision, because a laminate chosen for one style will end up as a procurement and testing burden for years. Colour and material choices are covered in more depth in our colour customisation guide, and sizing economics in the size customisation guide.

The politics of cutting a SKU

Every analytical argument above is correct and none of them are why cutting is hard. Cutting is hard because every SKU has an owner, and the owner’s incentives are not the company’s.

• A salesperson owns a SKU because a key account asked for it, and losing it feels like losing the account even when the account buys ninety-five per cent of its volume elsewhere.
• A founder owns a SKU because they designed it, and the emotional attachment is immune to a contribution table.
• A designer owns a SKU because it is the best-looking thing in the range, and aesthetics do not appear in the complexity score.
• A regional manager owns a SKU because their market is different, which is sometimes true and always asserted.
• Nobody owns the cost of complexity, which is why it always loses the argument against five specific, articulate owners.

Three mechanisms work. The first is to give complexity an owner: name the cost of the range as a line in someone’s objectives, because unowned costs lose every meeting. The second is to change the default: require a business case with the cost table above before any SKU is added, which stops the accretion at the source and is worth more than any cut. The third is to make cuts reversible in perception — a SKU can be retired rather than killed, with an agreed review date and an agreed trigger for bringing it back.

Retirement in particular dissolves most of the political resistance. A SKU that is "paused pending review" with a clear criterion for reinstatement is far easier to agree to than one that is "deleted", and in practice a paused SKU that nobody asks about for two quarters is deleted without a fight. The organisation gets the same outcome with none of the argument.

One more tactic: cut as a package rather than item by item. Negotiating forty SKUs individually gives forty opportunities for an owner to win. Presenting a range plan that removes the bottom quartile as a coherent strategy, with the reinvestment of freed capital stated explicitly, gives one argument that is much harder to defeat. The reinvestment matters — a cut that looks like decline mobilises opposition; a cut that funds a launch does not.

Why cutting before you have the numbers never works

The most common failure is a rationalisation project that starts with the analysis of which SKUs to cut and ends with a spreadsheet nobody acts on. The reason is almost always that the numbers were not credible to the people who had to agree.

Three specific credibility gaps kill these projects. Contribution is computed on revenue rather than margin, so a high-revenue, low-margin SKU looks like a keeper. Complexity is scored by feel rather than by a rubric, so every owner can argue their SKU is simple. And the cost of capital is ignored entirely, which is the largest single line for slow-moving stock and the one most often omitted because it does not appear in a P&L.

There is also a sequencing error. Projects that begin with "which SKUs should we cut" invite defence. Projects that begin with "here is what the current range costs to operate, per SKU, per year" produce a different conversation, because the cost is the finding rather than the proposal. Publish the cost table first, let it land for a month, and the cut list largely writes itself — with the additional benefit that the owners of expensive SKUs have already started defending them internally before the meeting.

The last requirement is a reinvestment plan agreed before the cut. Rationalisation that only removes product reads as retreat, and retreats get reversed as soon as revenue dips. A plan that says "we are removing fifteen SKUs and using the freed capital and attention to launch two platforms" reads as strategy, and strategy survives a bad quarter. That link between range decisions and capital is the same one explored in our brand building guide from the growth side.

Designing the replacement range

A rationalised range is not a smaller version of the old one. It is a designed structure with defined roles, and building it deliberately is what prevents the range from re-accreting within a year.

Assign every SKU a role before the range is finalised, and write the retirement date for seasonal and campaign items at the moment of approval rather than at the moment they stop selling. That single discipline — a pre-agreed end date — is the most effective defence against re-accretion, because it makes the decision once instead of every quarter.

Then check the range against three structural tests. Coverage: does it serve the price points your channels require? Laddering: is there a clear reason for a customer to trade up, visible in the product rather than only in the price? And platform economy: do the majority of units share a chassis, a laminate and a closure? A range that fails the third test will drift back into complexity within two seasons no matter how well the other two are done.

What to do with the tooling and material you free up

Cutting a SKU does not automatically recover its cost; the recovery comes from actively dealing with what it leaves behind. Four asset classes need a decision.

• Tooling: dedicated dies and fixtures for a deleted SKU are usually worthless except as scrap, but shared-platform tooling may still serve surviving items. Establish ownership and whether it can be modified rather than scrapped — a welding fixture for one geometry is sometimes adaptable to a related one at a fraction of new-tooling cost.
• Material: bespoke laminate and custom-coloured webbing from a deleted SKU is the real recovery opportunity. Convert it into a surviving product if the specification allows, or negotiate its return to the mill. Carrying it "in case" is the default outcome and the worst one.
• Components: standard hardware is genuinely reusable and should be consolidated into the platform store; custom-moulded parts usually are not, and should be written off decisively.
• Packaging and printed collateral: printed cartons, inserts and hangtags for deleted SKUs are pure waste. Count it once, because the number is usually large enough to change behaviour next time.

The ownership question on tooling deserves attention before the next programme, because it determines whether you can move the work or modify the tool later. Mould and fixture ownership, and what happens to it when a SKU is discontinued, is covered in our tooling and mould ownership guide, and the packaging side is treated in the packaging options guide.

Do the accounting honestly afterwards. Record the write-off in the same quarter as the decision rather than spreading it, because spreading it hides the cost of the original mistake and makes the next one easier to approve. A visible write-off is the cheapest form of organisational learning available.

SKU count and minimum order quantity: the compounding effect

SKU count interacts with minimum order quantities in a way that makes range indiscipline far more expensive than it looks, and it is the interaction most brands never model.

With a minimum of 500 pieces per style, each SKU carries a floor commitment of 500 units regardless of how fast it sells. A range of twenty-five SKUs therefore carries a minimum inventory commitment of twelve thousand five hundred units before any question of whether the market wants them. Reduce the range to twelve and the floor falls to six thousand — the same range breadth released several thousand units of working capital, with no change to anything else.

The compounding comes from colourways. If a minimum applies per style per colour in a custom programme, then five styles in four colours is twenty commitments, not five. That is why colour discipline matters so much more than it appears to: a colourway is not a variation, under a per-colour minimum it is a purchase commitment. Standardising colourways across the platform, or choosing materials where colour can be bought in a shared run, is one of the highest-return range decisions available.

The implication for range planning is direct: SKU count should be decided jointly with minimum quantity, not independently. A range that looks affordable at a unit level can be unaffordable at a commitment level, and the test is simply total commitment units against available working capital. That test belongs in the range plan, and it is the same arithmetic that drives the consolidation argument in our MOQ and production slot guide.

A ninety-day SKU rationalisation sequence

Rationalisation projects fail from scope creep rather than analysis. Ninety days is enough if the sequence is disciplined, and the order below is designed so that each stage produces something the next one needs.

Two rules make the sequence work. Publish the cost table before proposing any cut, because it converts the meeting from a defence of individual SKUs into a discussion of a finding. And socialise individually before presenting collectively, because an owner surprised in front of colleagues becomes an opponent for reasons that have nothing to do with the SKU.

Measure the outcome on three numbers rather than one: total inventory value, working capital released, and units of the range that still turn at an acceptable rate. Revenue will usually dip slightly in the first quarter and then recover as attention and capital concentrate on the survivors. A rationalisation that does not produce a visible improvement in inventory turns within two quarters was probably not executed, whatever the cut list said.

Keeping the range disciplined after the cut

The range will grow back unless the mechanism that grew it is changed. Four controls do most of the work, and all four are process rather than analysis.

• A business case gate: no SKU is added without the cost table completed and a named owner for its contribution. This is the single most effective control and it costs almost nothing to run.
• One-in-one-out on colourways and sizes: adding one requires retiring one, which forces the trade-off into the open rather than deferring it.
• Pre-agreed retirement dates for seasonal and campaign items, set at approval.
• A quarterly range review with a standing agenda: contribution table, aged stock, and SKUs with no reorder in two cycles. Reviews that only discuss new products are how ranges accrete.

Add one forward-looking discipline: an annual platform review. Once a year, ask which structural elements are shared across the range and which have drifted apart, because drift is gradual and invisible month to month. A platform that started as one chassis and one laminate can become four of each over three years without anyone deciding it.

The commercial frame worth keeping in front of the team is that range discipline is a growth tool, not a restraint. Brands with focused ranges launch faster, forecast better, hold less dead stock and put more engineering attention into fewer products — which is why their products are usually better. Every hour spent governing the range is an hour not spent liquidating it later.

If you want to plan a range against real production economics rather than assumptions, review our own process from first enquiry through sampling into bulk production and send us your proposed range, target volumes and channel mix. Minimum order quantity is 500 pieces per style, sampling takes 6–10 working days, bulk production runs 35–50 days, quotations are issued FOB Xiamen, and a shared-platform range can usually be quoted with fewer tooling lines and shorter material lead times than the same breadth built independently. The management case for reducing product complexity is set out in the research published by Harvard Business Review, and the operational planning methods follow the frameworks maintained by ASCM.

Frequently Asked Questions

Q1. What does an extra SKU really cost per year?

Typically between roughly thirteen hundred and eight thousand dollars once tooling amortisation, material minimums, incremental safety stock, dead stock provision, sampling, listing and planning time are counted. Most of it never appears as a line item.

Q2. Why does SKU count increase forecast error?

Because each item has less history and relatively more noise. Splitting the same demand across ten items instead of four increases total error, which is absorbed as stock-outs or safety stock.

Q3. Should I cut every low-contribution SKU?

No. Low-contribution items built on a shared platform in running materials cost little to hold and provide range credibility. Cut low-contribution, high-complexity items first.

Q4. What is platform design in a bag range?

A shared structural core — panel geometry, weld fixtures, base laminate, closure system — across several visually distinct products. Customers see a range; the factory sees one construction.

Q5. Which kinds of variety are cheap to add?

Colourways in running materials, size grades of an existing design, and print or branding artwork. All three touch the surface rather than the tooled, validated structure.

Q6. Which kinds of variety should be standardised instead?

Closure systems, base laminates and panel geometry. Each one carries tooling, procurement minimums and a validation regime, so varying them costs far more than the visible difference is worth.

Q7. How do I handle the politics of cutting someone’s SKU?

Give complexity an owner, require a business case before additions, and retire rather than delete — a paused SKU with a reinstatement criterion is far easier to agree to than a deleted one.

Q8. Why do rationalisation projects stall?

Usually because contribution was computed on revenue rather than margin, complexity was scored by feel, and the cost of capital was omitted. Publish the cost table before proposing any cut.

Q9. Should I cut in one wave or several?

Several. A single large deletion hides mistakes; you will not learn which two items should have been kept. Two waves with a quarter between them gives a read on each.

Q10. What is the substitution test?

For each candidate, ask what a customer who wanted it would buy instead. If the answer is another product in your range, the deletion is low-risk; if it is a competitor, you are exiting a demand segment deliberately.

Q11. How does a 500-piece minimum affect range decisions?

Each SKU carries a floor commitment. Twenty-five SKUs is a twelve thousand five hundred unit commitment before demand is known, and per-colour minimums multiply it further.

Q12. What should I do with tooling from a deleted SKU?

Check whether it can be modified for a related geometry before scrapping it, and confirm ownership terms. Shared-platform tooling often still serves surviving items; dedicated tooling usually does not.

Q13. How many colourways should a style carry?

Three to five for a mid-range product, with a one-in-one-out rule. Under a per-colour minimum, a colourway is a purchase commitment rather than a variation.

Q14. What roles should a designed range contain?

A hero or anchor, a volume core on a shared platform, an entry price point, seasonal or campaign items with retirement dates, and any channel-specific requirements — each with a defined job.

Q15. How long should a rationalisation project take?

Ninety days is realistic: two weeks of data, two weeks of costing, two weeks of ranking, two weeks of testing, two weeks of socialising, two weeks of execution.

Q16. Will revenue drop after cutting SKUs?

Usually slightly in the first quarter, then it recovers as capital and attention concentrate on the survivors. Judge the outcome on inventory turns and working capital released, not on revenue alone.

Q17. How do I stop the range growing back?

A business case gate for every addition, one-in-one-out on colourways and sizes, pre-agreed retirement dates for seasonal items, and a quarterly review with a standing agenda.

People Also Ask

What is SKU rationalization?

The structured removal of products that cost more in complexity than they earn in contribution, using a contribution-versus-complexity matrix rather than intuition.

How much does an extra SKU cost?

Commonly thirteen hundred to eight thousand dollars a year across tooling, material minimums, safety stock, dead stock, sampling, listing and planning time.

Which SKUs should be cut first?

Low contribution and high complexity: items with no recent reorder, bespoke laminates or hardware, custom tooling that does not pay for itself, and duplicates of better sellers.

Should I cut low-contribution colourways?

Usually not if they sit on a shared platform in a running material. They cost little to hold and provide range credibility that drives sales of the core.

What is platform or modular design?

A shared structural core across several visually distinct products. It gives customers variety while the factory produces one construction.

How do I stop SKU count growing back?

Require a business case for every addition, apply one-in-one-out on colourways, set retirement dates at approval, and hold a quarterly range review.