SKU profitability for DTC brands — landed cost done right.

01 The four components of landed cost

Landed cost is the fully-loaded unit cost of inventory at the point it is available for sale in the brand's warehouse or 3PL. Four components together. The supplier invoice alone is component one. Most brands stop there, and most brands are wrong by 18%–35% as a result.

Component one — factory cost

The supplier FOB invoice in the supplier's currency, converted at the transaction-date FX rate (or monthly average if applied consistently), with quantity discounts, amortised tooling, and factory-side inspection fees included. The correct cost basis is weighted average across POs — Settle's Product Cost Report and Finaloop's InventoryIQ both compute landed cost on WAC and post nightly to A2X. A single static "cost per item" in Shopify, never refreshed against actual invoices, is the most common silent failure here.

Component two — inbound freight and duties

Ocean or air freight, customs duties, import taxes, broker and compliance fees, insurance, drayage, port and documentation fees, demurrage. Allocated across the SKUs in a shipment by value (standard for freight and duty), by quantity (when SKUs are physically similar), or by weight or volume (when freight is weight-driven). This component typically runs 15%–30% of total landed cost on imported goods. It is also the component where the gap to true landed cost most often lives, because freight rates have moved 40%–80% peak-to-trough since 2022 and most brands refresh the SKU file annually if at all.

Component three — 3PL receiving and inbound handling

Inbound receiving fees, pallet-in, put-away — the work that gets the SKU from inbound dock to available-for-sale in the brand's warehouse system. Typically $0.05–$0.20 per unit at a mid-market 3PL. This is the component every brand ignores because the dollars per unit are small. The correct treatment is to capitalise inbound 3PL into inventory and keep outbound pick-and-pack as fulfilment expense below gross margin. Crossing the line — capitalising outbound or expensing inbound — distorts both gross margin and contribution margin, and the QoE will find it.

Component four — allocated overhead

The QA team's time, the supplier-management function, the inventory financing cost on working capital tied up in stock, the inventory shrinkage reserve, the logistics and operations payroll directly attributable to getting product ready for sale. The pragmatic DTC FP&A approach is a fixed dollar load per unit (annual pool divided by total units purchased) or a 5%–10% percent-of-landed-cost load. Strict GAAP keeps overhead out of book inventory cost; pragmatic FP&A includes it for pricing and channel decisions. We run both layers and reconcile them.

The four components, summed at SKU level on WAC, are the landed cost. The supplier invoice alone is component one. The discipline is to track all four, refresh on the cadence the volatility demands, and use the same allocation method across time so historic landed costs are comparable.

02 Component two is where the gap lives — freight and duties since 2022

The freight and duties component is where most of the gap between book and reality opens up. Drewry's World Container Index hit $10,377 per 40ft at the September 2021 pandemic peak. The 21 May 2026 reading was $2,076 — roughly 80% below peak but still 46% above the 2019 pre-pandemic baseline of $1,420. The Shanghai-Los Angeles spot lane sat at $2,713 per 40ft in the same May 2026 read, with the West Coast component having spiked to $11,197 ahead of a Chinese New Year squeeze before falling back. Sharp cycles, not straight-line trends — and the path from 2022 to 2026 has had at least three distinct freight regimes inside it.

The brands that get this most wrong are the ones whose finance team built a SKU file in 2021 or 2022 using peak-spot freight, then never refreshed. Their landed cost is too conservative for normal periods, they have been underpricing or over-promoting against an inflated cost basis, and they have been compounding the error every time a new SKU launched on the same template. The opposite failure — refreshing once at the 2023 trough and never adjusting — leaves the brand exposed when rates spike again. S&P Global's 2026 Ocean Freight Outlook calls this out explicitly: rates may ease as capacity expands, but "volatility remains firmly embedded" and the right model is banded (base contract, modest surcharge, stress spike), not single-point.

Duties are the second half of the component, and the 2025 tariff regime has made the work materially harder. The Section 122 global tariff added 10% ad valorem to most non-exempt imports from all countries. Section 301 China-specific duties continue at a 20% base (some lines 0%, some 100%+). Section 232 metals duties moved from 25% to 50% for steel articles and many derivatives, with copper and aluminum following the same path and lumber moving up in stages. The tariff stack on a women's woven cotton trouser imported from China now runs roughly 58.6% — 28.6% MFN plus 20% Section 301 plus 10% Section 122. On a face cream from China the stack is closer to 32.8%; on a stainless steel water bottle classified into a covered Section 232 line, total duty exposure can clear 80%. The point is not the specific rate; the point is that the rate that was right in 2023 is almost certainly wrong now.¹

HTS classification is the third sub-component and the one where most penalties accrue. The 10-digit HTS code drives the duty rate, the Section 301 exposure, and any Section 232 overlay. Misclassification errors run in both directions: a brand can overpay by defaulting to a high-duty "other" subheading when a more specific line exists, or it can underpay (and face back-duty assessment plus penalties) by picking a cheaper code that does not actually describe the goods. We see both errors in roughly equal measure in the rebuilds. The discipline is to classify at the full 10-digit level, maintain a classification memo per SKU under the General Rules of Interpretation, and review the top 20% of SKUs by duty spend at least annually.

Practically, the work product is a banded freight-and-duty model: by lane (at minimum Asia-USWC, Asia-USEC, Asia-Europe, plus a separate air or expedited line for emergency replenishment), by Incoterm, by HTS code, refreshed quarterly against actual carrier rates and confirmed duty assessments. Brands that run the refresh quarterly and capitalise the result back into the SKU file have the right number to price against. Brands that run it annually or never do not.

03 Component four is where the judgment lives — allocated overhead

Overhead allocation is the component most brands skip because it requires judgment about what is genuinely inventory-related versus what is period expense. The mechanical answer matters less than the discipline of doing it consistently. We use a three-layer model on every rebuild.

The inventory-related pool

Logistics and QA payroll attributable to receiving and inspecting product, supplier-management headcount, inventory systems and WMS, customs and trade-compliance work, the inventory financing cost on working capital tied up in stock at current debt rates. Annual pool divided by total units purchased — for a brand buying 1.5M units a year against a $500K inventory-related ops pool, that is $0.33 per unit. Same allocation across all SKUs unless there is a clear driver argument for a different basis.

The shrinkage and obsolescence reserve

Trailing-12 inventory shrink as a percent of inventory value, plus a forward-looking obsolescence reserve on slow-moving and seasonal SKUs. Apparel and beauty brands with high SKU complexity carry meaningfully higher reserves than supplements or commodity household — 2%–4% of inventory value at the high end, sub-1% at the low end. The reserve gets allocated by SKU based on the category-level shrink experience, not a blended brand-wide rate.

The FP&A overlay vs the GAAP book

Strict GAAP inventory cost includes only direct manufacturing and inbound-related costs. Pragmatic FP&A landed cost — what we use for pricing, range decisions, and channel mix — adds the overhead pool on top to capture the economic unit cost. We run both layers in parallel. Book inventory is GAAP-clean for the audit and the QoE; the pricing decisions get made off the FP&A overlay. The two reconcile on a single bridge: book unit cost plus overhead load equals FP&A landed cost.

Skipping the allocation produces high-confidence wrong SKU profitability for the SKUs that are most overhead-intensive — typically the slow-moving long-tail SKUs that consume disproportionate QA, supplier-management, and shrink reserve without driving proportional revenue. Range rationalisation done on supplier-invoice cost lets these SKUs survive longer than they should. Range rationalisation done on four-component landed cost surfaces them in week one.

04 The range-rationalisation conversation

Once the four-component landed cost is in place, the range-rationalisation conversation becomes data-driven. The same pattern shows up across the rebuilds: a long tail of SKUs that look acceptable on supplier-invoice gross margin but are contribution-negative or sub-15% CM3 once landed cost and variable cost are properly allocated. The percentage varies by catalog size and category.

In our sample of 18 rebuilds and consistent with the broader operator literature: lean catalogs under 50 SKUs run 10%–20% contribution-negative on true landed cost. Mid-size catalogs (50–250 SKUs) run 20%–35%. Wide catalogs (250+) run 30%–50%, with the worst case being a 700-SKU apparel brand where 48% of SKUs were CM-negative once size-and-colour variants were broken out separately. The Pareto pattern underneath the headline number is more useful than the headline: across DTC brands the top 10% of SKUs typically drive 60%–75% of contribution dollars, the next 20% drive another 15%–25%, and the bottom 70% contribute 0%–10% of CM dollars — frequently net-negative once allocated ad spend is included.

The four-bucket decision framework

1. 01
   Stars (CM3 ≥ 35%, top quartile of CM dollars): Promote and expand. These are the SKUs that earn paid traffic, inventory commitment, and merchandising real estate. The pricing-floor conversation defends them; the new-SKU pipeline extends the family.
2. 02
   Workhorses (CM3 25%–35%, meaningful volume): Keep and protect. These do not earn aggressive paid acquisition but they hold the assortment together and absorb fixed-cost overhead. Reprice at the margin, bundle to lift AOV, but do not cut.
3. 03
   Strategic gateways (low CM% but unlock LTV or attach): Keep, instrument, and review quarterly. Loss-leaders that exist to drive first-order conversion or to anchor a bundle are legitimate; loss-leaders that exist because nobody has looked at them in two years are not. The instrumentation question — does this SKU actually attach? — is what separates the two.
4. 04
   Dogs (CM3 < 20%, low volume, no strategic role): Discontinue. Phase out over 6–12 months for retail-listed items, faster for pure-DTC. Free the working capital, free the operational complexity, free the merchandising calendar. This is the bucket that typically holds 15%–30% of the SKU range in our rebuilds.

Hershey's Mexico case study is the canonical CPG analogue: a 25%–30% portfolio cut, 10–20 top SKUs driving over 80% of revenue, a 6–12 month phase-out for retail-listed items, and meaningful margin expansion on the surviving core.² AArete's published price-point retail engagement landed 2%–3% of gross margin from the rationalisation pass alone. In our DTC practice the numbers run similar: 200–400 bps of gross margin recovery on the surviving SKUs, plus the working-capital release from the discontinued tail.

05 Pricing decisions against the true cost

With the four-component landed cost in place, pricing decisions become defensible against a floor that actually exists. Three conversations change.

Promotional depth

The promotional-depth question — how deep can we go on this SKU and still hit our contribution target — has the right floor underneath it. We see brands that have been running 40%-off promotions on SKUs whose true CM3 at full price was 32%; the promo is destroying contribution every time it fires. With the right landed cost, the promo calendar gets rebuilt against a minimum-CM3 floor (we usually anchor at 20% on promo, 35% at full price) and the destructive promos get pulled before they ship.

New-SKU pricing

New-SKU launches get a pro-forma landed cost built before the buy commits, against banded freight (base, surcharge, stress) and current duty stack. The launch price is set to clear a 40% CM3 target at the base-case freight, with the stress-case freight still clearing 25% CM3. Brands that launch SKUs without this discipline reliably end up with a long tail of new product running at sub-15% contribution that nobody owns the decision to discontinue.

Wholesale and retail negotiation

Retail-buyer negotiations have a number the brand can defend. The buyer always pushes for the lower wholesale cost. The discipline of knowing the true landed cost — with freight, duty, and overhead capitalised — means the brand can defend the floor that gives the channel acceptable contribution without saying yes to a wholesale margin it cannot live with. The brands that lose the most contribution in retail negotiations are the brands that walked in with cost-of-goods-only.

06 FX and tariff sensitivity — the model that should live next to the SKU file

The single most useful artefact that comes out of the rebuild, alongside the SKU file itself, is a sensitivity model that decomposes landed cost into its drivers and lets the operator stress each one independently. The formula sits underneath every banded model we build:

Landed_unit_cost = (FOB_factory × FX_rate) × (1 + Tariff_stack) + Freight_per_unit + 3PL_inbound + Overhead_per_unit

Stress cases that should be run quarterly: USD-CNY ±5%, USD-VND ±5%, freight ±30% on each lane, Section 122 ±5%, Section 301 +25% (the tail risk on a renewed China escalation). The brands sourcing 60%+ from China that have not modelled the renewed-escalation scenario will be flat-footed when the announcement comes; the brands that have are already three months into supplier diversification or pricing-pass planning. We covered the CNY trajectory and the diversification implications in a separate piece; the relevant point here is that the SKU file has to be the input to the sensitivity, not the output.

The other model that should live next to the SKU file is the returns-adjusted contribution view. Returns are the second-most-mismodelled line in DTC finance after landed cost, and the two interact — a SKU with 22% return rate and 9% landed-cost gap is materially worse than the file suggests. The returns-reserve work travels with the landed-cost work in every rebuild; we wrote up the methodology separately for finance teams running both in parallel.

07 The refresh cadence and the operating rhythm

A four-component landed cost done once is a one-time benefit. The same work done on a quarterly cadence, integrated into the operating rhythm, is the structural advantage. The cadence we run on the engagements:

1. Monthly: refresh component one (factory cost) against actual PO receipts, WAC-weighted; refresh component three (3PL inbound) against actual 3PL invoices. Both are mechanical updates from accounting data.
2. Quarterly: refresh component two (freight and duties) against current lane rates and any tariff changes; refresh component four (overhead allocation) against current ops pool and unit volume. Both require manual review.
3. Annually: re-validate HTS classifications on top 20% of SKUs by duty spend; re-validate allocation methodology (value vs quantity vs weight) per cost type; re-validate overhead pool composition.
4. On every new-SKU launch: pro-forma four-component landed cost before the buy commits, against banded freight and current duty stack, with a stress-case CM3 floor.
5. On every range-rationalisation review (quarterly): rank SKUs by CM3 percentage and CM dollars, classify into the four-bucket framework, action the dogs and reprice the bottom workhorses.

The operating discipline is more important than the analytical sophistication. A brand that runs a basic four-component landed cost on a quarterly cadence with discipline will out-margin a brand that runs a sophisticated activity-based-costing model once and never refreshes. The SKU file is a living artefact, not a project deliverable. It pays for itself every quarter the freight market moves, every quarter a tariff line changes, every quarter a new SKU launches.

The work product matters into the sell-side process too. The QoE checklist we run with sellers in the 18-month exit-prep window opens with the landed-cost rebuild because the gross-margin defensibility line in a sale process turns directly on it. Brands that arrive at the process with a clean four-component landed cost, a documented refresh cadence, and a defensible range-rationalisation history clear 0.5–1.0 turns of EBITDA multiple above peers who do not. The buy-side QoE on consumer brands almost always finds the gap if the seller has not surfaced it first; the working-capital peg negotiation amplifies the damage if the diligence team's landed cost is materially different from the seller's.

08 Five questions for the next pricing review

1. Is the SKU file running on supplier-invoice cost only, or on four-component landed cost with all four components allocated at SKU level?
2. When was the freight-and-duties component last refreshed against current lane rates, current HTS classifications, and the full Section 122 / 301 / 232 tariff stack? If the answer is over a quarter ago, the file is wrong.
3. What percent of the SKU range is contribution-negative on true landed cost? If you do not know, the answer is almost always between 15% and 35%.
4. Is the new-SKU launch process running pro-forma four-component landed cost against banded freight scenarios before the buy commits, or are you discovering the margin after the SKU is in market?
5. Is the refresh cadence on the calendar — monthly factory and 3PL, quarterly freight/duty and overhead, annual HTS validation — or is the rebuild a one-time project?

The cost of getting this wrong is not the spreadsheet error. It is the eighteen months of pricing decisions, promotional depth, and range expansion made off the wrong number. The brands that have done the work — four-component landed cost, quarterly refresh, range-rationalisation calendar, banded sensitivity model — carry 200–400 bps of gross margin advantage over the brands that have not, and they carry it into every pricing review, every promo decision, every new-SKU launch, and every sell-side process they open.