How to architect revenue operations for an athletic footwear brand in 2027

The scenario that forces the rebuild

Picture a brand doing roughly $180M in annual net revenue: about 55% wholesale through national chains, regional run specialty, and international distributors; about 35% direct-to-consumer through its own site and four outlet doors; and about 10% through marketplaces like Amazon and Zalando. Three franchises — a daily trainer, a trail shoe, and a lifestyle retro — carry most of the volume, each in seasonal colorway drops, each in a men's and women's size run spanning roughly fourteen to eighteen sizes.

The problem surfaces the same way in every brand this size. The wholesale team runs its forecast in a spreadsheet keyed to account and style. The DTC team runs its forecast in the ecommerce platform keyed to SKU and channel. Planning runs allocation in an ERP keyed to a material number that only partially maps to either. Marketing measures acquisition in a paid-media dashboard that stops at the style level and has no concept of size. Finance closes the month in a fourth system where returns land as a lump credit two months after the sale that generated them.

Nobody is wrong, and everyone disagrees. A merchandiser asks why the trail shoe is late and gets three answers. A wholesale director commits an at-once fill to a key account without knowing DTC has already sold through half the size curve. A DTC lead runs a promotion on a colorway that wholesale is about to reorder at full margin, and the account calls to complain about price erosion inside a week. The pattern is not a people problem. It is an architecture problem: the operating unit of the business is a style-color-size, and none of the systems agree on what that is.

Athletic footwear makes this sharper than most categories for four structural reasons. First, size curves mean a style is never sold out — it is broken, and a broken size run still shows inventory on hand while converting at a fraction of the rate. Second, lead times from Vietnam, Indonesia, and China typically run four to six months for production plus three to six weeks of ocean transit, so the commitment is made long before the demand signal exists. Third, DTC return rates on footwear commonly land in the high twenties to mid thirties percent because fit is a physical variable a product page cannot resolve. Fourth, wholesale and DTC compete for the same units on the same calendar, and without a governing rule they will cannibalize each other while both hit their individual targets.

The architecture question is therefore not "which tools do we buy." It is: what is the smallest set of shared definitions, owned by one accountable team, that lets every channel decide from the same numbers — and what governance forces those decisions to be consistent when the channels want different things.

How the mechanism actually works

The working architecture has four layers, and the order matters. Skipping to layer three because a vendor demo looked good is the single most common way these builds fail.

Layer one — the identity spine. Everything hangs off a canonical product identity resolved to style, color, and size. In practice that means one internal SKU key that maps deterministically to the ERP material number, the GTIN or UPC on the box, the ecommerce platform's variant ID, each marketplace's ASIN or item ID, and the account-specific numbers that big wholesale partners assign. This mapping table is the most valuable asset revenue operations owns. Build it as a governed dimension with an owner, a change process, and a nightly validation job that flags any SKU appearing in a transaction without a mapping. The validation matters more than the initial build: new colorways arrive every drop, and an unmapped SKU silently disappears from every downstream report.

Alongside product, resolve two more entities. Account identity for wholesale — one canonical customer record where a chain's distribution centers, banners, and buying offices roll up to a parent, because margin and sell-through only make sense at the parent. And customer identity for DTC — an identity graph keyed on hashed email and order history so a guest checkout, an app purchase, and an outlet transaction collapse into one person for lifetime-value math.

Layer two — the demand signal. The spine is useless without sell-through. Wholesale reporting is the hard part: EDI 852 product activity data from accounts that send it, retailer portal exports from those that do not, and distributor reports on whatever cadence you can negotiate into the contract. Aim for weekly at the door-cluster and size level from your top accounts, which usually represent seventy to eighty percent of wholesale volume, and accept monthly style-level data from the long tail rather than delaying the build for perfect coverage. DTC and marketplace sell-through arrive natively. Land all of it in one warehouse table with the same grain: date, SKU, channel, location cluster, units, gross revenue, discount, returns.

Layer three — the decision engines. Three of them, all reading layer two. Demand forecasting produces a size-curve-aware unit forecast per SKU per channel across the buy horizon. Allocation decides which units go where when supply is short. Pricing and promotion governs markdown timing and depth across channels against a single calendar.

Layer four — the governance surface. Dashboards, alerts, and a standing weekly meeting where merchandising, planning, wholesale, DTC, and finance look at one screen. The technology is trivial; the discipline of a single screen is what actually changes behavior.

mermaid flowchart LR Q["Contested units in short supply"] --> R{"Confirmed wholesale PO<br/>to strategic account?"} R -->|Yes| S1["Fill in full"] R -->|No| T1{"Forecast contribution<br/>margin per unit"} T1 -->|"DTC higher"| U1["Allocate to DTC"] T1 -->|"Wholesale higher"| V1["Allocate to wholesale"] T1 -->|"Within 5 percent"| W1["Split by prior-season<br/>sell-through rate"] S1 --> X1["Reserve 5-10 percent<br/>for at-once"] U1 --> X1 V1 --> X1 W1 --> X1 X1 --> Y1["Release reserve on stated date"] Y1 --> Z1["Log every override for<br/>quarterly rule review"] </parameter> </invoke> </function_results>The override log in that final node is the part teams skip and the part that makes the rule survive. If overrides run above roughly ten percent of allocation decisions, the rule is wrong and should be rewritten rather than repeatedly bypassed.

Common pitfalls and how to avoid them

Aggregating away the size grain. The most expensive mistake. A style showing four thousand units on hand looks healthy until you see that ninety percent of it is sizes 7, 13, and 14. Every core table should carry size, and every sell-through dashboard should show a size-run health indicator — something as simple as the percentage of core sizes with more than two weeks of cover. Avoid it by making size a required column in the warehouse contract and failing the build when a source lacks it rather than silently rolling up.

Treating returns as a finance adjustment. If returns land in the model as a monthly reserve, DTC looks structurally more profitable than it is and the brand over-invests in paid acquisition against phantom margin. Fix it by attributing returns to the originating order and SKU, reporting DTC contribution margin net of returns and return shipping, and refusing to approve acquisition budgets against gross revenue.

Building dashboards before the spine. A dashboard on unreconciled data trains the organization to distrust the whole function, and that trust is hard to rebuild. Ship one narrow, correct view — weekly sell-through for the top ten accounts at size level — and expand only after it survives a quarter without a disputed number.

Letting wholesale and DTC run separate promotional calendars. A DTC flash sale on a style a key account just bought at full price generates a markdown claim, damages the relationship, and often costs more than the promotion earned. Maintain one pricing calendar covering all channels, with a blackout rule around account drop windows and a required approval whenever a DTC promotion touches a style with open wholesale orders.

Forecasting new franchises like carryover. A first-season launch has no history, and applying a carryover model to it produces a confident, wrong number that drives a bad buy. Use analog-based forecasting — pick two or three comparable prior launches, adjust for planned marketing spend and door count, and express the result as a low-mid-high range with an explicit reorder trigger rather than a single number.

Ignoring the long tail of wholesale reporting. Chasing perfect EDI coverage from every small account delays the build for months. Cover your top accounts weekly, model the tail from shipment data and periodic style-level reports, and label modeled figures clearly in the interface so nobody mistakes an estimate for a measurement.

No owner for the SKU mapping table. When new colorways arrive unmapped, they vanish from reporting and reappear as a variance at quarter close. Assign a named owner, run the nightly unmapped-SKU validation, and route failures to a queue with a service-level expectation of one business day.

Measuring channel performance without allowances. Wholesale gross margin excluding markdown money, co-op, and chargebacks overstates wholesale profitability and will push the brand toward the wrong channel mix. Load allowance accruals at account level and report a fully-loaded wholesale margin next to DTC contribution margin so the comparison is honest.

Over-indexing on tooling. No platform resolves whether wholesale or DTC gets the contested units. Write the rules first, then choose tools that can execute them.

Related questions

How long before a rebuilt revenue operations function pays for itself?

Most of the return comes from markdown avoidance and better buy accuracy rather than headcount savings. Expect the first measurable effect one full buy cycle after the forecast goes live — typically two to three quarters — because the commitment window sits months ahead of the sale.

Should DTC and wholesale share one demand forecast?

One forecast, channel-split outputs. A single unit forecast at style-color level, allocated across channels by rule, prevents the double-counting that happens when two teams independently forecast demand for the same physical inventory.

What is the minimum viable version of this architecture?

A canonical SKU mapping table, weekly size-level sell-through from your top accounts plus DTC, and one shared dashboard reviewed in a standing weekly meeting. That combination captures most of the value before any forecasting engine exists.

Does this change if the brand is DTC-only?

The spine and returns handling stay identical; allocation and channel-conflict governance mostly disappear. The size curve and returns-reason analysis actually matter more, since there is no wholesale channel to absorb broken size runs.

How do marketplaces fit into the size curve problem?

Marketplaces are an effective disposal path for broken runs because buyers self-select by size. Feed them prior-season remainders deliberately rather than letting them absorb current-season inventory and undercut your accounts.

FAQ

What is the first thing to build?

The canonical SKU spine at style-color-size grain, with a validated mapping to every downstream system identifier and a nightly job that flags unmapped SKUs. Nothing else in the architecture produces trustworthy output without it, and every month you delay it adds unmapped product from the drops that shipped in the meantime.

How much wholesale sell-through coverage is enough to start?

Roughly seventy percent of wholesale net revenue with weekly, size-level data is a workable starting point, ideally rising above ninety-five percent within two quarters. Concentrate on the top accounts first, model the tail explicitly, and label modeled figures in the interface so estimates are never mistaken for measurements.

Should we buy a retail planning suite or build on a warehouse?

Build the spine and reporting on a warehouse for control over footwear's size grain, and consider a packaged demand planning tool if your planners already know one. The mistake is buying a suite expecting it to solve identity resolution — that work is yours regardless of tooling.

How do we stop DTC promotions from damaging wholesale relationships?

One pricing calendar across all channels, a blackout window around account drop dates, and a required approval for any DTC promotion on a style with open wholesale orders. Enforce it in the promotion workflow rather than by policy memo, since memos lose to quarter-end revenue pressure.

What metrics should the weekly review actually show?

Sell-through rate by franchise and channel, size-run health for core styles, weeks of cover against the forecast, forecast variance since the last review, returns rate with reason-code mix for DTC, and open allowance accruals by account. Six numbers, one screen, same definitions for every team in the room.

How do we forecast a brand-new franchise with no history?

Use two or three analog launches, adjust for planned marketing spend and door count, and publish a low-mid-high range rather than a point estimate. Pair it with a defined reorder trigger and reserved short-lead capacity so an upside case is recoverable instead of becoming a stockout.

Sources

• https://www.census.gov/retail/index.html
• https://www.bls.gov/cpi/
• https://www.gs1.org/standards/id-keys/gtin
• https://x12.org/products/transaction-sets
• https://nrf.com/research
• https://www.bis.doc.gov/
• https://usfashionindustry.com/
• https://www.trade.gov/industry-analysis
• https://hbr.org/topic/subject/operations-and-supply-chain-management

Related on PULSE

• How to build a size-curve-aware demand forecast for a footwear brand
• Wholesale versus DTC allocation rules that survive a short supply season
• Reducing DTC footwear return rates with fit data and reason codes
• What belongs in a canonical SKU spine for apparel and footwear
• Designing a single cross-channel pricing and markdown calendar
• Measuring fully-loaded wholesale margin including allowances and chargebacks