Why is sell-out data essential for AI in fashion wholesale?

KEY TAKEAWAYS Summary by the editors

1. Sell-in data records what a brand ships to retailers; sell-out data records what those retailers sell to consumers, by product, size, store and period.
2. AI models for wholesale forecasting, re-orders and allocation need sell-out data because sell-in reflects retailer buying decisions and timing, not consumer demand.
3. Standard formats exist: GS1 Austria's fashion guideline for the EDIFACT SLSRPT message transmits sales quantities by GTIN, outlet and period between trading partners.
4. Under the EU Vertical Block Exemption Regulation in force since 1 June 2022, exchanging aggregated customer purchase data is generally acceptable in dual distribution, while future resale prices are on the list of problematic exchanges.
5. Sell-out data is most useful with stock data alongside it, because sales without stock cannot distinguish weak demand from sold-out sizes.

Sell-out data is essential for AI in fashion wholesale because it is the only direct measure of consumer demand in partner stores. Sell-in, what the brand ships, reflects the retailer's buying decisions, budgets and timing. A model trained only on sell-in learns how retailers order, not how consumers buy, and its forecasts, re-order suggestions and allocation advice inherit that blind spot.

What is the difference between sell-in and sell-out data?

The two can diverge sharply. A retailer may order heavily and then sell slowly, which looks like success in sell-in until the returns, cancellations and markdown support requests arrive next season. Equally, a style may sell out quickly in stores while sell-in shows nothing new because the retailer has no open budget to re-order.

Why does AI need sell-out data?

Most AI use cases in wholesale depend on knowing what consumers actually bought:

• Demand forecasting: next season's buy plans are better grounded in consumer sell-through than in what retailers ordered.
• Re-order and replenishment suggestions: a model can propose re-orders to an account only if it sees that account's sales and remaining stock.
• Size curves: retailer orders often follow standard size ratios, so only sell-out reveals each account's true size demand.
• Assortment recommendations for buyers: suggesting which styles suit an account requires knowing what has sold there before.
• Early warning: slow sell-through in partner stores signals markdown risk weeks before it appears in returns or order cancellations.

Sales alone are not enough. When a size sells out, sales stop, and a model reading only sales concludes demand fell. Research by Akchen and Caro on a footwear retailer found that nearly 25 percent of demand left unmet by stockouts shifted to adjacent sizes, which means raw sales misstate size demand in both directions. Sell-out data therefore needs stock on hand by product and location, ideally at the same frequency.

Read also

What makes a wholesale business AI-ready?

How do brands receive sell-out data from retailers?

Several routes exist, often in parallel for different accounts:

1. EDI sales reports: standard messages such as EDIFACT SLSRPT. GS1 Austria's guideline for the fashion sector describes it as carrying sales quantities by product (GTIN), sales location (outlet GLN) and period between trading partners.
2. Retailer portals and file exports: many department stores and marketplaces provide reports through supplier portals, often as spreadsheets in retailer-specific formats.
3. Concession and consignment models: where the brand owns the stock on the shop floor, it typically sees sales directly.
4. Shared platforms: some brands and retailers exchange data through common B2B or replenishment platforms with agreed data definitions.

The technical precondition is a shared product identifier. GS1 specifies that every product variation, meaning each size, each colour and each combination, needs its own GTIN. When brand and retailer both use the brand's GTINs, sell-out can be matched to the brand's article master without manual mapping. When the retailer uses its own codes, mapping errors become a major source of bad data.

What are the legal limits on sharing sell-out data?

Data exchange between a brand and its retailers is a vertical relationship, but many fashion brands also run their own stores and web shops, which makes them competitors of their retail customers (dual distribution). The EU's revised Vertical Block Exemption Regulation, Regulation (EU) 2022/720, in force since 1 June 2022, and its accompanying guidelines address information exchange in exactly this situation. According to an analysis by law firm Hogan Lovells, exchanges that are generally acceptable include aggregated information on customer purchases that does not relate to identified end users, and comparisons of sales performance among distributors on an anonymised basis. Information on the future prices at which supplier or buyer intend to sell is on the list of exchanges unlikely to be covered.

How can brands make retailers want to share?

Retailers share data when they get something back. Common value exchanges include automated replenishment that reduces their stock risk, size and assortment recommendations for their next order, faster re-order lead times, or shared markdown planning. Agreements should define which data is shared, at what level of aggregation, how often, who may use it for what purpose, and how long it is kept.

Smaller retailers often lack the systems to send structured data at all. For them, a simple, low-effort route, such as a regular stock and sales upload in a fixed template or a direct connection from their point-of-sale software, matters more than any formal standard. Coverage of the long tail is what turns a sample of large accounts into a usable picture of the market.

Read also

What is demand sensing and how does it work in fashion?

Where do sell-out initiatives typically fail?

• Data arrives in many formats and nobody owns the mapping to the brand's product master.
• Sales are shared without stock, so the data cannot support replenishment or size analysis.
• Reports arrive monthly, too late for in-season decisions.
• Coverage is patchy: a few large accounts share, the long tail of smaller retailers does not, and models over-weight the large accounts.
• The brand collects data but has no process to act on it, so retailers see no benefit and stop sending.

For AI, the lesson is straightforward: a forecasting or re-order model is only as good as its view of the consumer. For wholesale brands, that view is sell-out data, matched to clean product identifiers, accompanied by stock, and delivered often enough to act on.

Frequently asked questions

What is sell-out data in fashion?

Sell-out data records what a retailer sells to consumers, by product, size, colour, store or web shop and period. For a brand selling through wholesale partners, it is the closest measure of real consumer demand.

What is the difference between sell-in and sell-through?

Sell-in is what a brand ships to a retailer. Sell-through is the share of that stock the retailer sells to consumers over a period, calculated from sell-out data and stock.

How can a fashion brand get sell-out data from retailers?

Through EDI sales report messages such as SLSRPT, retailer supplier portals, concession or consignment models where the brand owns the stock, or shared B2B platforms. Most brands combine several routes and need clear data-sharing agreements.

Is it legal for brands and retailers to share sales data in the EU?

Generally yes, within limits. Under the 2022 Vertical Block Exemption Regulation, aggregated purchase data not relating to identified end users is generally acceptable, while exchanges about future resale prices are problematic, particularly in dual distribution. Legal advice is recommended.

GuideThe complete guide to AI in fashion merchandising and buyingRead the complete guide