The true cost of getting inventory replenishment wrong
The fashion industry is grappling with a costly paradox. On one hand, up to 30% of potential sales are lost to stockouts (Zalando 2023).
Out-of-stock sizes rank as the top complaint among shoppers, and inaccurate size-level forecasting can wipe out 20% of the monthly margin* – bad for profitability and customer experience.
On the other hand, brands are sitting on mountains of unsold goods, with some estimates valuing the industry's overstock at up to $140 billion.
The cost to store this idle stock is also getting more expensive, with warehousing costs up by 10% in 2023.
Overstock triggers markdowns: In H1 2024, discounting climbed by five percentage points YoY (State of Fashion 2025).
"Brands are spending millions of euros a year storing unsold items—and then paying again to bring them back through reverse logistics."
This isn’t just about margin. It’s a systemic problem rooted in outdated forecasting and fragmented stock management. In this video from our Business Casual series, Karolina Kozłowska (Director of Product at ZEOS) explains how replenishment decisions impact everything from stockouts to stranded inventory and why fashion brands often lose margin long before a product hits the shelf.
EU regulation & climate risk: A ticking clock for inventory replenishment
These operational inefficiencies are now becoming compliance risks, too. As new EU regulation takes shape, brands are facing increasing pressure to quantify – and justify – how they manage unsold stock.
Between tightening regulations and unpredictable demand, inventory replenishment must shift from static planning to dynamic control.
Ecodesign deadlines you can't ignore
Inventory efficiency is no longer just about cost – it's becoming a matter of compliance.
In July 2024, the European Union formally adopted the Ecodesign for Sustainable Products Regulation*, setting out two major legal milestones:
2025: Brands must publicly disclose how they manage excess stock
2026: It becomes illegal to destroy unsold textiles within the EU
The implications are clear: poor stock allocation decisions are no longer just costly—they can also impact a brand's sustainability credentials and compliance with emerging regulations.
And this is not just a European issue. In August 2024, California became the first US state to approve an Extended Producer Responsibility (EPR) law for textiles, requiring apparel brands to submit a plan for collection, repair, and recycling by July 2030. Global fashion players must now navigate a multi-regional compliance map.
All this comes as 60% of brands remain behind on their sustainability targets*.
Climate change: When the weather breaks the forecast
Climate change is warping demand and complicating inventory management.
While 2024 set global heat records, much of Europe saw its coldest summer in a decade, rendering simple seasonal forecasting insufficient. Stock intended for early heatwaves remained unsold, and replenishment for colder items was delayed.
"Though many brands may not yet think about it this way, climate change is increasingly an inventory management challenge."
Consumer behaviour is evolving, with traditional collections misaligned with modern habits that prioritise immediate, weather-sensitive purchases.
Customers expect 'buy now, wear now' gratification, complicating regional demand in the global online fashion market.
This dynamic makes calendar-based forecasting risky, leading to misalignment between supply and demand.
Why e-commerce broke traditional replenishment
Historically, replenishment was relatively predictable. Brands sent stock from a central warehouse to stores based on seasonal plans. But e-commerce disrupted this model. Now, brands must coordinate replenishment across:
Multiple channels and regions, each with distinct demand signals
High return rates that distort inventory accuracy
Siloed systems that obscure real-time visibility and slow down stock decisions.
What was once a linear supply chain has become a fluid, dynamic ecosystem. The shift from planned distribution to on-demand fulfilment has made replenishment far more complex – and much harder to get right.
How data-driven demand forecasting powers smarter stock replenishment
To navigate this complexity, brands need to move from static, seasonal planning to dynamic, data-driven inventory replenishment.
Many currently rely on a mix of ERP systems, inventory management tools, or in-house forecasting layers.
But few are built for the complexity of e-commerce in fashion – where short product lifecycles, high return rates, and fragmented demand across channels and regions are the norm.
In contrast, AI-powered forecasting uses a range of signals to continuously align supply with actual demand, including:
Weather forecasts to anticipate demand for weather-sensitive items.
Customer wishlists and search data to gauge interest in new and existing products.
Fashion trends and insights using tools like Zalando’s Trendspotter to identify emerging styles and trends in different regions.
Return patterns to account for expected resellable returns when replenishing, avoiding return-induced overstock.
Dynamic replenishment flips the script: instead of forecasting once per season, you make SKU-specific decisions on a rolling basis.
The result is a continuous feedback loop that connects demand sensing, forecasting, and execution in near real time – forming the backbone of a more responsive replenishment planning process.
Tackling the cold start: Replenishment without sales history
One of the toughest challenges in fashion replenishment is the cold start problem – forecasting demand for products with little or no sales history. These could be entirely new SKUs, seasonal reworks, or styles reintroduced to new markets.
Unlike replenishable core items, these products offer few clues in historical data alone. But there are several ways fashion brands are tackling this issue:
Attribute-based modelling: Some use machine learning models trained on product metadata (e.g. style, colour, silhouette, fabric) to infer likely demand based on past analogues.
Image recognition and computer vision: Others analyse product imagery to identify visual similarities to bestsellers or trending items, enhancing forecast precision before launch.
Test-and-reorder strategies: Some brands release small quantities to gather real-world demand signals before committing to larger restocks, especially in volatile categories.
One such approach is Zalando’s FashionDNA, which uses visual and contextual data to connect new SKUs with patterns across over 100 million fashion images. This enables confident forecasting before the first unit is sold, which is particularly useful for seasonal styles and fast-evolving categories.
As Karolina Kozłowska explains, the system is designed to:
"...analyse customer choices, preferences, current/future trends, identify characteristics of new items, compare them to existing ones (like past bestsellers), find similarities/differences, and forecast future demand accurately."
Ultimately, the right solution depends on a brand's product mix, data maturity, and forecasting model.
Building a single stock pool
Most brands manage fragmented inventory across multiple locations – from local warehouses in their home market to long-term contracts with 3PLs.
On top of that, many use "fulfilled by" models, delivering stock into marketplace networks like Amazon, Zalando, Otto, Bol, Allegro and others.
In fact, 11 of the 22 largest marketplaces in Europe offer their own fulfilment services.
Fragmented inventory creates friction: Stockouts on one channel, oversupply on another and duplicated safety buffers that increase cost without adding value.
One way to address this is by adopting a single stock pool – which doesn't mean one warehouse. It means treating all stock as one virtual inventory network, regardless of whether it's destined for:
Owned e-commerce
Marketplace platforms
Social commerce channels
Regional fulfilment nodes
Creating an agile inventory network enables you to:
Reallocate inventory across markets and channels based on real-time demand
Minimise excess safety stock and storage costs
Solve demand imbalances created by regional style preferences
When a style is trending in one region but lagging in another, stock doesn't have to sit still. It can be redirected to where it's moving, keeping availability high without overbuying.
"People are buying different things in Paris and Warsaw – so there are different replenishment decisions in every marketplace and every country."
In this model, replenishment becomes not just faster, but smarter. It transforms from reactive guesswork into a responsive, insight-led process that supports both profit and customer satisfaction.
Replenishment that moves with the market
For many fashion brands, replenishment still hinges on static forecasts, scattered systems, and manual guesswork. But that's no longer enough. Fragmented inventory leads to missed demand, unnecessary discounts and capital locked in unsold stock.
Fixing replenishment doesn't require a tech overhaul. It starts with sharper inputs: cleaner signals, shared logic across teams and planning models that reflect how fast demand moves in fashion.
With better signals – across platforms, regions and product types – brands can plan with more confidence, cut waste and protect availability where it matters most.
That shift doesn't happen overnight, but the brands that get it right will gain more than a margin. They'll earn loyalty, unlock flexibility and build a replenishment model that's ready for what's next.
Follow how inventory operations evolve – one decision at a time – via our monthly newsletter, Expertly Packed.
