A customer visits your store, finds the product they’re looking for, and then sees “out of stock.” That sale is lost in a matter of seconds. They’ll often go to a competitor instead, and they’ll hesitate before coming back.
A stockout therefore costs far more than restocking in advance, especially for an ecommerce retailer selling in Switzerland, France, and the EU.
The internal article "Integrating eCommerce Sites, Marketplaces and WMS in Switzerland". cites data from McKinsey. It shows that real-time inventory management reduces stockouts by up to 35%.
The central question of this article is simple: how to forecast ecommerce sales to avoid stockouts.
We address this question methodically, covering definitions, the data to gather, forecasting methods, and the crucial role of real-time inventory visibility.
The highest-performing companies invest in demand forecasting, inventory accuracy, and predictive fulfillment to improve customer loyalty and product availability. according to a McKinsey study
This topic applies equally to sellers on their own websites and those on Swiss marketplaces such as Galaxus.
A stockout costs more than early restocking
A stockout costs more than early restocking
Stockouts account for 65.6% of total inventory distortion; excess inventory accounts for 34.4% in 2026, according to IHL.
Reordering in advance has a known cost: a few extra days of tied-up cash.
A stockout, on the other hand, incurs diffuse costs that are difficult to measure. Not only do you lose an immediate sale, but you also lose customer trust and, on a marketplace, your seller ranking.
Google’s search and Shopping algorithms, as well as the criteria used by Galaxus (Switzerland’s largest marketplace), factor product availability into their rankings.
The math almost always favors the same approach. Reordering a few days early is less expensive than a week of stockouts on a fast-moving product.
Note that approximately 6.3% of global sales are affected by stockouts, according to IHL, Sales forecasting is specifically designed to find this balance: ordering enough, neither too early nor too late.
Demand, sales, and inventory forecasting: don’t confuse these 3 concepts
These three types of forecasts answer different questions. Mixing them up leads to ordering the wrong quantity at the wrong time. Here’s a clear definition of each.
Demand forecasting
Demand forecasting estimates the market’s appetite for a product, regardless of your ability to sell it. It measures how many customers would like to buy it over a given period.
It relies on signals such as the sales velocity of your best-selling products, seasonal peaks, and the product life cycle (launch, growth, maturity), as detailed in the internal article "2026 Trends: Ecommerce Sales Forecast, Zero Stockouts"
Sales forecasting
Sale forecasting, defined in detail in this IBM article, determines the physical quantities to be held to meet projected demand. It links the forecast to actual conditions: current inventory levels, supplier lead times, and storage locations.
Sales forecasting therefore breaks down the performance of each distribution channel. It is used to plan your sales targets and budgets.
Inventory forecasting
Sales forecasting, as detailed in this IBM article, determines the physical quantities that need to be held to meet projected demand. It links the forecast to actual conditions: current inventory levels, supplier lead times, and storage locations.
A warehouse management system (WMS) is software that manages and optimizes these inventory movements. Inventory forecasting informs replenishment decisions.
Confusing these three concepts leads to tangible errors. You may have high estimated demand, but a saturated distribution channel and poorly allocated physical inventory.
Always separate the three before approving a supplier order.
The true cost of stockouts (and overstocking)
Online demand is more volatile than in-store demand. The internal article "E-logistics: Optimizing Logistics for eCommerce" points out that seasonal peaks, marketing campaigns, and trends make demand difficult to anticipate. This volatility amplifies two opposing risks: stockouts and overstocking.
Lost sales and customer trust
The first consequence of a stockout is the immediate loss of a sale. The second, more long-term consequence affects brand reputation and customer trust.
Every order canceled due to a stockout erodes the customer relationship, as highlighted in the internal article "Integrating eCommerce Sites, Marketplaces and WMS in Switzerland".
A disappointed customer rarely returns to the same place.
Penalties on marketplaces and advertising
On a marketplace, recurring out-of-stock situations lower your seller ranking,
as Galaxus points out. The same internal article notes that overselling and out-of-stock situations harm your reputation as a seller.
Your listings lose visibility, and your advertising budget is spent driving traffic to products you cannot deliver.
The hidden cost of overstocking
The flip side of the problem is overstocking. Too much inventory ties up cash flow and increases storage costs.
The Keboola tool indicates that AI-powered automated forecasting and replenishment reduce stockouts by up to 30% and decrease overstocking by up to 25% [1].
The Saras Analytics guide adds that accurate forecasting reduces overstocking, stockouts, rush-order costs, and unnecessary discounts [2].
Situation
Business Impact
Cost to the Company
Occasional stockout
Lost sale and customer redirected to a competitor
Immediate loss of revenue and wasted customer acquisition costs
Repeated stockouts on a marketplace
Lower seller ranking and loss of visibility
Sustained drop in sales, less profitable advertising budget
Overstocking
Slow-moving products, forced discounts
Tied-up cash, storage costs, and overstock—up to 25% of which is avoidable
Before choosing a tool, estimate—even roughly—the cost of a stockout for your own catalog. Multiply the margin of a best-selling product by the number of days it’s likely to be out of stock. This figure will justify your forecast budget.
Data to gather before forecasting your sales
A forecast is only as good as the quality of the data that feeds it. Gather three categories of data before starting any project.
Sales history by SKU and channel
Consolidate your metrics at the SKU (Stock Keeping Unit, a product’s unique identifier) level: sales velocity, product lifecycle, and performance by channel (your own website, Swiss marketplaces such as Galaxus).
The internal article "Integrating eCommerce Sites, Marketplaces and WMS in Switzerland". details these key metrics.
The Darwin AI guide recommends consolidating at least two years of transactional data at the SKU level, including returns and cancellations [3].
Real-time inventory levels (warehouses, 3PL, transit)
Connect your inventory levels across all locations: your own warehouses, 3PL (Third-Party Logistics) providers, and goods in transit.
Without this unified view, a discrepancy arises between the displayed inventory and the actual inventory. This discrepancy skews all your replenishment calculations. A forecast based on inaccurate inventory leads to incorrect orders.
Promotional calendars and external factors
Include external factors that influence demand. The Darwin AI guide cites promotional calendars, price history, marketing spend, web traffic, weather, and search trends [4].
The 42Signals guide summarizes these requirements into three categories: sales history, current market indicators, and future events [5].
It also emphasizes the need for preliminary data cleansing to filter out non-representative one-off spikes, such as an isolated clearance sale.
Here is a minimum checklist of items to gather before getting started:
- 2 years of sales data by SKU and channel, including returns and cancellations
- Real-time inventory levels across all locations
- Replenishment lead times by supplier
- Promotional calendar and known sales events
- Traffic data and marketing spend
Methods, KPIs, and safety stock for accurate forecasting
Once the data has been collected, several forecasting methods can be used. The choice depends on the stability of demand and the volume of data.
Analysis of historical trends and seasonal adjustments
For products with stable demand, basic methods are often sufficient. The Sellerchamp guide recommends applying a 3- or 6-month moving average, or exponential smoothing, to your weekly sales [6]. These calculations highlight the underlying trend.
Seasonal and event-based adjustments then correct this baseline. Sellerchamp recommends factoring in calendar events such as Black Friday, sales, and the holiday season.
The internal article "Integrating eCommerce Sites, Marketplaces and WMS in Switzerland". lists these same seasonal peaks. Compare the previous year’s figures with current trends.
Artificial intelligence forecasting
Artificial intelligence and machine learning identify complex patterns in large volumes of data.
The Darwin AI guide states that these systems generate forecasts that are 30 to 50% more accurate than traditional statistical methods [7]. They also handle SKU-level forecasting for thousands of products simultaneously.
To measure accuracy, use the MAPE (Mean Absolute Percentage Error) metric, which expresses the average deviation between the forecast and actual results.
Darwin AI suggests aiming for an improvement of at least 20% over your existing method before validating a new model.
Safety stock and reorder point
Safety stock is a buffer quantity held to account for unforeseen events, such as supplier delays or sudden spikes in demand.
The reorder point is the inventory threshold that triggers automatic restocking.
The internal article "Integrating eCommerce Sites, Marketplaces and WMS in Switzerland". explains this threshold mechanism. The Sellerchamp guide provides a general rule of thumb: allow for 20 to 30% buffer stock on highly variable products [8].
For a new product with no sales history, pre-orders allow you to gauge demand. The Ecwid guide recommends opening pre-orders before investing in inventory [9].
Method
Principle
Ideal for
Limitations
Moving average / exponential smoothing
Project the trend based on past sales
Products with stable demand
Does not capture seasonal peaks
Seasonal adjustments
Adjust the baseline based on calendar events
Products subject to sales and holidays
Relies on reliable historical data
AI/ML forecasting
Detect complex patterns in large volumes
Extensive product catalogs, volatile demand
Requires clean and well-organized data
Pre-orders for new products
Gauge actual demand before placing an order
Launches with no historical data
Works only for products available for pre-order
Set an automatic reorder point by SKU rather than reacting after a stockout occurs. This threshold turns a manual decision into a rule that’s applied every day.
Real-time inventory visibility: the gap that software tools don’t cover
Purely software-based forecasting tools analyze historical data. They don’t see the actual status of physical inventory in the warehouse. This is the blind spot that direct integration with a WMS resolves.
A WMS connected to all sales channels
The workflow is simple: as soon as an order is confirmed on your website or a marketplace, it is transmitted to the WMS.
The WMS updates inventory in real time across all connected channels (Shopify, WooCommerce, Magento, marketplaces), as described in the internal article "Integrating eCommerce Sites, Marketplaces and WMS in Switzerland".
The internal article "Integrating eCommerce Sites, Marketplaces and WMS in Switzerland". confirms this central role, featuring marketplace connectors such as Lengow and Iziflux.
This workflow improves forecasting itself. Reliable, continuously synchronized inventory data eliminates discrepancies between displayed and actual inventory. Your restocking calculations are then based on accurate data.
Warehouses in Switzerland, France, and across Europe to refine forecasting
Strategically located warehouses in Switzerland and neighboring France allow products to be stored where they sell the fastest.
The internal article "Integrating eCommerce Sites, Marketplaces and WMS in Switzerland". presents this real-time view, synchronized across both regions, with a single inventory count.
The "Guide to Integrating eCommerce and Warehouse Management (WMS) in Switzerland" specifies that this approach covers both B2B and B2C flows as well as marketplaces such as Galaxus.
It is this combination that explains the reduction in stockouts of up to 35% mentioned above: accurate forecasting, driven by actual inventory levels.
Many Swiss and French fulfillment providers emphasize the speed of logistics execution. Few offer a forecasting tool directly connected to warehouse operations.
Emaloja distinguishes itself through this combination of WMS technology and physical infrastructure.
Check whether your current WMS reports inventory levels in real time to your forecasting tool. This is a prerequisite before investing in more sophisticated forecasting software.
What tools should you use to forecast your sales and inventory?
Three categories of tools complement each other: dedicated forecasting software, AI-powered centralized data tools, and a WMS solution integrated with physical logistics.
Dedicated forecasting software focuses on planning.
Onir is an example designed for ecommerce teams in Switzerland, Germany, and Austria [10].It helps forecast demand, plan inventory, and execute restocking actions using practical workflows.
AI-powered centralized data tools aggregate data from various sources. Keboola centralizes sales, ERP, supplier, and marketing data for AI-driven forecasting and automated restocking [11]. It transforms reactive management into proactive decision-making.
For small businesses, no-code automation is often sufficient.
The internal article "Inventory optimization with Artificial Intelligence (AI)". mentions connectors between a spreadsheet and an AI tool, or visualization tools, set up with little or no code via Zapier.
Tool Type
Example
Primary Use Case
Limitations for Ecommerce Businesses
Dedicated Forecasting Software
Onir
Forecast demand and plan restocking
Remains disconnected from physical inventory without a WMS
Centralized AI Data Tool
Keboola
Centralize sales, ERP, and marketing data for AI
Requires clean, structured data
No-code automation
Zapier
Connect spreadsheets and AI tools on a tight budget
Limited capacity for large volumes
Logistics-integrated WMS
Emaloja
Synchronize actual inventory with sales channels
Requires a connected warehouse infrastructure
These software tools remain disconnected from the reality of the warehouse unless a real-time WMS feeds them data. Choose a forecasting tool based on its ability to connect to an existing WMS, not on the number of features it offers.
Real-world example: anticipating a surge in demand before Black Friday
Let’s take the example of a Swiss ecommerce retailer that sells on its own website and on Galaxus. Four weeks before Black Friday, sales of a flagship product surge significantly. Without proper planning, the retailer will run out of stock right in the middle of the campaign.
The retailer cross-references two sources. First, the sales history for the same product during the same period the previous year. Second, the real-time inventory levels reported by the WMS.
This cross-referencing allows the retailer to trigger a restock before reaching the reorder point, rather than after running out of stock. The order arrives before the peak, not during it.
During the campaign, each sale updates the available inventory in real time across all connected channels. The internal article "2026 Trends: Ecommerce Sales Forecast, Zero Stockouts" describes this automatic update, which prevents overselling between the website and Galaxus.
If stock shortages become a reality, effective customer communication can minimize the damage.
The same internal article recommends immediately informing customers of the restocking timeframe, suggesting a similar alternative product, and offering a notification when the item is back in stock.
This way, you retain the potential sale and maintain customer contact.
Turn this scenario into an internal checklist to be followed before every known peak period (sales, Black Friday, the holiday season): compare historical data, verify actual inventory levels, expedite restocking, and prepare messages about stock shortages.
FAQ on ecommerce sales forecasting and stockouts
Combine an analysis of your historical data with a real-time view of inventory. Consolidate at least two years of sales data by SKU and by channel, then factor in seasonal and promotional trends. Set an automatic reorder point for each product to trigger restocking before stock runs out. The key is to link these forecasts to a WMS that reflects actual physical inventory.
The Sellerchamp guide suggests a general rule of 20 to 30% safety stock for highly variable products [12]. Adjust this percentage based on your supplier’s replenishment lead time: the longer the lead time, the higher the buffer should be. A fast-moving product with a short replenishment lead time can tolerate a lower buffer.
Use pre-orders to gauge demand before committing to inventory, as recommended by the Ecwid guide [6]. You can also group the new product with similar items that are already selling—by color, size, or material—to use their demand profile as a guide. Market research complements these signals at launch.
Update your forecast at least once a month for a stable product catalog, and weekly as a sales peak approaches. Inventory, on the other hand, should be monitored continuously using the WMS, without waiting for the next forecasting cycle. The more volatile the demand, the more frequently you should update it.
Yes. Recurring out-of-stock situations lower your seller ranking and reduce the visibility of your listings, as explained in the internal article "Integrating eCommerce Sites, Marketplaces and WMS in Switzerland". The algorithm favors reliable sellers who can deliver on time. Real-time inventory synchronization between your website and the marketplace prevents overselling, which triggers these penalties.
Yes. No-code automations connect a spreadsheet to an AI tool without writing code, using solutions like Zapier, as described in the internal article "Inventory optimization with Artificial Intelligence (AI)". A small store can therefore automate its forecasting without a technical team. The key is to start with clean sales data and reliable inventory figures.
In summary: building a zero stockout strategy
Here are the key points of this method:
- Clear separation of demand forecasting, sales, and inventory
- Quantifying the actual cost of stockouts and overstocking for your product catalog
- Minimum data requirements (2 years of sales data per SKU, real-time inventory, promotional calendar)
- A method tailored to volume—ranging from exponential smoothing to AI—with a reorder point per SKU
- A forecasting tool linked to a WMS that reflects actual physical inventory
Accurate forecasting is no longer a luxury but a prerequisite for profitability, as highlighted in the internal article "2026 Trends: Ecommerce Sales Forecast, Zero Stockouts"
Emaloja combines WMS technology with a physical logistics infrastructure in Switzerland, France, and Europe.
The group draws on the experience of STAR Logistique, which has been in operation since 1988, as detailed in the internal article “Why emaloja Is the Leader in Ecommerce and Logistics in Switzerland in 2026.”
The coverage of warehouses in Switzerland, France, and Europe is presented on the ecommerce page.
To apply this method to your product catalog, request a supply chain audit or a customized forecast from emaloja.
You’ll then know if your current inventory is replenishing quickly enough to avoid the next stockout.
Swiss eCommerce Sales Forecasting to Eliminate Stockouts