One stock ledger for webshop, marketplaces and stores cuts promotional overselling by about 70%

The retailer sells through its own brand webshop, third-party e-commerce marketplaces and physical retail channels, and kept adding new sales channels as the business grew.

Each platform had its own order back office and its own stock settings.

Although the company used an ERP for purchasing and finance, a great deal of manual work remained between channel orders, warehouse operations and after-sales returns.

During promotions, some products were prone to overselling, while some returned items sat back in the warehouse for days before becoming available for sale again.

The challenge

Stock figures differed across channels

Each platform synchronised stock at a different frequency.

When inventory included reserved, awaiting-inspection and unsellable items, the available quantity shown on platforms was easily distorted.

Order processing was fragmented

Order statuses, shipping information and cancellation rules differed from platform to platform.

Operations staff had to switch between multiple back offices.

Returns did not reach the inventory and finance systems in time

Once a customer initiated a return, the refund, the physical return of goods, quality inspection and restocking were often handled by different people.

SHEPHARD's digital solution

Step 1: Establish unified product and channel mapping

Using the company's ERP product master data as the foundation, we maintain the mapping between internal SKUs, platform SKUs, product variants and sales channels.

We defined which system owns product information, pricing, stock and order status.

Step 2: Build unified order management

Orders, cancellations and refund events are received through platform APIs or other supported integration methods.

Orders from every channel follow a single business model while retaining each platform's original order number.

Step 3: Define rules for available-to-sell stock

Physical stock, reserved quantities, quality-blocked stock and channel allocation rules all feed into a single available-to-sell calculation.

Fast-moving products use appropriate event-based synchronisation and stock buffer strategies to reduce the risk of overselling caused by data latency.

Step 4: Build a reverse logistics process

Customer returns are managed through RMA (return merchandise authorisation).

Return requests, return shipping, warehouse receipt, quality inspection, stock disposition and refund status are all linked.

Depending on the item's condition, it is restocked, refurbished, scrapped or handled in another way.

Step 5: Build product and channel performance analytics

Revenue, fulfilment costs, refund rates and true gross margin are analysed by channel, product, campaign and order.

This lets the company assess the profitability of each channel, rather than simply comparing sales volumes.

Implementation approach

We first unified SKU and stock data, then connected the main sales channels.

Order fulfilment, reverse logistics and channel financial analysis were introduced step by step.

Before major promotions, order peaks, API rate limits, stock sync latency and duplicate events must be stress-tested.

Business value

The company now has a single view of orders and stock, and can coordinate resources across sales channels more effectively.

Returns no longer stop at the customer service system; they now flow through to warehouse stock, finance refunds and performance analysis.

Results (approximate): Overselling during promotions approx. −70%; order sync latency from hours to minutes; returns processing time 7→3 days.

Metrics tracked on an ongoing basis: Overselling rate, order sync latency, order fulfilment cycle time, returns processing time, channel contribution margin, days inventory outstanding.

What this case teaches us

Omnichannel retail is not the same as opening several sales platforms at once.

True omnichannel digitalisation means every channel shares trustworthy product, order, inventory and customer service capabilities.

*To protect client confidentiality, company details in this case study have been anonymised; screenshots are illustrative and results are rounded approximations.*