Update: We originally published this article in 2012 as a brief look at how Nike responded to a major inventory management problem. The more useful story, however, is not simply that Nike experienced a software failure and recovered. It is how forecasting, system integration, product complexity, and implementation decisions combined to create excess inventory in some products and shortages in others.
Inventory software can help a business make better purchasing and replenishment decisions. But software alone cannot compensate for unreliable data, disconnected systems, rushed implementation, or recommendations that are not reviewed against real-world conditions.
Nike’s experience remains a useful reminder that even a sophisticated company can run into serious inventory problems when technology and operations are not aligned.
What Happened to Nike’s Inventory?
In 2000, Nike began using new demand-and-supply planning software as part of a larger effort to modernize its supply chain technology. The goal was straightforward: use demand forecasts to determine how much footwear should be produced and where inventory would be needed. Instead, the implementation contributed to the wrong inventory mix.
In its 2001 annual report, Nike said difficulties implementing the new planning software created two problems: excess footwear inventory overall and insufficient inventory for key products that retailers had ordered. Those inventory issues also contributed to a decline in the company’s gross margin. Contemporary reporting placed the resulting sales shortfall at approximately $80 million to $100 million.
Nike did not simply lack inventory. It had too much of some footwear and too little of the products customers actually wanted. Inventory management is not only about having enough stock. It is about having the right products, in the right quantities, at the right time.
The Problem Went Beyond a Software Bug
It is tempting to describe the incident as a forecasting-software failure. The available records suggest the implementation problems were broader. A federal court opinion related to the incident described complaints about inconsistent rules between software components, manual integration work, difficulty handling Nike’s large number of SKUs, poor communication with existing systems, and slow performance.
An inventory system does not operate by itself. It must exchange dependable information with product records, purchasing workflows, supplier activity, order systems, and existing technology. When those connections do not work as expected, even a sophisticated forecasting model may produce recommendations based on incomplete or poorly interpreted information.
The lesson is not that businesses should avoid inventory software. It is that the implementation must receive as much attention as the software selection.
Lesson 1: Test the System With Real Operational Complexity
A software demonstration may show that a system can process an order, update a quantity, or generate a forecast. That does not necessarily prove it can handle the full complexity of the business. Testing should reflect actual conditions, including:
- The number of products and variations
- Real order volume
- Multiple warehouses or suppliers
- Seasonal demand changes
- Returns and inventory adjustments
- Connections with existing systems
- Exceptions such as partial receipts and backorders
Nike’s product volume and supply-chain complexity were much greater than those of a typical eCommerce company. The principle still applies at a smaller scale. A merchant with 500 SKUs across three sales channels has different needs from a merchant with 20 products on one storefront. The system should be tested against the business that will actually use it, not only a simplified example.
Our guide to choosing inventory management software for eCommerce explains the operational questions businesses should consider before selecting a system.
Lesson 2: Do Not Treat Integration as an Afterthought
Inventory information often exists across several platforms. The storefront records sales. A purchasing system tracks supplier orders. The warehouse records receipts and movements. Accounting software tracks inventory value and cost. Shipping activity confirms what left the building. If those systems do not exchange information reliably, employees may have to reconcile reports or enter the same data more than once. That creates more opportunities for delays and errors.
Before implementation, businesses should understand:
- Which system owns each type of data
- How often information will update
- What happens when an update fails
- How duplicate or conflicting records are handled
- Which workflows still require manual action
- Who is responsible for resolving exceptions
Ordoro’s partners and integrations connect eCommerce sales channels and business systems with the inventory, order, and fulfillment workflows around them.
Lesson 3: Be Careful With Customization
Businesses often have legitimate reasons to customize software. Their products, suppliers, approval processes, or fulfillment requirements may not fit a completely standard workflow. But heavy customization can make an implementation harder to test, maintain, and update.
Every custom rule introduces another decision the system must process correctly. Custom connections may also behave differently when the software or another connected platform changes. Before requesting customization, determine whether the business requirement is truly essential or whether an existing workflow can accomplish the same goal.
The objective should not be to recreate every historical process inside a new platform. A software change can also be an opportunity to simplify procedures that have become unnecessarily complicated.
Lesson 4: Roll Out Major Changes in Manageable Stages
A phased implementation allows a business to test workflows, identify problems, and train employees before expanding the system across the entire operation. A merchant might begin with one warehouse, product category, sales channel, or inventory process. Once the data and workflow are performing reliably, the business can add more complexity. A staged rollout may take longer than switching everything at once, but it limits the amount of inventory and customer activity exposed to an undiscovered problem.
Businesses should also maintain a clear transition plan. Employees need to know when the old system stops being authoritative, when the new system becomes the source of truth, and how activity completed during the changeover will be reconciled.
Lesson 5: Forecasts Need Reliable Inventory Data
Forecasting tools use historical information and current signals to estimate future demand. Those recommendations become less dependable when the underlying records contain inaccurate receipts, unexplained adjustments, duplicated products, inconsistent SKUs, or stockouts that were not documented correctly.
For example, a product may appear to have weak demand because it was unavailable for several weeks. A forecasting system looking only at recorded sales may recommend ordering less, even though the real problem was that customers could not buy it.
Before relying heavily on forecasting or automated replenishment, businesses should establish dependable procedures for:
- Receiving inventory
- Recording returns and damage
- Managing product and SKU data
- Counting physical stock
- Investigating discrepancies
- Tracking supplier lead times
Our inventory management best practices guide covers the operational foundation needed to support reliable inventory decisions.
Lesson 6: Automated Recommendations Still Need Human Review
Software can analyze more information than an employee could review manually. It can identify sales patterns, calculate reorder points, and flag unusual changes. But a recommendation is not the same as a decision.
An employee may know that a supplier is discontinuing a product, a marketing campaign has been canceled, a large wholesale order is coming, or a recent sales increase was caused by a temporary promotion. Human review gives those recommendations business context.
Businesses should define which decisions can happen automatically and which require approval. A routine quantity update may be safe to automate, while an unusually large purchase recommendation may need review before money is committed. The same principle applies to exception alerts. Software can flag unusual activity, but someone must investigate what happened and determine the appropriate response.
Lesson 7: Train Employees on the New Workflow
A technically successful implementation can still fail if employees do not understand how their responsibilities have changed. Training should explain more than where to click. Employees need to know:
- Which system should be used for each task
- How inventory quantities change
- How to handle exceptions
- When manual changes are appropriate
- Who can approve adjustments
- How to report a system or data problem
Employees should also understand how their work affects other teams. Receiving the wrong quantity can distort purchasing recommendations. An undocumented return can affect available stock. A manual workaround may prevent another system from receiving the correct update.
Our guide to inventory management training for eCommerce teams explains how to align training with receiving, purchasing, warehouse, customer service, and management roles.
Lesson 8: Monitor the Results After Launch
Going live is not the end of an inventory software implementation. Businesses should monitor whether the new system is producing the expected operational improvements. Useful indicators may include:
- Stockouts and overselling
- Excess or slow-moving inventory
- Receiving discrepancies
- Unexplained adjustments
- Forecast accuracy
- Order delays
- Manual workarounds
- Supplier and warehouse exceptions
Unexpected changes should be investigated early. A growing number of emergency purchase orders, repeated inventory corrections, or employees returning to spreadsheets may indicate that part of the workflow is not working as intended.
Our article on inventory inefficiency warning signs covers several patterns that may signal a process or system needs attention.
What Smaller eCommerce Businesses Can Learn From Nike
Most eCommerce companies will never manage Nike’s product volume or global supply chain. They can still make the same types of mistakes on a smaller scale. nA merchant may connect a new storefront without confirming how SKUs match. It may import inaccurate starting quantities, automate purchasing before supplier lead times are correct, or give employees conflicting instructions during a system change.
The financial impact may not reach $100 million, but the operational consequences can still be serious. The most important lessons are practical:
- Choose software based on the real workflow.
- Confirm that product and inventory data are accurate.
- Test integrations and exceptions before launch.
- Introduce major changes in manageable stages.
- Train employees on the complete process.
- Review automated recommendations before acting on them.
- Monitor results after implementation.
Technology works best when it supports a clearly understood inventory process.
Build Your Inventory System on a Dependable Foundation
Inventory software should connect with the way products are purchased, received, sold, and fulfilled. It should also give teams the visibility needed to review exceptions before small problems become expensive ones. Ordoro helps eCommerce businesses connect inventory, orders, purchasing, warehouses, and fulfillment in one platform. See How Ordoro Works
About the author
Jagath Narayan is the CEO and co-founder of Ordoro, the #1 ecommerce platform for retailers growing from 10 orders/day to 10,000 orders/day. Follow him on LinkedIn to learn more about Entrepreneurship and Ecommerce.
