The retail chain's inventory manager had the same conversation every quarter. In the same meeting, the merchandising team complained about stockouts on the top 20 SKUs while the CFO complained about excess inventory consuming working capital. Both problems were real. Both were getting worse. And both, everyone agreed, were caused by inaccurate demand forecasting.
The prescribed solution was a better forecasting system. A six-figure software purchase was approved. Eighteen months later, the stockouts and excess inventory were still there.
The forecasting system was not the problem. The logic of ordering based on forecasts was.
One Root Cause, Two Symptoms
Simultaneous stockouts and excess inventory look like two different problems. The obvious conclusion is that the forecast is wrong — fix the forecast and both problems resolve. The Theory of Constraints diagnosis is more precise: the root cause is not forecast accuracy. It is forecast dependency.
Any prediction of future demand at the SKU level, over a planning horizon of weeks or months, will be wrong in ways that cannot be corrected by better statistical methods. You can improve the forecast from 60% to 70% accurate. You will still have stockouts on fast-movers and excess on slow-movers, because the error distribution is asymmetric: you tend to understock what is selling fastest and overstock what is not. Both the shortages and the excess are the predictable output of a push-based ordering system. They are not two problems. They are one problem with two symptoms.
The Pull Replenishment Model
TOC replenishment does not improve the forecast. It eliminates the need for short-cycle, SKU-level forecasting entirely.
Every stocking location maintains a target buffer for each SKU — sized to cover replenishment lead time from the upstream source, adjusted for demand variability. When actual inventory falls below two-thirds of the target buffer (yellow zone), a replenishment order is triggered. The quantity ordered equals what was actually consumed since the last replenishment — not a predicted order quantity. You replace what was sold. You do not predict what will be sold.
The result: fast-movers are replenished frequently and in proportion to actual sales. Slow-movers are replenished only when inventory actually depletes. Both the stockout and the excess problem resolve from the same mechanism change.
The Evidence
SportZone, a sports retail chain operating in Portugal and Spain, implemented TOC pull replenishment over 18 months. Their results, presented at TOCICO 2018: inventory turns improved from 2.8× to 4.9×. Stockout rate reduced 62%. Gross margin improved 3.2 percentage points, primarily from fewer markdowns on excess inventory.
Dr. Reddy's Laboratories ran the same model across a global pharmaceutical supply chain: two central warehouses, 32 regional warehouses, and more than 2,000 distributors. The system ran daily replenishment calculations on actual consumption, automatically generating orders via SAP when buffer levels penetrated yellow or red zones. The result: simultaneous reduction of excess inventory and improvement in product availability.
The Mabin and Balderstone meta-analysis of 82 TOC implementations found median inventory reductions of 49%. For a company carrying $20 million in inventory, that is nearly $10 million in freed working capital — with no revenue sacrifice required. This is the number that turns the conversation from an operations memo into a CFO priority.
Where Inventory Should Live
Pull replenishment inverts the traditional logic of where to hold inventory. Push models build inventory downstream — at stores or regional warehouses closest to customers. TOC logic holds the buffer upstream (central warehouse) and sends small, frequent replenishments downstream as consumption occurs.
Upstream concentration works because it aggregates demand. A central buffer serves many downstream locations simultaneously, absorbing variability across the entire demand pool rather than each downstream location protecting itself independently with its own safety stock. Total system inventory falls while availability improves. These are not competing outcomes — they are the same mechanism producing two results.
The Competitive Opening
When Panasonic implemented pull replenishment with selected distributors while competitors continued push-based ordering, the distributors began actively preferring Panasonic — not because of price or product features, but because of inventory economics. Lower tied-up capital. Fewer stockouts. Higher turns. A commodity product became a differentiated offering without any product changes.
A manufacturer who commits to maintaining a distributor's buffer — through vendor-managed inventory or a formal pull replenishment agreement — is offering something no push-based competitor can match. The offer: your inventory investment decreases, your availability improves, and you never need to place an order. Competitors competing on price cannot replicate this without changing their entire supply chain model. This is the basis of a decisive competitive edge in distribution.
What Changes in the Organization
The operational change is technically straightforward: smaller, more frequent replenishments triggered by actual consumption. The organizational change takes longer. Purchasing teams measured on order accuracy and inventory levels need new metrics. Store managers who have been forecasting and ordering need a different mandate.
Forecasting does not disappear. Long-range aggregate demand planning for capacity and capital decisions remains essential. What is replaced is the short-cycle, SKU-level forecasting that drives ordering decisions — and the errors that forecasting inevitably introduces. Framing the change this way reduces resistance from planning teams who reasonably fear their function is being eliminated.
This Week's Action
Pull your inventory data for the top 50 SKUs by revenue. For each one, calculate weeks of supply on hand and count stockouts in the last 90 days. If you find any SKU where you currently hold more than six weeks of supply and have still experienced at least one stockout, you are looking at the diagnostic signature of push-based ordering. That data is the starting point for the conversation about pull replenishment — and for calculating what the working capital release could be worth.