The company had been selling automotive parts for years and had been consistently profitable. Sales continued to grow and, on the surface, the business appeared to be doing well.
But something was increasingly worrying its manager. Whenever it was time to pay suppliers, payroll, and other obligations, the cash simply was not there. Month after month, the company had to rely on bank credit lines to cover short-term needs.
The company was selling a lot. It was profitable. But its debt kept growing.
How could it be selling more than ever and, at the same time, have less and less cash?
A consultant friend sat down with him to review the business. After analyzing the information, an answer began to emerge—one that was not hidden in sales, or necessarily at the bank. It was several yards farther back: in the warehouses.
There were vehicle accessories that had gone unsold for months. Parts imported from Japan had been purchased in anticipation of demand that never materialized. Hundreds of SKUs were taking up space and consuming capital without moving.
But an even more troubling contradiction emerged. While a significant portion of inventory remained immobilized, some of the highest-demand products were experiencing constant stockouts. Customers were ready to buy, but the right product was not available.
The company had, at the same time, too much inventory and too little inventory.
And meanwhile, it continued financing itself through the banks.
The Cost We Do Not See
Immobilized inventory costs far more than its purchase price. It consumes working capital, indirectly generates financing costs, occupies warehouse space, requires handling and control, creates insurance costs, and increases the risk of deterioration, obsolescence, or expiration.
But the opposite extreme is costly as well. Not having the right product available means lost sales, dissatisfied customers, and eventually lost market share.
That is why the objective should not simply be to reduce inventory. The objective is to have the right inventory, in the right quantity, at the right time.
This Is Where Data Science Changes the Conversation
Today, each SKU can be analyzed using variables such as historical demand, variability, margin, turnover, days without movement, inventory levels, stockout frequency, replenishment lead times, and supplier behavior.
Machine Learning can help forecast demand, while techniques such as clustering can uncover groups of products with similar behaviors that often remain hidden among thousands of SKUs.
We might find high-turnover, high-margin products with insufficient inventory; expensive items with very low turnover; virtually dead stock; products with erratic demand; and healthy SKUs that simply need to be maintained.
We then stop asking, “How much inventory should we reduce?” and begin asking, “What should we reduce, what should we protect, and where do we actually need to increase inventory?”
That difference can transform working capital.
From Data to Decisions
The analysis does not end with identifying excess inventory. It should help us decide what to buy, how much to buy, when to buy it, which products to liquidate or discontinue, and, especially, why we accumulated inventory we did not need in the first place.
Perhaps the problem lies in forecasting. Perhaps in minimum order quantities. In replenishment lead times. In purchasing policies. Or simply in historical decisions that no one ever questioned again.
Many of these questions can also be answered through traditional analysis. In fact, we have done it that way for years. The difference is not only whether we can perform the analysis, but how long it takes and how often we can repeat it.
Manually reviewing thousands of SKUs—studying their turnover, demand behavior, margins, inventory levels, and obsolescence risk—can require many hours, days, or even weeks. And by the time we finish, the inventory has already started changing again.
Technology changes that equation. A properly designed model can repeat in minutes an analysis that once required many hours of work and update it every time new information becomes available. We can even go one step further and create an intelligent agent that continuously monitors inventory, detects deviations, anticipates potential stockouts, identifies unnecessary accumulation, and generates alerts when an SKU begins to move away from the levels we consider healthy.
In this way, data is no longer used only to correct inventory once the problem already exists; it begins helping us prevent the problem from happening again. The goal is not to replace the knowledge of the people who manage the business, but to enable that knowledge to be applied continuously across thousands of SKUs, with a speed and consistency that would be extremely difficult to achieve manually.
Once the model is stabilized, the impact can go far beyond freeing up cash: less obsolescence, fewer stockouts, better purchasing decisions, higher turnover, and a significant reduction in the warehouse space required. In some cases, it may even make it possible to redesign the physical distribution or eliminate facilities that are no longer necessary.
The company in our example may not have needed more financing. It needed to recover the money it already had trapped on its shelves.
Look at Your Warehouses Before Calling the Bank
How much of your inventory has not moved in 6, 12, or 24 months? How much capital is trapped there? How many sales are you simultaneously losing because of stockouts?
Do not wait for financial debt to give you the answer.
Measure the excess. Identify the critical products. Quantify the trapped cash. Find the causes and act.
“Because every dollar of inventory should have a reason for being there. And if it does not, it is time to ask what that money could be doing elsewhere in your company.”