Inventory turns and working capital: what the measure leaves out

01 / FIELD NOTE

Keep the decision tied to the operating context.

Days of inventory, or days sales of inventory, is calculated as average inventory divided by cost of goods sold, multiplied by the number of days in the period. It answers a simple question: at the current rate of sale, how long would the stock on hand last. A lower figure means capital is committed for less time, which is the point of watching it.

Because it is a ratio, it moves for two different reasons. Average inventory can fall, or cost of goods sold can rise. The first is the outcome a business wants. The second can happen for reasons that have nothing to do with inventory management — a price change, a mix shift, a good quarter — and a ratio that improves for that reason has not made the operation leaner. Reading the measure without knowing which lever moved is how a working-capital metric gets mistaken for an operational one.

The numerator is where the real constraint sits, and it is not a single decision. Average inventory is the sum of cycle stock, safety stock, in-transit stock and stock that is not really available at all. Only the last of those is pure waste, and it is also the one that a quantity-based record is worst at revealing.

Safety stock exists to cover uncertainty, and a substantial part of the uncertainty is the record itself. A business holds extra because the figure it orders against may be wrong, and the cost of that buffer is charged to a demand forecast that was never the reason for it. Improving the accuracy of the record reduces the uncertainty the buffer was sized for, which reduces the buffer — and that is a working-capital effect produced by a data change rather than a commercial one.

Stock that is physically present but not available is the most expensive category because it is invisible to the measure as usually calculated. Goods in the wrong location, goods reserved beyond their order, goods waiting on a quality decision, goods returned and never dispositioned. They count as inventory on the balance sheet while being unusable for sale, so the business is paying to hold them without the possibility of a return. Only unit-level visibility distinguishes them from stock that is genuinely available.

Excess at the other end is equally distorted. Goods that sell slowly sit in the numerator at cost, and their cost of capital, storage, insurance and eventual markdown is real. Identifying them early enough to act — to move them, promote them, or return them — depends on knowing their age and their rate of movement, which a periodic count does not provide and a frequently counted record does.

Turnover is also where the relationship between inventory and demand becomes visible. A stockout and an overstock can exist in the same business at the same time, in different categories, and an average figure conceals both. Splitting the measure by category, and reading it against an accurate count for that category, is what turns one number into a set of decisions.

Demand distortion amplifies the problem upstream. Small variations in retail demand, passed through ordering rules and lead times, grow as they move back through distribution and manufacturing, so that a modest change at the till becomes a large swing in production. The mechanism is information delay: each stage orders against its own reading of demand rather than against actual consumption. Shortening the delay by sharing real movement data is what dampens the swing, and that requires the movement data to exist and to be trusted.

Where the record is the bottleneck, the sequence matters. Reducing safety stock before the record is trustworthy converts a buffer against uncertainty into a shortage, and the shortage will be attributed to the forecasting rather than to the record. Improving visibility first, then holding the buffer down, keeps the cause and the effect in the right order.

The measures worth tracking alongside days of inventory are the ones that explain it: the proportion of stock recorded in the wrong place, the age profile of what is on hand, the share of stock that is present but unavailable, and the time between a movement and its posting. Those are leading indicators, and unlike a ratio they cannot improve without the operation actually changing.

What a business should be sceptical of is any claim that a technology reduces days of inventory by a stated proportion. The ratio is the product of purchasing policy, demand, lead times, assortment breadth and record quality, and a single intervention moves one of those. What a better record does is make the lever visible and let a decision be made deliberately, which is a different and more durable claim than a projected saving.