Inventory is the working-capital line that grows without a meeting. There’s no quarterly review where someone proposes increasing it. There’s no executive sign-off when a site manager doubles a reorder threshold to feel safe. It just compounds — quietly, across plants and yards and warehouses, until a CFO runs the cash-conversion-cycle math one quarter and realizes the operating line is propping up something that shouldn’t need propping.
The math is the easy part. The reason it stays hidden is the system that’s supposed to track it.
The math nobody runs out loud
The inventory line on a typical mid-market manufacturer’s balance sheet is 60–90 days of working capital. In construction, materials inventory plus equipment-in-transit can run 45–75 days. In rail and ground transport, spare parts inventory across yards is 90–120 days, and you can’t reduce it below a certain floor without risking a service event. In hospitality, FF&E spares and consumables together run 30–60 days.
Across all of these, the recoverable portion — inventory you could release without disrupting operations — is typically 30–50% of the carrying total. That’s three to six weeks of working capital sitting in over-stocks, duplicates, dead stock, and reorder thresholds set higher than the data justifies. At a 200-million-revenue mid-market company, that translates to $5–15M of cash that could go to the credit line, the dividend, or the next acquisition.
Nobody runs that math out loud because the system that holds the inventory data isn’t capable of producing the answer. The math requires:
- One queryable balance per part, across every location
- Movement history per part, with seasonality removed
- Reorder rules visible alongside actual reorder behavior
- The ability to see, by SKU, where the over-stock concentrates
Spreadsheets and most ERP inventory modules can produce one of these. Producing all four in the same query, on the same day, against the same data — that’s where the systems break, and where the working capital hides.
Why it grows invisibly
Inventory grows for understandable reasons. Each one is reasonable on its own. The compounding is what does the damage.
Site managers set reorder thresholds defensively. A stock-out at site 7 last quarter triggered a customer issue. The site manager raised the reorder point on every related part to make sure it doesn’t happen again. The data didn’t justify it; the memory of the stock-out did. There’s no central review of reorder thresholds, so the change persists.
Cross-site duplication is invisible. Site 3 over-orders a part. Site 9 is short of the same part. They don’t talk because the system doesn’t show them each other’s balances. Site 9 orders fresh from the supplier instead of asking site 3 to ship internally. Both balances grow.
Dead stock accumulates because nobody owns the write-down. Parts that supported equipment generations now retired sit on the shelf indefinitely. Each one is a small carrying cost. Multiplied across thousands of SKUs across plants, the cumulative cost is real. But identifying them requires running utilization queries against the inventory data the operations team owns, against the equipment data the maintenance team owns. Those datasets don’t talk.
ERP inventory modules report by location, not by network. A controller asks “how much of part 8841-A do we have, network-wide?” — and the answer requires running the same report across six plants, exporting to Excel, summing manually. The query that should take seconds takes a half-day, and a half-day means the question doesn’t get asked weekly.
Five questions a COO should be able to answer in one query
These are the questions that diagnose where the working capital is hiding. None of them should take more than a few seconds to answer. Most ERP inventory modules can’t answer any of them.
- By SKU, what’s the network-wide on-hand balance, broken down by location? If this requires aggregating across multiple reports, the over-stock and the under-stock at different sites are invisible.
- Which parts have moved less than once in the last 12 months? Dead stock candidates. The list is usually 5–15% of total SKUs and 1–3% of carrying value, and writing it off releases capital and shelf space.
- For each reorder rule in effect, what was the actual demand pattern that produced it? Most reorder thresholds were set once, defensively, and never revisited. Comparing the rule to the demand pattern surfaces the over-conservative rules that quietly inflate inventory.
- What’s the gap between what the inventory ledger says and what the most recent cycle count found? If the gap is large or growing, the data driving every other working-capital decision is wrong, and every reorder decision compounds the error.
- Of the inventory currently on hand, what percentage was ordered against an open work order or production run? If the percentage is low, you’re stocking against the calendar, not against actual operational need.
What changes when the answers are one query away
The platform shift required to answer these questions in seconds rather than days isn’t about a better inventory module. It’s about the inventory data living in the same operating picture as the asset register, the maintenance schedule, and the supplier graph — instead of in a parallel system that has to be reconciled against everything else weekly.
When an inventory ledger is append-only and immutable, the gap between book and physical surfaces in real time, not at quarter-end. When reorder rules sit with the asset and the location they support, they get reviewed when the asset’s utilization changes — not in a quarterly batch. When part movements across the network are queryable in one shot, the cross-site duplication that hides the working capital becomes visible.
The recoverable working capital doesn’t move because of a procurement initiative or a new metric on the COO’s dashboard. It moves because the data that hid it stops being able to.