Inventory Shrinkage in Multi-Site Consigned Stock: The Five Signals We Look For

Shrinkage in consigned inventory is harder to see than shrinkage in owned inventory because the manufacturer never physically holds the unit. The signals are in the data, not on the shelf.

What shrinkage actually costs

The National Retail Federation puts general shrinkage at 1.6 percent of revenue. In consigned medical device and industrial networks, we routinely see the number run two to four times higher because consumption reporting is delayed and physical counts happen infrequently.

For a manufacturer with $500M in consigned revenue, a shrinkage rate of 4 percent is $20M in annual loss. That number is almost never carried on the P&L as shrinkage. It hides inside COGS, write-offs, and cycle-count adjustments.

The five signals we look for in a first assessment

1. Location-level stock-out variance

The first signal is direct. If ten locations serve similar customer profiles and three of them stock out five times as often as the others, the stock is either being consumed and not reported, walking off site, or being transferred without a record.

2. Product-line billing discrepancies

The second signal compares consumption records against distributor billing by product line. A healthy product line shows a tight ratio between units consumed and units billed. A product line with a widening ratio over three or more quarters is leaking.

3. Transfer-record decay

Transfer records decay when field reps move stock between locations without logging the movement. The signal is a rising count of negative on-hand corrections at the destination location paired with unexplained positive corrections at the source.

4. Cycle-count miss patterns

Not all cycle-count misses are equal. Random misses suggest a counting error. Clustered misses on the same SKU family, same location, or same rep suggest a systemic issue that deserves investigation.

5. Freight-in vs freight-out imbalance

The fifth signal is often the most quietly damaging. When freight-in to a region exceeds freight-out plus recorded consumption over a rolling six-month window, product is arriving and disappearing.

How the assessment leads to a defensible business case

The five signals are not opinions. They are calculated from consumption, billing, and transfer records that the manufacturer already owns. The assessment output is a ranked list of locations and product lines by estimated recoverable dollars, with a confidence band from Monte Carlo simulation.

Sponsors ask a version of this question in every steering committee:

"You are about to fund a data platform. Does anyone know if it works?"

The five-signal assessment answers that question before the platform is funded. If the signals do not add up to a defensible business case, the project should not proceed.

Tying back to a real engagement

For the US orthopedic bracing and mobility device manufacturer described in our inventory optimization case study, the five-signal assessment surfaced $19M in annual loss across chargebacks, shrinkage, and dead stock. The department in question was carrying 20 percent of total company revenue. Within six months of standing up the reconciliation platform and running Monte Carlo placement, $4M was returned to profit and the shrinkage rate on the pilot region dropped to under 2 percent.

Frequently asked questions

Is consigned inventory shrinkage covered by insurance?

Usually only if it can be traced to a specific loss event. Slow, distributed shrinkage from reporting decay is almost never covered.

How long does a five-signal assessment take?

Four to six weeks for a multi-site network, assuming access to consumption, billing, and transfer extracts.

Do we need to install anything at customer sites?

No. The assessment runs entirely on data the manufacturer already generates. Field process changes come later, and only where the data justifies them.

What is a realistic shrinkage recovery target?

For a network that has never been assessed, 40 to 60 percent of identified shrinkage is typically recoverable within twelve months. The remaining loss requires physical process change.

How DATA4AI helps: We run the five-signal assessment, build the reconciled data platform, and quantify the recoverable pool before any long-cycle infrastructure spend. See our manufacturing practice and our service offerings, or book a discovery call.