The system says you have twelve. The shelf has none.
Phantom inventory is stock your records say exists and your shelf does not have. It blocks replenishment silently, and it is one of the most expensive data problems in retail because every system downstream trusts the wrong number.
Replenishment logic reorders when recorded stock falls below a threshold. If the record is wrong and reads higher than reality, the reorder never triggers. The shelf stays empty, sales stay at zero, and because zero sales looks like weak demand, the system may reduce the order further. The error protects itself.
Every phantom unit was created by a real event
Phantom inventory is not random data corruption. It is the accumulated residue of ordinary operational events that were never recorded.
Theft not yet counted
Stock leaves without a transaction. The record keeps it until a count finds the discrepancy, often months later.
Damage and disposal
Product broken, spoiled or written off in practice but never written off in the system.
Receiving errors
A delivery booked in full that arrived short, which creates units that never existed on site at all.
Misplacement
Stock physically present but in the wrong location, on the wrong fixture or stranded in the back room. Present in the building, absent from the shelf.
Scanning errors
The wrong item scanned at the till depletes one product’s record and leaves another overstated. One mistake creates two errors.
Returns handling
Items returned to stock that were never resaleable, or returned to the wrong code entirely.
The error is self-concealing
Most data errors get noticed because something downstream breaks loudly. This one does the opposite: it produces a plausible-looking result.
| Step | What happens | Why nobody catches it |
|---|---|---|
| 1 | Record overstates stock on hand. | Nothing has visibly failed yet. |
| 2 | Reorder point is never reached, so no order is raised. | The system is behaving exactly as configured. |
| 3 | Shelf empties and sales fall to zero. | Zero sales reads as weak demand, not as absence. |
| 4 | Forecast lowers future orders for the line. | The forecast is responding correctly to the data it has. |
| 5 | Line is reviewed for delisting on poor performance. | Every number supports the decision. All of them are downstream of one wrong figure. |
By the time anyone investigates, the evidence points at the product rather than the record. This is how healthy lines get delisted, and why phantom inventory costs more than the units it misplaces.
Comparing what the record claims against what the shelf shows
Observe the shelf continuously
Vision on your existing cameras establishes whether a facing is stocked, gapped or empty through trading hours, without anyone walking the aisle.
Compare against recorded stock
The signal is the contradiction: the record says units are available while the shelf has been empty for hours. That combination is the definition of phantom inventory.
Raise it as an exception
A specific product at a specific location with a specific discrepancy, which someone can verify in minutes rather than a full stock take.
Correct the record and act on the pattern
One correction fixes a line. The repeat pattern across products and locations tells you whether the underlying cause is shrinkage, receiving or process, which is the fix worth having.
Three losses from one wrong number
Sales that never happen
The direct loss, and the invisible one. An empty shelf generates no transaction to appear in any report.
Decisions made on false data
Forecasts, range reviews and supplier conversations built on sales figures that measured availability rather than demand.
Effort spent counting
Blanket stock takes exist largely because nobody trusts the numbers. Targeted exceptions reduce how much of that is necessary.
Phantom inventory FAQs
Will more frequent stock counts solve this?
They reduce it and are expensive to sustain. Counting tells you the record was wrong at the moment of counting, and drift restarts immediately afterwards. Continuous observation catches the discrepancy as it appears, and lets you count where there is reason to rather than everywhere on a cycle.
Is this the same as shrinkage?
Related but not identical. Shrinkage is stock lost. Phantom inventory is stock the record still believes in, whatever the reason. Theft creates it, but so do receiving errors and misplacement, which are not losses at all.
Do we need new cameras?
Usually not. A-Eye is built to run on existing CCTV. Whether current coverage is usable for shelf work depends on angles and framing, which we assess per site rather than assume.
Does this integrate with Odoo?
Yes. Discrepancies raise as exceptions against the relevant product and location, so corrections happen in your system of record rather than in a separate tool.
What about stock in the back room?
A shelf camera sees the shelf. If the record is right and the units are in the stockroom, that is a replenishment problem rather than a data problem, and it is worth distinguishing because the fix is completely different.
How quickly would we see anything useful?
Discrepancy patterns tend to surface within the first weeks, because the same products and locations recur. Whether the underlying cause is fixed that quickly depends on what it turns out to be.
Find out how much of your stock exists only on paper
If your shelves empty while the system reports stock available, that gap is measurable. We will look at your current camera coverage and stock data and tell you what can be compared today.