A picking error is usually detected outside the warehouse, when the customer opens the box. By then it has produced a reshipment, a credit memo, a customer service call and an inventory discrepancy, because the goods shipped in excess or short no longer match the record. In e-commerce the impact is greater: the 2026 Online Sales Study by AMVO, Mexico's online sales association, reports that receiving incomplete or damaged products is the top complaint of online shoppers in Mexico. The article on the cost of inventory errors covers what that costs; this one covers how to reduce it.
First step: classify the errors
The usual response to picking errors is a meeting that asks for more attention, and its effect is temporary. A lasting result requires identifying the type of error, because each type has a different cause.
| Error type | What the customer sees | Most frequent cause |
|---|---|---|
| Wrong item | A different model, color or size arrived | Look-alike products side by side, typed codes, unrecorded locations |
| Wrong quantity | 12 arrived instead of 1, or 1 instead of 12 | Unclear unit of measure (case versus each) |
| Wrong lot | A lot with shorter shelf life, or a held lot, arrived | Lot chosen by the person rather than the system |
| Short shipment | Lines or units are missing | A short pick that was neither recorded nor reallocated |
| Wrong package | The whole order went to another customer | Shipping labels swapped at packing or loading |
For two to four weeks, log every claim with its type, the product, the location it came from and the shift. With 50 to 100 cases, a pattern emerges: a small number of products and locations usually accounts for most of them.
Recurring causes
Look-alike products next to each other
Two sizes of the same product, a 500 ml and a 1 liter, in adjacent locations encourage errors, especially when the cartons share a color. The physical fix is to separate them: different aisles or levels, alternating with products that look nothing alike. Placing each product by velocity and by similarity to others is part of slotting.
Manual code entry
When a product code is typed instead of scanned, the system records what was typed, not what was picked. If a product has no readable barcode, the problem is in receiving or labeling, not in picking. Label anything that arrives without a code at receiving.
Case versus each
The most common cause of wrong quantities is an ambiguous unit: the order says "12" without specifying whether it means 12 eaches or 12 cases of 12. Define one unit of measure per product in the catalog and, if it sells in several, the conversion factor. The GS1 General Specifications assign a different GTIN to each packaging level (each, inner pack, case) precisely so that a scan can tell a case from an each. On the picker's screen, quantity should always carry its unit: "Pick 6 ea", never just "6".
Lot chosen by the picker
In food, beverages, cosmetics or chemicals, the issue rule (FIFO or FEFO, first expired, first out) cannot depend on the picker checking dates on each case. When the system allocates the lot and the handheld validates it, the right lot ships even if a different one sits at the front of the rack.
Unrecorded short picks
When a picker cannot find the full quantity and closes the task as if it were complete, the error has two effects: the customer receives less, and the system keeps stock that is not in the location. Recording a short pick should be easier than omitting it; this is a process design decision, not a matter of discipline.
Effective controls
- Scan the location and the product on every pick. It prevents more errors than any other control because it compares what was taken with what was ordered, at that moment.
- Show quantity with its unit. For products sold by the case and by the each, use different barcodes for each pack level.
- Allocate the lot from the system, and validate the scanned lot.
- Record short picks on the spot, with reallocation to another location if stock exists.
- Verify at packing. Scanning each item into the package stops a picking error before it ships. For high-value orders, add package weight as a second check.
- Count locations with recurring errors more often. A wrong location record produces picking errors even when the picker follows the process correctly.
How to measure picking accuracy
Picking accuracy is the share of picked lines without error over total picked lines. Measure it by line, not by order: a 40-line order with one error and a one-line order with one error do not weigh the same. Separate errors caught at packing (internal) from errors reported by customers (external). The first measures the effectiveness of verification; the second, the errors that reach the customer.
How Chainlink handles it
In Chainlink WMS, guided picking walks the picker through each pick in route order. The app shows the location, product, lot and quantity and asks the picker to scan the location and the product to confirm. If a different location is scanned, it reports a wrong location and points to the correct one; with strict FEFO, if a lot other than the allocated one is scanned, it asks for the lot that expires first. If the location holds less than expected, the picker records the short pick and Chainlink tries to reallocate it to another location with stock. If there is none, the short pick stays pending until a user with an Administrative license approves it.
At packing, each item is scanned into the package and Chainlink accepts only what was picked for the order, never more than was picked. Each package is weighed when it is closed. In the Professional and Enterprise plans, advanced analytics show picking accuracy alongside inventory accuracy.
Frequently asked questions
What is the most common cause of picking errors?
It depends on the warehouse, but two causes show up almost everywhere: look-alike products in neighboring locations and case-versus-each confusion. Classifying claims by type for a few weeks shows which one predominates in your operation.
Is a packing check sufficient?
A packing check stops errors, but it detects them late: the picker must return to the rack and the order is delayed. Validation at picking prevents the error; verification at packing stops the one not detected earlier.
How is picking accuracy calculated?
Picked lines without error divided by total picked lines in a period. Keep errors caught in the warehouse separate from errors reported by customers.
Share this article: LinkedIn
