Returns inspection means confirming, item by item, what came back from the store and in what condition: the right product, intact packaging and enough shelf life left on the best-before date. With vision AI on a wearable, the worker who is already handling the product can identify it, read the date and check for damage in one motion, and the result goes to SAP under a rule you define.
Below we look at why store returns are so hard to check consistently, what manual and fixed-station approaches miss, and how returned product inspection with TIM INSPECT runs the check on AXION in the worker's hand. We finish with the practical points to settle before a pilot.
Why store returns are hard to inspect
In fast-moving consumer goods and retail distribution, returns rarely arrive in a neat, uniform flow. A single return receipt can include different products, single items, multipacks and full cartons. Some packs are crushed, some are close to their date and some have lost or hidden their barcode. Each of these needs a decision: back to stock, write-off, return to supplier or another route defined by your business rules.
Two things make that decision expensive. First, it takes time: reading a printed date, turning a pack to find the barcode and judging whether a dent is acceptable all happen by hand. Second, it varies: the same scratch or deformation is judged differently from one person to the next and from one shift to the next. Inconsistent decisions show up later as disputes with stores or suppliers and as stock that does not match what is on the shelf.
Traditional returns checks and their limits
- Visual check and paper or terminal entry: the date is read by eye and damage is noted according to personal judgement. It is slow, and the result depends on who is doing it.
- Barcode-only processing: scanning confirms which product came back, but not its condition. Products without a readable barcode stop the flow.
- Fixed camera station: a fixed camera is good at identifying and counting products. But a barcode can end up underneath, a date can sit inside a fold, packs lie on top of each other and cartons arrive closed. A camera that looks from one angle cannot turn the product over.
The common gap is condition. Identifying a return is the easier half; checking damage and date reliably, the same way every time, is where most processes break down.
How vision AI checks returns in the worker's hand
In the FMCG deployment described on the solution page, the check runs on AXION, a wearable computer, worn by the worker who receives the return. TIM INSPECT provides the image analysis and the link to SAP. The flow has four steps.
1. Scan the return receipt
AXION reads the number on the receipt. The return record opens in SAP and the product list appears on screen with target, scanned and remaining quantities. The process can also start without a receipt.
2. Identify the product, with or without a barcode
If a barcode is visible, it is read. If not, the product is recognised from its packaging, logo, colour and printed text. For similar products, the distinguishing detail such as flavour, weight or volume has to be visible. When the system is not sure, it does not guess; it asks to see the product again. A photo is added to the record as evidence.
3. Check the date and the packaging
The worker turns the product to show the right face. AXION reads the best-before date, calculates the days left and looks for the damage types you have defined, such as burst packs, deformation or scratches.
4. Apply your rule and complete the record
If there is damage or a date that is too close, AXION alerts the worker on screen and by voice, and your rule decides the outcome. A line does not close without confirmation. When no lines are open, product, quantity, date and damage status go to SAP with the confirmed overall result.
Because one return number links the receipt, the product images, any corrections and the confirmation, missing, extra or wrong products become visible while the return is being received rather than during a later reconciliation. The solution page reports a decision time under one second, 100% accuracy in returns and 100% customer satisfaction for this deployment, based on the Thread In Motion project report.
What to plan for before a returns inspection pilot
Write down your damage and date rules
The system applies your rule, so the rule needs to exist in a usable form. Decide which damage types count as a return reason, how many days before the best-before date a product is treated differently and what happens in each case. Packaged goods typically need definitions for burst packs and deformation; other products may need surface damage such as scratches.
Collect product images for identification
Visual identification learns from product photos. Your own trained users can add photos when new products are introduced, and model updates go live after approval, so the catalogue does not depend on an outside team each time the range changes.
Decide where a fixed camera is enough
Not every point in the returns area needs a person turning the product. Where identification and counting are enough, a fixed camera can do the job; AXION covers the condition checks. The two can work together in the same flow.
Lighting and the work surface
Date printing is small and often low-contrast. Good, even lighting at the receiving table and a clear surface to hold products against make both reading and damage detection more dependable. If you also want to document damage on cartons and pallets as they arrive, damage detection covers that step.
SAP integration
Agree which SAP objects hold the return record and the expected products, and which fields the result should update. The worker completes the return in one flow while the record runs in the background.
FAQ
What is returned product inspection?
It is the check of store returns for product identity, packaging damage and best-before date. With TIM INSPECT, the check runs on AXION and the confirmed result goes to SAP with the return record.
How is a product without a barcode identified?
From the look of its packaging: logo, colour and printed text. If the system is not sure, it asks to see the product again instead of making up a result.
Who decides what counts as damage?
You do. You define the damage types, the date thresholds and the outcome for each case, and TIM INSPECT applies the same rule to every product.
Would a fixed camera not be enough?
For identification and counting, yes. For condition checks, a hand that turns the product is needed, because barcodes, dates and damage are not always on the face a camera sees. That is why condition checks run on AXION.
Next step
To review your product groups, damage definitions and date rules, you can request a demo and go through the returned product inspection flow with our team.