How to read a pre-shipment inspection report

An inspection report usually arrives within a day of the visit, and the supplier is waiting for your answer. Reports from different companies are laid out differently, but most carry the same parts. This is a way to read them in order. It describes common practice; your own inspection instructions decide what the report must contain.

1. The overall result

Most reports open with a result such as pass, fail or pending. Read it last. The result is the inspection company's summary against the criteria it was given. If those criteria were incomplete, a pass means less than it appears to.

A pending result commonly means the inspector found something the instructions did not cover and is leaving the decision to you.

2. What was inspected

Check the order number, the product, the factory address and the date. Then check the quantities: how many units were ordered, how many were finished, and how many were packed when the inspector arrived. A pre-shipment inspection of a half-finished order is a sample of the half that existed.

3. Sampling

The report should state the lot size, the sampling plan and the number of units pulled. With an AQL plan it will also give, for each class of defect, the number of defects at which the lot is accepted and the number at which it is rejected. You can check those figures for your lot size with the AQL calculator at aqlchart.com.

Look at how the sample was chosen. Cartons should have been picked by the inspector from across the lot, not presented by the factory.

4. The defect table

This is the core of the report. Defects are counted by class:

• Critical — could harm the user or breaks a regulation.
• Major — likely to cause a return or a complaint.
• Minor — a departure from the specification that most users would accept.

Compare the count in each class with the accept and reject numbers. Then read the descriptions. Ten minor defects of ten kinds are a different situation from ten units with the same fault, which points to one cause in production.

Check that the classes used are the ones you defined. If you gave no defect list, the inspector applied the company's default one.

5. Checks and tests on site

Measurements, weight, function tests, barcode scans, carton drop tests: each should show what was done, on how many units, and the result. A line that reads "not performed" needs a reason — no equipment at the factory is a common one.

On-site tests are not laboratory tests. A report cannot tell you that a product meets a safety regulation; that needs a test report from a laboratory.

6. Packing and labelling

Labels, warnings, barcodes, carton marks and packing method are compared with your packing specification. Errors here are cheap to fix before shipment and expensive after.

7. Photographs

Photos should show the lot as found, the sampling, each defect type, the tests, the labels and the cartons. Use them to check the descriptions, and look for what is missing: no photo of the full lot may mean the goods were not all there.

8. Remarks

Read the inspector's remarks closely. This is where things outside the criteria are recorded: goods not fully packed, a production line different from the one audited, cartons already sealed on arrival.

Deciding

There are three usual decisions: ship, rework and re-inspect, or reject. The report supports the decision; it does not make it. If you ask for rework, say which defects, on how many units, and whether a second inspection follows.

Questions

Can I accept a lot that failed?

Yes. The decision is the buyer's. Some buyers accept a failed lot with a price adjustment or after sorting; record what was agreed in writing.

Does a pass mean every unit is good?

No. The inspector checked a sample. A pass means the sample met the acceptance numbers of the sampling plan.

What if the report has no defect classification?

Ask for it. Without classes and acceptance numbers the result cannot be tied to anything you agreed with the supplier.

---

Sources: this article describes common practice in inspection reporting and cites no external data.