How to Prepare for a Final Random Inspection | Specifications, AQL Setup, Packaging Files

For a Final Random Inspection (FRI), have the order 100% produced and at least 80% packed. Send one final set of files: the PO and SKU list, product specification, approved sample details, AQL and defect rules, test instructions, packaging artwork, barcode data, and carton requirements. For example, a 4,000-unit lot at General Inspection Level II gives a 200-unit main sample in the UTS AQL table. If barcode verification is included, every checked barcode must scan successfully.

100% Produced, 80% Packed

Book the Final Random Inspection when production is finished and at least 80% of the goods are packed.

Check these points before the inspection date:

100% of the ordered goods have been produced;
at least 80% are already packed;
every SKU, color, size, and version is available;
finished goods are at the inspection location;
packed cartons can be opened for random sampling;
approved samples and reference files are available if required;
equipment needed for agreed on-site tests is ready.

A 10,000-piece order is not ready just because 10,000 pieces appear on the production record. If 1,500 pieces are still being assembled, or one SKU is stored at another location, the finished lot available for inspection is incomplete.

The same problem happens with packing. If only 40% of the goods are packed, carton quantities, assortment, retail packaging, labels, shipping marks, and final pack-out cannot be checked across a properly prepared shipment.

SKU List

Give each SKU its own quantity and packaging reference. Don’t send only the total order quantity.

SKU	Model	Variant	Order Qty	Barcode	Packaging
A01	Model A	Black / M	1,000	Code A01	Rev. C
A02	Model A	Blue / M	800	Code A02	Rev. C
B01	Model B	Black / L	1,200	Code B01	Rev. B

The three rows above total 3,000 units. That total can still be correct while the order is wrong. If A01 has 1,100 pieces and A02 has only 700, the shipment still contains 3,000 pieces, but the SKU quantities do not match the order.

Use the same list to connect each SKU with:

model number;
buyer item number;
color and size;
ordered quantity;
plug or voltage version;
language version;
accessory set;
barcode;
retail packaging;
carton marking.

This matters even more on mixed orders. Six models sold in four colors can create 24 product combinations before size, plug type, or market version is added.

If an earlier During Product Inspection found a wrong label, loose fitting, missing accessory, or dimension problem, put that item on the FRI recheck list after the factory says it has been corrected.

Final Spec

Send one approved version. Don’t leave three specifications, two artwork files, and an old sample in the inspection folder and expect the inspection team to pick the right one.

File	Use During FRI
Purchase order	SKU, quantity, model and assortment
Product specification	Material, components, function and workmanship
Drawing / size chart	Dimensions and tolerances
Approved sample	Agreed appearance, color, construction or workmanship
Packaging artwork	Print, label, barcode, model information and warning text
Inspection instruction	AQL, defect rules, test method and special checks

If the sample has a black screw but the latest specification says silver, settle that before FRI. We should not guess which version the buyer wants.

The same rule applies when something changed after sample approval. A first sample may use one zipper, adapter, fabric, screw, cable, insert, or carton. If another version was later approved for mass production, write that change into the final file.

Also state what the approved sample is approved for. A sample may be:

a color reference;
an appearance reference;
a workmanship reference;
a construction reference;
a full product reference.

A sample approved for color is not automatically the standard for every dimension, component, or function.

When the physical sample still needs approval before bulk production, Sample Evaluation can be used to compare it with the drawing, specification, function, and packing requirements earlier in the order.

Measurements

Write numbers, not phrases such as “normal size,” “good weight,” or “correct length.”

Check	Example Requirement
Overall width	500 ± 3 mm
Cable length	≥ 1.5 m
Hole spacing	80 ± 1 mm
Net weight	450 ± 10 g

These numbers are examples, not UTS acceptance limits. Use the tolerances from the buyer-approved drawing, specification, product standard, or other agreed file.

If a drawing says 500 mm but gives no tolerance, we can record an actual result such as 498 mm. We should not decide that ±3 mm is acceptable unless an approved requirement says so.

For a drawing with 30 or 40 dimensions, mark the ones that matter most during FRI. Typical priorities include:

overall size;
mounting-hole position;
part-to-part fit;
cable or hose length;
garment measurements;
critical thickness;
net weight;
clearances that affect assembly or function.

A five-minute measurement that checks an important assembly point is more useful than spending the same time on a dimension that has no effect on fit, function, or buyer acceptance.

Test Methods

“Function test: pass” is too vague. Write down what the product must actually do.

For each special on-site test, give us:

test item;
test method;
sample quantity;
expected result;
failure condition;
equipment required;
whether the test damages the sample.

For a timer, for example, the instruction could require the unit to power on, accept a time setting, count down, sound the alarm, and reset correctly. That gives five clear points instead of one vague instruction.

An assembly test needs the same level of detail. State which parts are assembled, which tools are used, and what to check after assembly: fit, alignment, stability, fastening, movement, or function.

Keep factory-site inspection separate from laboratory testing. ISO/IEC 17020:2026 sets requirements for bodies performing inspection.^[1] ISO/IEC 17025:2017 applies to testing and calibration laboratories.^[2]

Chemical composition, restricted substances, some electrical safety tests, flammability, endurance, and other controlled tests may need laboratory methods instead of a short on-site check. Our article on product testing and product inspection shows where the two scopes differ.

AQL Numbers

Set the lot size, inspection level, AQL values, and defect categories before FRI.

ISO 2859-1:2026 covers acceptance sampling plans for inspection by attributes and uses Acceptance Quality Limit (AQL) to index the sampling schemes.^[3] ANSI/ASQ Z1.4 is also widely used for attribute sampling and includes normal, tightened, and reduced plans.^[4]

Before the inspection, confirm:

lot size;
inspection level;
major-defect AQL;
minor-defect AQL;
critical-defect rule;
any special sample size.

Here is the example used in the UTS ANSI Sampling Tables:

Input	Example
Lot size	4,000 units
Inspection level	General Inspection Level II
Code letter	L
Main sample	200 units
AQL example	2.5
Acceptance number	10
Rejection number	11

So a buyer should not simply write “check 20 pieces” for that 4,000-unit example. Under the stated plan, the main sample is 200 units.

At AQL 2.5 in this example, 10 is the acceptance number and 11 is the rejection number for the 200-unit sample. That does not mean “2.5% of the whole order may be defective.” AQL works together with the sampling table.

Defect Rules

Decide what counts as critical, major, and minor before the result is known.

Class	Typical Use
Critical	A condition that meets the buyer's defined critical rule, often involving a serious safety or other unacceptable risk
Major	A defect that can affect function, durability, assembly, important appearance, saleability, or a key specification
Minor	A lower-level appearance or workmanship issue that normally does not stop intended use

The product decides the context. A loose decorative trim and a loose structural part are not the same problem. A small mark under a product and a large mark across the main visible surface are not the same either.

Use a product-specific defect list when you already know where problems happen. For example:

wrong barcode found on an earlier batch;
dimension corrected during production;
component changed after sample approval;
wrong accessory found during DPI;
loose screws or fittings found earlier;
wrong manual language used during packing.

If one condition is unacceptable at any frequency under the buyer's rules, write that clearly instead of leaving it inside a normal cosmetic AQL limit.

Sample Sizes

A 200-unit AQL sample does not mean every test needs 200 pieces.

Check	What to Set
Workmanship	Main AQL sample
Dimensions	Measurement items and sample quantity
Function	Method, quantity and expected result
Assembly	Quantity and assembly steps
Destructive test	Approved quantity and method
Packaging	Units and cartons to check

A destructive test may use only a few samples because the tested units cannot be sold afterward. A basic function check may use another agreed quantity. Measurements may focus on a smaller set of units and several important dimensions.

Write these quantities before inspection. Otherwise a buyer may expect 200 function tests while the working instruction only calls for a much smaller sample.

If every unit must be checked for the agreed points, that is not a normal random-sampling FRI. Depending on the order, a Full Inspection may fit that requirement better.

Packaging Files

Match each SKU to the correct product label, retail pack, inner pack, and master carton.

Level	Check Against
Product	Model, product label, rating information and markings
Retail pack	Box or bag, manual, accessories, insert, label and barcode
Inner pack	Quantity, assortment, divider and inner label
Master carton	Quantity, SKU, carton number, shipping mark and packing method

Useful files include:

retail box artwork;
product labels;
rating labels;
polybag markings;
hangtags;
manuals;
warnings;
inserts;
barcode data;
inner-carton labels;
master-carton marks.

Keep only the active version. If Rev. C was approved on August 20, an old Rev. A file should not sit beside it with no status mark.

Watch for cross-SKU errors. A correct Model A product can still have the wrong Model B retail box, Model C manual, or Model D barcode.

Pack-out

Count the assortment, not just the total.

Example: one carton contains 100 pieces and the approved mix is:

25 black;
25 white;
25 blue;
25 red.

If the carton actually contains 40 black, 20 white, 20 blue, and 20 red, it still has 100 pieces. The carton count is right; the assortment is wrong.

Here is another simple example. An order has 100 cartons with 24 units per carton, so the expected packed quantity is 2,400 units. If five cartons contain only 23 units, the shipment is five units short even though all 100 cartons are present.

Use the same check for:

sizes;
colors;
models;
plug versions;
manual languages;
accessories;
market-specific packaging.

Accessory lists need exact quantities too. If one retail box should contain one main unit, one adapter, one cable, one manual, and one accessory kit, list all five items. A product can work perfectly and still be packed incorrectly.

For master cartons, provide the approved requirements for:

SKU or model;
quantity per carton;
PO number;
carton number;
shipping mark;
carton dimensions if required;
gross and net weight if required;
handling marks;
protection and packing method.

If a carton drop test is required, define the method and sample quantity before inspection.

Barcodes

For barcode or QR-code checks included in the FRI scope, the checked samples need a 100% scan-success rate.

We check three things:

Print: the code is clear and not visibly damaged, blurred, covered, or badly printed;
Scan: every checked code scans successfully;
Data: the scan result matches the approved SKU, model, barcode, or label information.

GS1 lists print quality, contrast, damage, position, and packaging interference among the common reasons a barcode may not scan properly.^[5]

If 39 of 40 checked barcodes scan, that is 97.5%. It is still not 100%. The unreadable code should be recorded, corrected, and rechecked according to the buyer's requirement.

Routine UTS product inspection does not use barcode grading or scoring as the normal acceptance method. We check whether it is clearly printed, whether it scans, and whether the returned data is correct.

Missing Files

No approved requirement, no invented limit.

If no tolerance is supplied, record the actual measurement instead of making up a tolerance.
If no final artwork is supplied, document the actual packaging instead of comparing it with an old version.
If the sample and specification conflict, get the final approved requirement before using either as the inspection basis.
If no special test method is supplied, do not create a new pass/fail rule on site.
If barcode reference data is missing, a scan may show whether the code works, but the returned data cannot be confirmed against information that was never provided.

For example, a measured length of 498 mm against a 500 mm target is a recorded difference. Without an approved tolerance, it should not automatically become a “failed dimension” based on a limit we created ourselves.

The same applies to labels. If the actual carton says “Model A” but no approved carton artwork was supplied, the report can show what was found; it should not claim the mark matches a file that was never available.

Final File Pack

Keep the working FRI pack short and current.

File	What to Include
PO + SKU list	Quantity by model, color, size and version
Product specification	Material, construction, dimensions, function and tolerance
Approved sample record	What the sample is approved to represent
AQL instruction	Inspection level, AQL values and defect rules
Test instruction	Method, quantity, result and equipment
Packaging artwork	Final box, label, manual, insert and warning files
Barcode list	Correct barcode matched to each SKU
Packing requirement	Assortment, accessories, carton quantity and carton marks
Recheck list	Problems corrected during sample approval, IPI, DPI or earlier production

Our office team reviews these requirements before arranging the inspection. After FRI, our team provides the recorded sample findings, defect counts, photos, measurements, agreed test results, barcode findings, and packing checks for buyer review.

If the order needs correction after FRI, the next check should focus on what was changed. A buyer dealing with a failed FRI can use a defined recheck or full-inspection plan instead of repeating the same inspection without a clear corrective scope.

Finally

Before FRI, check four numbers first: production should be 100% complete, at least 80% should be packed, the lot quantity should match the PO, and each SKU quantity should add up to the total. Then lock the final specification, tolerances, AQL settings, test quantities, artwork, barcode list, and carton rules. In the UTS AQL example, a 4,000-unit lot at General Inspection Level II gives a 200-unit main sample; other tests can use different agreed quantities. If barcode verification is included, all checked codes must scan successfully. Any missing requirement should be recorded as missing, not replaced with a rule made up during inspection.