Incoming inspection is the check a cosmetics factory runs on bottles, pumps, caps, tubes, labels and cartons when they arrive, before any product is filled into them. AQL sampling is the statistical rule that sets how many pieces to examine and how many defects a lot may contain and still pass. Most cosmetics factories inspect to ISO 2859-1 at general inspection level II, with an acceptance quality limit of 0 for critical defects, 1.0 to 1.5 for major and 2.5 to 4.0 for minor. This tutorial covers skincare, haircare and colour cosmetics packed by OEM and ODM factories for any market; costs are illustrative. First step: write a defect list per component and attach it to the purchase order.
Key takeaways
- AQL is a sampling rule, not a quality promise. At AQL 2.5 with a 200-piece sample, a lot that is really 5% defective is accepted about 58% of the time. If a defect would stop your filling line, set it as critical with an acceptance number of zero and inspect the lot before it ships, not after.
- The defect list is the contract. A lot can pass AQL and still be useless if nobody wrote down that a pump must deliver 0.2 ml per stroke or that the cap torque must sit between 8 and 12 in-lb. The classification of critical, major and minor defects, with a photo or a sample for each, is the document that makes inspection results enforceable.
- A defect found at receipt costs cents; the same defect found at filling costs dollars. On 10,000 pumps with 3% leakers, sorting at incoming is roughly USD 150 to 250 of labour; discovering the leakers during the fill means line downtime, lost bulk and a late delivery, and the realistic bill is closer to USD 2,000 to 3,500 before any customer returns.
- Four parties can inspect, and the brand should decide which. The component supplier’s outgoing check, the filling factory’s incoming check, a third-party inspector at the component plant and the brand’s own golden-sample comparison each prove different things. For a first order of a new component, use two of them.
Who is this tutorial for?
This tutorial is for brand owners, product managers and junior quality staff who buy finished cosmetics from an OEM (Original Equipment Manufacturer) or ODM (Original Design Manufacturer) factory and either supply their own packaging components to it or want to understand how the factory checks the components it buys on their behalf. It assumes no statistics background. It is most useful when you are placing a first order with a new bottle, pump or tube supplier, when a batch has just been delayed by a packaging problem, or when you are writing the quality section of a supply agreement. If your question is which pack to choose in the first place, start with our guide to cosmetic packaging and filling options and come back here once the component is specified.
What is incoming inspection for cosmetic packaging, and who does it?
Incoming inspection is the formal examination of a delivered lot of packaging components against an agreed specification, carried out before the lot is released to the filling line, and it ends with one of four decisions: accept, sort, rework or reject. Under ISO 22716, the cosmetics good manufacturing practice (GMP) standard, packaging materials must be received, identified, quarantined and checked against acceptance criteria before use, and the same principle appears in the US FDA’s draft cosmetic GMP guidance and in HSA Singapore’s voluntary GMP certification scheme (both pages checked 23 October 2026). A factory that is GMP-certified therefore already has an incoming inspection procedure. What it usually does not have is your defect list, your approved sample and your view on which defects matter commercially, and that is the gap this tutorial closes.
Four parties can carry out some form of inspection, and they do not substitute for each other. The component supplier inspects its own output before shipping and may issue a certificate of conformance (CoC) stating that the lot meets the specification; this is useful as a record but it is the supplier marking its own work. The filling factory inspects at receipt against its own standard procedure, which is normally a visual and dimensional check on a sample, plus whatever functional tests its line experience has taught it to run. A third-party inspection company can inspect at the component supplier’s plant before the lot leaves, which is the only option that catches a bad lot before freight has been paid on it. And the brand itself can compare pre-shipment samples against the golden sample it signed off at development. Table 1 sets the four routes side by side.
Table 1. Who can inspect a packaging lot, and what each route proves.
| Route | What it proves | What it misses | Typical cost | Lead-time impact | Best when |
|---|---|---|---|---|---|
| Component supplier’s outgoing check plus CoC | The supplier believes the lot meets its own understanding of the spec | Anything not on the supplier’s own checklist; fit to your filling line; your decoration standard | Included in component price | None | Repeat orders of a proven component from a supplier with audit history |
| Filling factory’s incoming inspection | The lot will run on this line and matches the factory’s retained sample | Defects the factory does not class as its problem (print colour, carton finish); lots already paid and shipped | Included in fill price for standard checks; functional tests may be charged | 1–3 days after receipt | Every lot, as the baseline; the brand supplies the defect list |
| Third-party inspection at the component plant | An independent AQL result on the actual lot before it ships, with photos and measurements | Problems that appear in transit; compatibility with the bulk | USD 250–400 per inspector-day plus travel, illustrative | 2–5 days to book; the lot waits for the report | First order of a new component, overseas supplier, decorated or functional parts (pumps, droppers, airless) |
| Brand’s own golden-sample comparison | The pre-shipment or first-off samples match what you approved | Lot-level variation; everything you cannot measure at a desk | Your time plus courier | 3–7 days for samples to travel | Any lot where decoration, colour or a new mould is involved |
The practical rule for a first production order is to use at least two routes: the factory’s incoming inspection as the baseline, and either a third-party inspection at the component plant or a courier of first-off samples to you, depending on whether the risk is functional (pumps, droppers, airless packs) or cosmetic (print, colour, finish). If you source components yourself rather than through the factory, you also inherit the coordination between routes, a cost that our comparison of buying from a factory, a trading company or a sourcing agent sets against the margin you save.
How does AQL sampling work, and which AQL should you set for bottles, pumps and cartons?
AQL stands for acceptance quality limit: the worst average percentage of defective pieces that a sampling plan will accept most of the time, and the number is chosen by the buyer, not measured by the inspector. The standard that cosmetics factories and inspection companies use is ISO 2859-1 (the equivalent US standard is ANSI/ASQ Z1.4), which the British Standards Institution describes as sampling procedures for inspection by attributes, indexed by AQL, for lot-by-lot inspection (BSI Knowledge, checked 23 October 2026). The US National Institute of Standards and Technology explains the underlying idea in its engineering statistics handbook: a lot at the AQL should be accepted with high probability, and a lot that is much worse should be rejected with high probability, with the sample size and acceptance number set to make both true.
The mechanics take three steps. First, the lot size and the inspection level give a sample size code letter; nearly all cosmetics inspection uses general inspection level II, which is the standard’s default. Second, the code letter gives the sample size. Third, the sample size and the AQL together give an acceptance number (Ac) and a rejection number (Re): if the inspector finds Ac or fewer defects of that class in the sample, the lot is accepted; at Re or more it is rejected. Because critical, major and minor defects carry different AQLs, one sample is examined three times against three thresholds, and the lot must pass all three. Table 2 reproduces the single-sampling, normal-inspection values for the lot sizes most common in cosmetics.
Table 2. ISO 2859-1 / ANSI/ASQ Z1.4 single sampling, normal inspection, general level II. Ac / Re = accept on this many or fewer / reject on this many or more.
| Lot size (pieces) | Code letter | Sample size | AQL 0.65 | AQL 1.0 | AQL 2.5 | AQL 4.0 |
|---|---|---|---|---|---|---|
| 281 to 500 | H | 50 | 1 / 2 (uses the 80-piece plan) | 1 / 2 | 3 / 4 | 5 / 6 |
| 501 to 1,200 | J | 80 | 1 / 2 | 2 / 3 | 5 / 6 | 7 / 8 |
| 1,201 to 3,200 | K | 125 | 2 / 3 | 3 / 4 | 7 / 8 | 10 / 11 |
| 3,201 to 10,000 | L | 200 | 3 / 4 | 5 / 6 | 10 / 11 | 14 / 15 |
| 10,001 to 35,000 | M | 315 | 5 / 6 | 7 / 8 | 14 / 15 | 21 / 22 |
Which AQL to set is a commercial decision that depends on what a defect does downstream. The convention in cosmetics is critical 0, major 1.0 or 1.5, minor 2.5 or 4.0; general consumer-goods inspection often runs looser at 0 / 2.5 / 4.0, and a brand that leaves the choice to the inspection company will usually get the looser set. A critical AQL of 0 means Ac = 0 at every sample size: one critical defect rejects the lot. That is appropriate for anything that makes the product unsafe or unsaleable (a cracked bottle, a pump that does not prime, a wrong barcode, a missing ingredient list) and it is the setting most brand owners mean when they say they want “zero defects”.
What AQL does not do is guarantee a defect rate, and this is the point most often misunderstood. The sampling plan is a probability filter, and the probability of accepting a lot slides smoothly as the lot gets worse. Table 3 shows the arithmetic for a 200-piece sample (lots of 3,201 to 10,000), calculated from the binomial distribution; the NIST handbook’s operating characteristic curve is the same calculation drawn as a graph.
Table 3. Probability that a 200-piece sample accepts a lot, by the lot’s true defect rate (our calculation, binomial, single sampling).
| True defect rate in the lot | AQL 0 (Ac 0) | AQL 0.65 (Ac 3) | AQL 1.0 (Ac 5) | AQL 2.5 (Ac 10) |
|---|---|---|---|---|
| 0.5% | 37% | 98% | 100% | 100% |
| 1% | 13% | 86% | 98% | 100% |
| 2% | 2% | 43% | 79% | 100% |
| 3% | 0% | 15% | 44% | 96% |
| 5% | 0% | 1% | 6% | 58% |
| 8% | 0% | 0% | 0% | 7% |
Two things follow. A lot that is genuinely 3% defective passes a major-defect AQL of 2.5 almost every time and a major AQL of 1.0 less than half the time, so the choice between 1.0 and 2.5 is not cosmetic: it is the difference between receiving 3% faulty pumps as normal and sending them back. And an AQL of 0 is harsh on good suppliers, because a lot that is only 0.5% defective is rejected nearly two times in three; reserve it for defects that are genuinely critical, or the supplier will price the rejections into the component. If a component is both critical and high-volume, a pump or dropper for example, a tighter plan is cheaper than a looser one followed by 100% sorting at the factory, which is the usual outcome when a bad lot reaches the filling line.
Which defects count as critical, major and minor on cosmetic components?
A defect is critical if it makes the product unsafe, illegal or unusable, major if a customer would reject or return the unit, and minor if it is visible on close inspection but would not stop a sale, and the same physical fault can sit in different classes for different products. A 1 mm print scratch on the back of a 400 ml shampoo bottle is minor; the same scratch across the brand name on a USD 90 serum is major. The defect list therefore has to be written per component and per product, agreed with both the component supplier and the filling factory, and illustrated with photographs or limit samples that show the boundary between acceptable and not. Table 4 gives a starting classification for the components most private-label brands buy; treat it as a template to edit, not a standard.
Table 4. Starting defect classification by component (edit per product; attach photos or limit samples).
| Component | Critical (AQL 0) | Major (AQL 1.0–1.5) | Minor (AQL 2.5–4.0) | Key measurement |
|---|---|---|---|---|
| Bottle or jar (glass, PET, PP) | Crack, chip, hole, sharp edge; wrong neck finish; wrong material or resin code | Visible contamination inside; deformed shoulder; wall thickness below spec; colour outside approved range | Light scuff, flow line, small bubble in glass, faint mould seam | Neck finish (e.g. 24/410) with go/no-go gauge; weight; height; wall thickness |
| Pump, sprayer, dropper | Does not prime or dispense; leaks when inverted; wrong dip-tube length; thread does not seat | Output per stroke outside spec; stiff or sticking actuator; spray pattern off; dip tube cut at angle | Cosmetic mark on collar; slight colour variation on actuator | Output per stroke (weigh 10 strokes); priming strokes; inverted leak test; dip-tube length |
| Cap, closure, disc-top | Cross-threaded or will not close; liner missing; broken hinge | Torque outside range; liner loose; orifice blocked; visible sink mark on top | Minor flash on edge; slight gloss variation | Application and removal torque; liner presence; orifice diameter |
| Tube (laminate, PE) | Seal or seam leak; wrong orifice; cap will not seat; print of wrong artwork version | Print misregistration above 0.5 mm; colour outside tolerance; crushed or kinked body | Faint scratch on body; slight print density variation | Orifice diameter; print registration; colour against standard under D65 light |
| Label and decoration (screen, hot foil, pad) | Wrong text, barcode, batch or language version; missing mandatory element; unreadable barcode | Skewed label above 1–2 degrees; bubbles, wrinkles; foil lifting under tape test; rub-off | Dust spot under 0.3 mm; slight colour drift within tolerance | Barcode scan and verification grade; adhesion (cross-hatch tape test); rub resistance; alignment |
| Carton, leaflet, shipper | Wrong artwork or market version; missing regulatory text; carton will not fold or close | Crease cracking on colour; glue flap open; scuffing on face panel; wrong board weight | Faint mark on inside; slight colour variation on non-face panel | Board weight (gsm); fold and crush; colour; glue adhesion |
The right-hand column matters as much as the classification, because an inspector can only reject what can be measured or compared. For every functional test, write down the method, the specification and the number of pieces to test. Pump output is weighed over ten strokes after priming and expressed as an average per stroke with a tolerance (a 0.2 ml lotion pump is commonly specified at ±10%). Decoration adhesion is checked by the cross-hatch tape method in ASTM D3359; a transit test on the packed shipper follows an ISTA procedure or ASTM D4169, which ISTA lists alongside its own. Barcodes are checked with a verifier that reports a grade against the GS1 General Specifications, not just a phone scan, because a code that scans on a phone at 10 cm can still fail at a retail checkout. (All four pages checked 23 October 2026.) What a label must carry in the first place, and who supplies each element, is covered in our packaging and label requirements checklist; how the artwork gets to a print-ready state without a wrong-version carton reaching inspection is the subject of our artwork approval tutorial.
One class of defect cannot be found at incoming inspection at all: incompatibility between the pack and the formula. A pump that passes every test on water can gum up on a silicone-rich serum; a PET bottle can stress-crack with a high-fragrance toner; a hot-foil logo can lift when the carton is stored next to an oil-based product. These are discovered by filling the real bulk into the real component and storing it, which is why compatibility is part of stability testing and belongs in the pre-launch testing checklist, months before the first production lot of components is inspected.
How do you run an incoming inspection step by step?
An incoming inspection follows the same eight steps whether it is done by the factory, an inspector or your own staff: confirm the lot, draw a random sample, inspect against the defect list, measure, test function, record, decide, and feed the result back to the supplier. The procedure below is written for a brand that wants to specify what the factory does or to run a check itself on first-off samples; a GMP-certified factory will already have most of it in its own format.
- Confirm what you are inspecting. Match the delivery note to the purchase order: component code, artwork version, quantity, supplier batch number. One supplier batch delivered on one date is one lot. If a delivery contains two supplier batches, treat them as two lots with two samples, because a mould change or a resin lot change between them is exactly what you are looking for.
- Quarantine the lot. Under GMP the lot is labelled as not released and stored apart from approved stock until the decision is recorded. Ask to see where your factory keeps unreleased packaging; a lot already on the filling line before inspection is a finding, not a formality.
- Draw a random sample across the whole delivery. Open cartons from different pallets and different positions, not the top layer of the nearest pallet. A common working rule is to open at least the square root of the number of cartons plus one, and to draw pieces from each. The sample size comes from Table 2; for a 6,000-piece lot that is 200 pieces from about nine or more cartons.
- Inspect visually against the defect list and the approved sample. Examine each piece under consistent lighting for the time it takes to see every surface; for decorated items that is typically a few seconds per piece, at arm’s length and then close. Tally defects by class. A piece with two defects counts once, in the more serious class.
- Measure. Take the key dimensions in Table 4 on a sub-sample, usually 5 to 10 pieces for dimensions and 10 to 20 for weight, and compare with the specification and the drawing. Neck finishes are checked with the factory’s go/no-go gauge, because a bottle that is 0.3 mm out on the thread will not seal with the cap you have already bought.
- Test function. Pumps, sprayers, droppers and closures get their functional tests on a defined sub-sample: priming and output on 10 to 20 pumps, torque on 10 closures, an inverted leak test on filled or water-filled units over 24 hours. Record the numbers, not just pass or fail, so that a drift between lots is visible.
- Record and decide. The report states lot identity, sample size, defects found by class against Ac/Re, measurements, functional results, photographs of every rejected piece, and one of four dispositions: accept; accept after 100% sort (who sorts, who pays); rework (for example re-labelling) with a re-inspection; or reject and return. Sign and date it. A certificate of analysis or inspection report you cannot read is as good as none, and our tutorial on reading a spec sheet and CoA walks through the lines to check.
- Close the loop. Keep retained samples of every lot, good and bad, for the life of the product plus a year. Send the supplier the report and ask for a corrective action on any rejection or on two consecutive lots at the same major defect. Under ISO 2859-1, two rejections in five consecutive lots also move the supplier to tightened inspection, which is the formal version of the same idea.
Timing is the step brands most often get wrong. Components must arrive early enough for the inspection, a possible sort and a possible replacement before the filling slot, which in practice means two to four weeks ahead for a domestic supplier and six to eight for an overseas one. A lot that arrives the day before filling can only be accepted or the slot lost, and the factory will usually accept, because the alternative is an idle line. The filling line’s own constraints, and what a lost slot costs in changeover, are set out in our guide to fill tolerance and filling-line minimums.
What does a packaging defect cost, and who pays for it?
The cost of a component defect is set almost entirely by where it is found: at the component plant it costs a report, at incoming it costs a sort, at the filling line it costs downtime and bulk, and at the customer it costs the brand. Table 5 works through one illustrative case, 10,000 lotion pumps for a 200 ml body lotion where 3% (300 pumps) leak when inverted. The figures are prepared by oemhallmark from quotes and batch records we have seen; they are not survey data, and your own numbers will differ with labour rates and product value.
Table 5. The same defect found at four points: 10,000 pumps, 3% leakers, 200 ml lotion at USD 1.40 bulk cost per unit, filled 2,500 units per hour (illustrative).
| Found at | What happens | Direct cost, illustrative | Schedule impact | Who normally pays |
|---|---|---|---|---|
| Component plant (third-party inspection before shipping) | Lot rejected; supplier sorts or remakes before freight | Inspection fee USD 300–400; no component loss | 3–10 days if the supplier sorts; a remake adds the mould cycle | Brand pays the inspection; supplier pays the sort or remake |
| Filling factory incoming inspection | Lot fails AQL 1.0 (Ac 7 of 315 at level II); 100% inverted leak test on all 10,000 | Sort labour about USD 150–250 (10,000 pieces at a few hundred per hour); 300 replacement pumps and freight USD 100–200 | 2–4 days; filling slot usually holds if components arrived early | Supplier pays replacements if the spec and AQL were on the purchase order; sort labour is negotiated |
| During filling | Leakers found when filled units are inverted at pack-out; line stops; filled units with bad pumps must be opened, refilled and recapped | Line downtime 4–6 hours at USD 100–200 per hour; 300 units of bulk and labour at about USD 2 per unit; re-fill and re-pack labour; realistic total USD 2,000–3,500 | Batch late 1–2 weeks; next filling slot may be lost to another customer | Contested: factory argues it fills what it is given; brand argues GMP incoming inspection should have caught it |
| At the customer | Leaking parcels, returns, reviews, a marketplace listing suspended; possible recall if product is contaminated | Returns at USD 8–15 per unit landed plus replacements; review damage not quantifiable; a recall is a separate order of magnitude | Weeks; relaunch cost | Brand, almost entirely; recovery from the supplier is limited to the component value unless the contract says otherwise |
The table is why the quality clauses in a supply agreement matter more for packaging than for bulk: bulk is tested before release under GMP as a matter of course, while packaging is only checked as thoroughly as somebody specified. Six clauses do most of the work. The component specification, drawing and defect list are attached to the purchase order by reference. The sampling standard, inspection level and AQLs per class are stated. The approved sample is named, dated and held by both parties. Rejection follows a written procedure with a time limit for the supplier to respond. Replacement lead time and who pays return freight are fixed. And liability for consequential loss, the downtime and bulk in the third row of Table 5, is either capped or expressly included, because it is otherwise excluded by most suppliers’ terms. Worked wording for these clauses, and for the inspection standard most often cited, is in our examples of OEM cosmetics contracts, and the related question of when a deposit should be released against inspection is covered in the payment terms guide.
Finally, inspection is also the point at which a brand learns whether its factory is as careful as its certificate suggests. When you compare manufacturers in the oemhallmark directory, ask each one for its incoming inspection procedure for packaging, a blank inspection report, and the AQLs it applies by default; a factory that answers with a form and numbers is operating a system, and one that answers “we check everything” is not. The fuller version of that visit is our factory audit checklist, where packaging receipt is one of the ten areas.
Questions to ask the factory or component supplier before the first lot
- Which sampling standard, inspection level and AQLs do you apply to packaging components by default, and will you apply ours if they are tighter?
- Where is unreleased packaging stored, and what label does it carry?
- Which functional tests do you run on pumps, closures and tubes at receipt, on how many pieces, and are the results recorded as numbers?
- Who holds the approved sample for each component, and how is it referenced on the purchase order?
- If a lot fails, who sorts, who pays for the sort, and how many days do you need before the filling slot is lost?
- Can we see a completed incoming inspection report for a comparable component, with the customer’s name removed?
Mistakes to avoid
- Specifying “AQL 2.5” with no classes. A single number applied to all defects means a cracked bottle and a scuffed one count the same. Always set three classes with three AQLs.
- Approving a golden sample that was hand-picked. Ask for the sample to be drawn from a production run, not from the supplier’s showcase, and keep two: one with you, one with the factory.
- Sampling from the top of one pallet. A supplier that knows how inspections are done will put the best cartons where they will be opened. Draw across pallets and positions.
- Testing pumps dry. Priming and output must be tested with water or the bulk, on units assembled to the bottle they will be sold in, with the dip tube cut to length.
- Letting components arrive the day before filling. It converts every inspection result into “accept”, because the alternative is losing the slot.
- Treating a passed AQL as a defect-free lot. It is not; see Table 3. Keep retained samples and track defects per lot so that a supplier drifting from 0.5% to 2% is visible before it becomes 5%.
- Forgetting the printed batch code. If the factory prints the batch number and expiry onto the component or carton at filling, the legibility and position of that print are part of final inspection, not incoming inspection; the format itself is explained in our piece on shelf life, PAO and batch codes.
Frequently asked questions
What AQL is normal for cosmetic packaging?
The common setting in cosmetics is critical 0, major 1.0 or 1.5 and minor 2.5 or 4.0, at ISO 2859-1 general inspection level II. General consumer-goods inspection often uses 0 / 2.5 / 4.0, which accepts about twice as many major defects. Choose the tighter major AQL for functional parts such as pumps, droppers and airless packs, and write the chosen values on the purchase order so that both the supplier and the inspector apply the same plan.
Does passing an AQL inspection mean the lot has no defects?
No. AQL is a sampling rule that accepts lots at or near the stated defect rate with high probability and rejects much worse lots with high probability; lots in between are accepted or rejected by chance. At AQL 2.5 on a 200-piece sample, a lot that is 5% defective is still accepted about 58% of the time. If a particular defect must never reach the line, class it as critical with an acceptance number of zero, and consider 100% inspection of that one characteristic.
Who should inspect packaging components, the factory or the brand?
Both, in different ways. A GMP-certified filling factory must inspect components at receipt and will do so against its own procedure; the brand’s job is to supply the defect list, the approved sample and the AQLs so that the factory inspects to the brand’s standard. For a first order of a new or decorated component, add either a third-party inspection at the component plant before shipping or a courier of first-off samples to you, so that a bad lot is caught before freight and filling time are spent on it.
How many pieces are inspected in a lot of 5,000 bottles?
Under ISO 2859-1 single sampling at general inspection level II, a lot of 3,201 to 10,000 pieces has code letter L and a sample of 200 pieces. At AQL 1.0 the lot is accepted with 5 or fewer major defects in those 200 and rejected at 6 or more; at AQL 2.5 the numbers are 10 and 11. Dimensional and functional tests are usually run on a smaller sub-sample of 5 to 20 pieces drawn from the same 200.
What is the difference between incoming inspection and pre-shipment inspection?
Incoming inspection is done on packaging components when they arrive at the filling factory, before use. Pre-shipment inspection is done on finished, filled goods before they leave the factory for the brand or its warehouse, and covers fill weight, coding, assembly, cartoning and pallet build as well as the components. A brand that relies only on pre-shipment inspection finds packaging defects after the bulk has been filled into them, which is the most expensive point in Table 5 short of the customer.
Can a supplier refuse our AQL?
A supplier can decline to accept a tighter AQL than its process can deliver, and it is better to learn that at quotation than at rejection. Ask the supplier what defect rate its process normally achieves on the component, request its last few outgoing inspection reports, and set the AQL just inside that. A supplier that will not state any defect rate is telling you how it will respond to a rejection.
Sources and limitations
Standards and regulatory points were checked on 23 October 2026 against the BSI Knowledge entry for ISO 2859-1 (sampling procedures for inspection by attributes), the NIST/SEMATECH Engineering Statistics Handbook sections on acceptance sampling and operating characteristic curves, ASTM D3359 (tape test for adhesion) and ASTM D4169 (performance testing of shipping containers), the ISTA test-procedure list, the GS1 General Specifications, the US FDA’s draft guidance on cosmetic good manufacturing practices and its page on the Modernization of Cosmetics Regulation Act, HSA Singapore’s GMP certification page, the European Commission’s cosmetics legislation page for Regulation (EC) No 1223/2009 and the UK Cosmetic Products Enforcement Regulations 2013. The sample sizes and acceptance numbers in Table 2 are our transcription of the single-sampling normal-inspection tables common to ISO 2859-1 and ANSI/ASQ Z1.4 for the lot sizes shown; the full standard covers other inspection levels, tightened and reduced inspection and double and multiple sampling, which are not reproduced here. The acceptance probabilities in Table 3 are our own binomial calculation. The defect classification in Table 4 is a template drawn from inspection reports and specifications we have seen, not a published standard, and it must be edited for each product. The costs, labour rates, line speeds and timescales in Table 5 and the text are illustrative figures prepared by oemhallmark; they are not survey data and no named factory, supplier or inspection company is described. ISO 22716 and ISO 2859-1 are paid standards and are named rather than linked.
This article is general information for beauty brand owners and is not legal, regulatory or engineering advice. Confirm current requirements with the relevant authority, the full text of the standards cited and your own advisers before you finalise a component specification, inspection plan or supply agreement.