SAMPLING PLAN FOR PRODUCT INSPECTION The different types of sampling plans for QC inspections If an inspector controls the

quality of your products in China, he probably checks only a portion of the whole batch But how does he decide how many pieces to pick for his inspection? In other words, what sampling plan does he follow?

Page7

The most commonly used is the single-stage sampling plan by attributes. What does it mean, and what are the other options? A single-stage sampling plan dictates that a certain number of pieces (n) should be drawn and inspected. That number n depends mainly on the size of the batch (and also on the inspection level). If the number of defects is under the AQL limit, the result is passed. In a double-stage sampling plan, the inspector would start by taking a smaller number of sample (n1). If the number of defects in n1 is above certain limits, more samples are picked. An accept on zero plan is a weird animal that is usually not a good idea in China. Some importers, who are sensitive to legal litigation by their customers, accept batches only if no defective unit is found. The only advantage is the fewer samples need to be checked. More in-depth information here. A sampling plan by attributes classifies the samples as either non defective or defective. There is nothing in between. A sampling plan by variables allows for a finer evaluation. For example, the length of the product is measured, and the exact findings are taken into account when a decision is made. A rectifying sampling plan is applicable if the defects that are found can be corrected immediately. It is not very different from a sampling plan by attributes, but it takes into account that the batch is of higher quality after the inspection and, in case of inspection failure, the whole batch should be inspected. Continuous sampling is the best plan when products are made individually in a continuous flow. It makes no sense to pick samples inside each batch. It consists of several phases: At the beginning, each piece is checked (thats the screening). After a certain number of pieces were found satisfactory, only certain pieces are checked randomly (thats the sampling). If a piece is defective: back to screening. If you want to use one of these plans, you should visit the excellent SQC Online website and get the numbers you will need. And if you want more in-depth information about these plans, you should read Practical Acceptance Sampling: A Hands-On Guide. It was pointed out to me a few months back by Etienne Charlier. I pasted this books description (from the publisher) below. I should add that it is a college textbook first, before being a practitioners guide. Acceptance sampling plans provide criteria and decision rules for determining whether to accept or reject a batch based on a sample. They are therefore widely used by manufacturers, suppliers, contractors and subcontractors, and service providers in a wide range of industries. The book introduces readers to the most popular sampling plans, including Military Standards and civilian ISO and ANSI/ASQC/BS standards. It covers the design, choice and performance evaluation of different types of plans, including single- and double-stage plans, rectifying and non-rectifying plans, plans for pass/fail and continuous measurements, continuous sampling plans, and more. What do you think?

Back to basics: what is the AQL? The AQL tables are statistical tools at the disposal of buyers (for product inspections). They help determine two key elements: How many samples should be inspected? Where is the limit between acceptability and refusal, when it comes to defective products?

Page7

The need for an objective measurement of quality In virtually every production batch, there will be defective products. It is true even after the manufacturer has checked each individual product and has repaired the defective ones. Thus, in a supplier/buyer relationship, the supplier cannot be expected to deliver defect-free goods. However, the buyer wants to control the quality of purchased goods, since he does not want too many defects. But what does too many mean? How to set the limit between acceptability and refusal in a way that can be agreed upon and measured? Definition and application of AQL The limit, as described above, is called the AQL. It stands for Acceptance Quality Limit, and is defined as the quality level that is the worst tolerable (ISO 2859 standard). For example: I want no more than 1.5% defective items in the whole order quantity means the AQL is 1.5%. In practice, three types of defects are distinguished. For most consumer goods, the limits are: 0% for critical defects (totally unacceptable: a user might get harmed, or regulations are not respected). 2.5% for major defects (these products would usually not be considered acceptable by the end user). 4.0% for minor defects (there is some departure from specifications, but most users would not mind it). These proportions vary in function of the product and its market. Components used in building an airplane are subject to much lower AQL limits. Getting familiar with the AQL tables Before using the AQL tables, you should decide on three parameters: The lot size. If you ordered different products, the quantity of each product is a lot size, and separate inspections should be carried out for each lot. If you ordered only one product, the lot size is your total order quantity. The inspection level. Different inspection levels will command different number of samples to inspect. In this article, we will stick to the so-called level II, under normal severity. The AQL level appropriate for your market. If your customers accept very few defects, you might want to set a lower AQL for both major and minor defects. There are basically two tables. The first one tells you which code letter to use. Then, the code letter will give you the sample size and the maximum numbers of defects that can be accepted. First table: sample size code letters

Page7
How to read this table? It is very easy. If you follow my example, I assume your lot size is comprised between 3,201pcs and 10,000pcs, and that your inspection level is II. Consequently, the code letter is L. Second table: single sampling plans for level II inspection (normal severity) How to read this table? Your code letter is L, so you will have to draw 200pcs randomly from the total lot size. Besides, I assume you have set your AQL at 2.5% for major defects and 4.0% for minor defects. Therefore, here are the limits: the products are accepted if NO MORE than 10 major defects AND NO MORE than 14 minor defects are found. For example, if you find 15 major defects and 12 minor defects, the products are refused. If you find 3 major defects and 7 minor defects, they are accepted. Note: in quality inspections, the number of defects is only one of the criteria. It is sometimes called quality, or quality findings. The other criteria are usually on the inspectors checklist, which typically includes: Packaging conformity (barcodes, inner packing, cartons, shipping marks).

Product conformity (aspect, workmanship). If all the products are in red color instead of orange, there is no need to count each sample as a defect. It makes more sense to refuse for product conformity. Specific tests defined in the inspection checklist (they might not be performed on all inspected samples if they are time-consuming or destructive). To go further

Page7

Related articles: all the basics an importer should know about quality control are explained in the articles listed on this page. Author: Renaud Anjoran is an expert in quality assurance and quality control. His agency helps buyers in China. More info here. Frequently asked questions about AQL Q: What are the reduced and tightened inspection severities? A: They are designed to be used in very specific situations, when a producer is particularly reliable, or on the contrary fails too often. In practice, these severities are used in less than 1% of QC inspections. The normal severity already allows for a good variation of sample sizes. In the vast majority of cases, third-party inspectors follow only the normal severity. Q: So, basically I have to authorize the factory to produce some defects? A: Yes, some defects, since a factory cannot reasonnably be expected to turn out 100% good quality. However, it does not mean the buyer tolerates everything as long as the number of defects are below the AQL limits. Please see below the note issued in the ISO2859 standard: Although individual lots with quality as bad as the acceptance quality limit may be accepted with fairly high probability, the designation of an acceptance quality limit does not suggest that this is a desirable quality level. Sampling schemes [...] are designed to encourage suppliers to have process averages consistently better than the AQL. Q: Based on my AQL, I calculated the proportion of defects authorized. Why dont they correspond to the maximum number of defects authorized? A: It is true. In our example above, 2.5% of 200 samples is 5 samples, but we accept the goods even if 10 samples are found with a major defect. Why this difference? There are heavy statistics behind this issue. To rejection (based on the random element when drawing the sample) checked) would be accepted. And, in the same logic, the consumer statisticians had to account for these risks, thats why the numbers were make it simple, the producer runs a risk of even though his products (if they were all runs a risk of accepting bad products. The adjusted and seem not to make sense.

Q: Why not just say, well check 10% of the quantity, or whatever percentage deemed appropriate? A: Here again, the statisticians tell us it is not that simple. As we go up in the total quantity, the proportion of products checked can decrease, for the same confidence in the inspection results.

As you can see in the chart below, the number of samples to check (vertical axis) increases at a slower pace than the total quantity (horizontal axis). Q: How to choose an AQL limit for my products? A: See this article. How to choose an AQL limit?

A few months ago, I wrote an explanation about the AQL: what it is and how to use it. I also listed a few frequent questions, but it seems like I forgot one of them:

What AQL tolerance is suitable to my products? Unfortunately, it is hard to base this on statistical reasoning. So I dont have any definite answer.

Page7

The right AQL depends on two things, I think: The market you are selling into, The kind of risk the users run by using/consuming/getting close to your product. 1. The market The most common AQL chosen by importers is 2.5% for major defects, 4.0 for minor defects, and 0.1 for critical defects. It is considered the standard tolerance for most consumer products sold in supermarkets in North America and in Europe. Based on this standard, you can adjust an AQL that is a bit stricter (say, 1.0/2.5/0.10) if you sell your products in a high-end boutique channel. Or a bit looser (say, 4.0/6.5/0.1) for sale on a low-end market. 2. The user risk For many car and plane parts, and for pharmaceuticals, the accepted defect rate is much lower than 1%. Thats because bad part might cause death. Specialists often use six sigma tools to reduce the opportunities for defects in every process along the manufacturing cycle. I am not very familiar with these industries. I guess there are rules of thumb specific to each field. To sum up: There are no guidelines for deciding what AQL limits to choose. You have to decide what your tolerance. If the whole batch should not contain more than 1.0% of a certain kind of defect (over the long run), then the AQL should be 1.0% for this kind of defect. Is it clear?

Back to basics: what is an inspection level? This article introduces the different options available to buyers, when it comes to the representativity of inspection findings. Inspection level II (under normal severity) is appropriate for most inspections. But it is sometimes necessary to increaseor or reducethe number of samples to check. The need for sampling, rather than 100% checking When controlling the quality of a batch of products, it is not practical to inspect 100% of them (unless the quantity is very small). Inspecting a large number of products takes a long time: it is expensive, and inspectors are less effective as they get tired. Actually, a 100% check does not yield that much more information than inspecting a statistically representative sample. The question becomes: how many products to check? Why different inspection levels? There is a fairly obvious principle in statistical quality control: the greater the order quantity, the higher the number of samples to check.

But should the number of samples ONLY depend on the order quantity? What if this factory had many quality problems recently, and you suspect there are many defects? In this case, you might want more products to be checked. On the other hand, if an inspection requires tests that end up in product destruction, shouldnt the sample size be drastically reduced? And if the quality issues are always present on all the products of a given batch (for reasons inherent to processes at work), why not check only a few samples? For these reasons, different levels are proposed by MIL-STD 105 E (the widely recognized standard for statistical quality control). It is usually the buyers responsibility to choose the inspection levelmore samples to check means more chances to reject bad products when they are bad, but it also means more days (and dollars) spent in inspection. The 3 general inspection levels Level I Has this supplier passed most previous inspections? Do you feel confident in their products quality? Instead of doing no quality control, buyers can check less samples by opting for a level-I inspection. However, settling on this level by default, in order to spend less time/money on inspections, is very risky. The likelihood of finding quality problems is lower than generally recommended. Level II------It is the most widely used inspection level, to be used by default. Level III If a supplier recently had quality problems, this level is appropriate. More samples are inspected, and a batch of products will (most probably) be rejected if it is below the quality criteria defined by the buyer. Some buyers opt for level-III inspections for high-value products. It can also be interesting for small quantities, where the inspection would take only one day whatever the level chosen. The 4 special inspection levels These special levels can be applied in cases where only very few samples can be checked. Four additional special levels, S-1, S-2, S-3 and S-4 [] may be used where relatively small sample sizes are necessary and larger sampling risks can be tolerated (ISO 2859 standard). Under S-3 level, the number of samples to check is lower than under S-4, and so on. In practice: for consumer goods, quality control is usually performed under the general levels. The special levels are used only for certain tests that either take lots of time or destroy the samples. Another situation where special levels are appropriate is a container-loading supervisionto have an idea of what is inside the cartons, without spending too much time at that checking. Two examples to get an clearer understanding Lets say you have ordered 5,000 pcs of a product. In the table below, you can see how many samples would be drawn under each of the 7 inspection levels.General inspection levels Special inspection levels I 80pcs II III S-1 S-2 8pcs S-3 20pcs S-4 32pcs

Page7

200pcs 315pcs 5pcs

As you can see, the numbers of samples to check vary from 5pcs to 315pcs. But a trained inspector might be able to do it in one day, whatever the inspection level you choose. Now lets say you have ordered 40,000pcs of a product. Again, you can see the differences in sample sizes.General inspection levels Special inspection levels I II III S-1 S-2 S-3 S-4

200pcs 500pcs 800pcs 8pcs

13pcs

32pcs

80pcs

In this case, the inspection might take one day of work (for S-1, S-2, S-3, S-4, or reduced level), two days (under level II), or three days (under level III). Sampling plan calculations for quality inspections Most inspections only involve 1 or 2 references, and can be conducted in one man-day (i.e. one inspector on site for one day). However, in certain situations, deciding on a sample plan is very difficultsee my previous article about shipments including many references. This is one of the most difficult things to explain to my clients. Fortunately, my designer managed to explain it all nicely in this illustration:

Page7

Page7

Page7

Page7
How to arrange reference samples for QC inspections? If you pay for quality inspections on your products before they are shipped, you should try to make sure the inspectors have a reference sample in hand. It is one of these things that seem obvious, but that dont get done in many cases due to logistical obstacles. Where do these samples come from? Here are the two most frequent cases: 1. The factory develops pre-production samples until a perfect samples is approved by the customer. At that point, one sample needs to be sent to the inspection company and, ideally, another one should go back to the factory to make sure the right prototype is their model.

2. The customer asks for production samples (or shipment samples, which are basically production samples with full individual packing), and sends one sample back for QC purposes. Logistics & timing obstacles As mentioned above, the samples need to be sent back to the producing country (lets say China). But where, specifically?

Page7

Directly in the factory? Some clients have enough trust in the suppliers and accept this option. But the factory might tamper with the samples In the inspection companys office? This sounds safer. But how do the samples get in the right inspectors hands at the right time? It is not feasible when inspectors live alone and far from the office, and if they are out in the field when the Fedex guy shows up. There is also a timing problem. If production is on the way for 10 days when the finished samples are ready, if it takes 4 days for the customer to receive and approve the samples, and if it takes another 3 days for the samples to get back to China, the samples might arrive too late. This is why, in most cases, the pre-production sample are used for QC purposes. One option: sealing the pre-production samples in the factory Once a pre-production sample is approved, it can be sent to the factory and sealed (or stamped, or vacuum packed). The goal is that nobody can play with the reference samples. However, the ideal would be the production/shipment samples for two reasons: They are naturally closer to what the inspectors will see. It eliminates discussions such as but the customer accepted this when we went into production or but the colors are never exactly the same in the bulk and on pre-production swatches. Another option: quick review of production samples The factory sends the samples to someone in China (it might be an inspection company, a sourcing agent) who compares all the samples. If they look the same and are the same size, same weight etc., one sample is sent to the purchaser. The other one remains in China. (In case they are not similar, the supplier needs to re-send another set of samples). Once the purchaser receives the sample and he approves it, the good sample is already in China. The supplier is happy because he didnt have to pay Fedex or DHL to send the samples overseas. The client still needs to pay the international courier cost once (same cost). Anybody has other experiences/suggestions? How your inspectors can fail to notice quality problems Performing a quality inspection prior to a shipment is like a filter. It wont catch 100% of the problems 100% of the time. But it should catch the BIG problems over 90% of the time. More and more importers are disillusioned about the value of QC inspections. Many of them think (and, depending on who they work with, they are not necessarily wrong) that way, way less than 90% of BIG problems are detected. And they usually think that bribery is the main reason for this lack of effectiveness. I dont agree. I listed below the major reasons why Chinese inspectors sometimes fail to notice quality problems that are BIG in their clients eyes. 1. Laziness and lack of discipline Most professional inspectors are guilty of this.

For example, they dont pick cartons from every side and every height of the pile, to save time but also to avoid bothering the factory. There are actually 2 separate issues here: 1.1 They take shortcuts to go faster

Page7

A very common example: during the visual check or the testing, they often dont check as many samples as they should. As a result, their report is based on their guesses after checking a sample size that is much smaller than their QC plan calls for. The statistical plan is not respected, and in the worst cases the report means nothing at all. 1.2 They are unwilling to bother the factory Many of them are tired of fighting. They want to work in a relaxed atmosphere, have a nice lunch, and avoid justifying themselves after complaints from the supplier. For instance, they let the factory check (and repair!) the samples by themselves during the unpacking process. They pretend not to see this little game. And, of course, many problems are corrected before inspection. Is this bribery? Yes and no. More about this phenomenon here. 2. Lack of training When they need to check garments, some quality control agencies see no problem in sending inspectors who have a very superficial knowledge of textile products. Some sophisticated clients can suspect it after reading the QC report, but it is difficult to prove. These inspectors are very slow in taking measurements, so they ask the factory to measure and they write what they are told. Oh, and there are many issues that their untrained eye just cant notice. This is just an example. The same thing happens with electronics, ceramics, furniture, etc. 3. Wrong calibration As I wrote last week, QC Inspectors cannot follow different quality standards. Send an inspector who is used to checking cheap products for monitoring a production of goods to be sold in high-end boutiques, and what will he conclude? Everything is fine. The reason is, it is so much nicer to what he usually sees. Unfortunately, the high-end buyer might see defects that are important to him and that were completely undetected during the inspection. If you work with QC firms that work for cheap buyers (promotional goods, mass markets in South America), dont ask them to check high-quality products. 4. Poor timing of inspection Many buyers dont understand why ONLY a final inspection (after production is over) can confirm the average quality of a batch. When everything is packed, the quantity can be counted, and products can be picked randomly. The factory cant play games. What often happens is that suppliers are late. The job gets done when only 20% of the goods are completed, and maybe 1 or 2 cartons are packed. It is easy for a factory to hide the defective products and claim they are not made yet. What to do in this case? A re-inspection at the suppliers charge. Naturally, most Chinese suppliers pretend not to understand why they should pay for a re-inspection in this case, since the job got done.

So you should think now about the instructions to your QC agency/staff: should they abort the mission in such cases? 5. Rushed job

When an inspector is under intense timing pressure, he is forced to take all the shortcuts he can. There are actually 3 separate issues:

Page7

5.1 The inspectors manager decides on an unrealistic sampling plan This happens when the client is allowed to choose the sampling size in an online booking system, or when the QC company does not dare to say no to a client. It happens often in peak periods, like before Chinese New Year. I have seen sampling plans including 8 different product references to check for conformity, and a total of 315 pieces to check visually, in 1 man-day! The result is the same as in point 1.1: the job is not done fully, and the report means nothing. 5.2 The inspector himself compresses the time of the job After a few rushed jobs (see point 5.1), inspectors know they can get done in 2 hours. Some of them conclude that they can do one job in the morning (for their employer) and one in the afternoon (on a freelance basis). This is a highly dangerous habit! 6. Corruption Yes, corruption happens. Especially in situations with the following characteristics: The sums at stake are quite large, The importer puts a lot of pressure on the supplier to ship in time, When there is a problem that is acceptable (according to the inspectors judgment), but is not in conformity with the clients specifications. I already described the way corruption works in a previous article. The point is, it is not as frequent as most people think. 7. Unclear or incomplete specifications If you dont communicate detailed requirements, and if you dont ensure that the inspector has a reference sample in hand, you are asking for trouble. This may seem obvious. But tens (hundreds?) of unprofessional importers commit this mistake. In the China sourcing game, letting another party guessing some information is never a good idea. *** Oh, and how about factory interference? I am not including it in this list, because an inspector who doesnt cut corners (see point 1.1 above) will avoid it. And, if it cant be avoided, the job should be aborted. What do you think? Am I forgetting an important element? Procedure if a problem is detected during an in-process inspection Two years ago, I drew a flow chart to guide importers in case a final QC inspection was failed. But the best is to detect problems during an in-process inspection (before the products are finished). At this stage, if an issue is caught by an inspector, there are 2 objectives: To make sure the same problem wont be present on the pieces that will go through that same process later;

To make sure the pieces that already bear this problem are dealt with in the right way. So I drew another flow chart, specifically for in-process inspections:

Page7
What do you think? Anything to add to this procedure? Procedure for dealing with failed inspections Quality inspections are only a way to collect information, and present it in a way that allows the buyer to take a decision. Last month I wrote a post about the case where the buyer refuses the products and asks for rework. In that article (re-inspections and corrective actions), I explained what to do: Analyze the situation and send clear guidelines to the supplier,

If necessary, change the scope of the inspection. But I keep explaining this over and over to my clients. So I drew a flow chart that can be used as a procedure for dealing with failed inspections:

Page7
A few notes, for using this flow chart with highest effectiveness: The buyer should anticipate all these steps (can it be repaired? if not, what to do?), and then communicate with the supplier accordingly. In the best case scenario, the supplier communicates the percentages of (1) the pieces that are classified as fine after the sorting job, and (2) the proportion of pieces that are not acceptable but that can be reworked. For very sensitive production runs, the buyer can send an inspector to check the reliability of the sorting job and the effectiveness of the repair job. The factorys incentive is to sort out quickly the biggest issues and repair them grossly to be a little closer to the standard (cha bu duo). The worst for them is to short-ship or reproduce because it means materials were wasted (and it generally costs much more than labor). Why we shouldnt check 10% of the order quantity Some of my new clients, when they book their first inspection, tell me please check 10% of the order quantity. Here is what I typically respond:

Usually we follow the statistical tables that everybody in the industry works with. They were developed specifically for quality control, and they have become a standard that even your suppliers have heard of. Nearly 100% of importers follow this standard. But we certainly can apply your inspection plan if you explain it to me. If we randomly select 10% of the order quantity for checking, what is the maximum number of major and minor defects that you will accept? Generally this is enough to convince them to follow the relevant standards. But some of them actually stick to their guns, and give me an upper limit on the number of defects. In this case, here is what I say:

Page7

We can follow your suggestion. But in case the inspection is failed, it will be easy for your supplier to question the scientific validity of the inspection plan. The maximum number of defects you gave me does not take into account the risk that the whole quantity is acceptable while the samples we have selected are worse than the average. Neither have you considered the risk of the opposite happening (accepting a lot because of samples above average). You would need to choose a confidence level, calculate the impact of both risks, and come up with limits on the numbers of defects. Then they realize they are better off not reinventing the wheel. We are also better off, as a new inspection plan creates confusion and can lead inspectors to make mistakes. Another classic is trying to optimize the work done. Lets say the product is rather simple, and an inspector can check up to 315 samples of in one day. The order quantity is 3,000 pcs, so even a tightened-level inspection would require us to draw only 200 samples. A buyer willing to optimize the inspectors job is likely to ask us to check 315 samples anyway. Once again, the statistical tables dont allow for this situation. For 90% of consumer goods (excluding food) sourced in China, following the normal inspection level is fine. Instead of inventing their own procedure, importers would be better inspired to spend more time communicating with their suppliers and preventing mistakes.