To inspect or not to inspect?

A tale about quality

The following conversation occurred recently while working with a manufacturing client.  I had observed that they were performing inspection and significant testing of the entire batch of a particular part supplied by a 3rd party organisation.  The part in question was a wiring loom, a standard part they use in some quantity. The conversation went along the following lines:

“Why do you inspect them?”

“The part is used in a safety product that we make. Because it’s a safety product, we have to make 100% sure that the part is OK, so we test them all.”

“Does your supplier test them?”

“Not sure, but we wouldn’t trust their tests anyway; it’s too important.”

“How many fail?”

“Not many, but some have in the past.”

“Are the defects the same issues or variable?”

“I can’t recall.”

“Do you keep records?”

“There is an NCR system, but no one uses it.”

“Why not?”

“Management doesn’t do anything with it. It’s a waste of time.”

“How do you test the part?”

“We get a set of parts and connect it up to make sure it does what it’s supposed to.”

“Is that on a standard test jig?”

“No, we just set something up as required.”

“Is there a standard test procedure?”

“No, we know what it’s supposed to do and just make sure it works OK.”

I’m sure you can already see that there are a number of problems here, and, while these conversations can be a little difficult for the client, they can be very informative. Debates with clients about the importance and necessity of inspection are commonplace, largely because I specifically ask about inspection regimes to learn about the maturity of their quality systems. In most cases, I uncover a lot of misconceptions and misapplications of quality approaches. In this case, I empathise wholeheartedly with the sentiment of wanting to assure a high standard of reliability in their product, especially given the safety nature of its application, however, their approach of 100% inspection will ultimately fail to live up to expectations.

Let’s go to some first principles about quality, and for this, I am unashamedly going to draw upon W E Deming, one of the pioneers of quality management, who was highly critical of reliance on inspection as a method for ensuring quality. Deming argued that inspection is reactive and does not add value because it only detects defects after they have already been produced. Instead, Deming advocated for prevention through process control and continuous improvement.

Some Key Points Deming made about Inspection:

1. “100% inspection is ineffective" I’ve put this first as I think it is the key point. Even rigorous inspections will miss defects! This is especially so for complex or high-volume production. So, if defects exist, some will inevitably go through to customers. Because of this, 100% inspection regimes will never achieve what they are intended to do. Typically, if defects are found, it is only the defects that are known to occur that are discovered. New and unknown defects, even if obvious in hindsight, can easily be missed.
2. "Cease dependence on mass inspection" (Point 3 of Deming's 14 Points) Deming argued that quality should be built into the process rather than checked at the end and that relying on final inspection leads to waste because defective products have already incurred costs in materials, labour, and time.
3. "Inspection does not improve quality" Deming described inspection as merely sorting good products from bad, rather than eliminating the causes of defects. Instead of inspection, organisations should focus on statistical process control (SPC) to monitor quality during production.
4. "Quality comes not from inspection but from improvement of the process and replacing inspection with quality control and process improvement" Deming emphasised understanding variation and using statistical methods to reduce defects at the source. Workers should be empowered to identify and fix problems early, rather than relying on inspectors at the end of production. Organizations should use control charts, root cause analysis, and worker involvement to prevent defects rather than detect them after the fact. Quality improvement should be seen as a business-wide responsibility, not just a function of quality control departments.

All of this is not to say immediately cease your inspections. If you have products that you believe include defects, you need to put some gates in place to minimise the amount going to customers (remembering, though, that you won’t catch them all). However, you must have a longer-term plan to reduce your dependency on inspection gates by implementing alternative approaches. Once confidence in product quality is reassured, you can reduce or eliminate the inspections. The alternatives to inspection include:

• Using statistical sampling instead of 100% inspection.
• Focussing on upstream process improvements rather than catching defects at the end.
• Train workers in quality management principles so they can address problems proactively.
• Implement a continuous improvement culture (Kaizen) instead of relying on quality checks.

Deming's core message was that quality must be designed into the process, not inspected in. He viewed excessive reliance on inspection as a symptom of poor process control and encouraged businesses to focus on systemic improvements that prevent defects from occurring in the first place. For my client, I recommended to first engage with their supplier to find out what testing was happening and how it was done. Quality problems must be controlled at their source, so it makes sense to work with the supplier to ensure that their processes can guarantee quality at the required level. They are, after all, specialists in making their products and should be accountable for their quality. Some sampling might still be done while confidence in quality is regained, with the sampling rate reduced as confidence allows. I also recommended as a general principle that the NCR system be re-invigorated and used with discipline, with reports back to team on outcomes. Recording defects and analysing this data is where quality starts!