Organizational Performance Part 39: Understanding the Accuracy of 100% Inspection | Operational Excellence Quick Hits

Quick Hits share weekly tips and techniques on topics related to Operational Excellence. This week’s theme relates to understanding the accuracy of inspection. We hope you enjoy the information presented!

, Organizational Performance Part 39: Understanding the Accuracy of 100% Inspection | Operational Excellence Quick Hits, Future State Engineering
, Organizational Performance Part 39: Understanding the Accuracy of 100% Inspection | Operational Excellence Quick Hits, Future State Engineering

Speaker 1: (00:05)
In today’s session, we’re going to continue on the series of establishing the right mindset within an organization and continue to challenge the underlying assumptions about different concepts or beliefs in organizations. Today’s session, we’re going to talk about 100% inspection. So a lot of times when I go into companies, they think 100% inspection is 100% accurate, so I want to challenge that belief. And first of all, let’s define what is 100% inspection. So, it means that every item is inspected. In other words, all elements or features of a particular item are compared against a predetermined specification of defined requirements. So it doesn’t matter what it is. It could be a piece of paper, it could be an application, it could be a product. It could be a service where we’re looking at, was the service provided as expected? So a lot of companies you inspection as a means to determine if the process was being performed correctly.

Speaker 1: (01:09)
If an item meets all the specifications of the defined requirements, then the item is acceptable. And then, if the item being inspected fails to meet one of the specifications of the defined requirements, then the item is rejected. And again, we’re looking at the requirements from the customer’s perspective.

Speaker 1: (01:31)
When we look at inspection, there can be two types of errors. So I want to describe these two types of errors so people are familiar with the types of errors that can happen in an inspection process. So the first is what we call type one error. And what that means is, accepting a defective item as a good item. So this type of error increases the risk that the customer will receive a defective item. The other type of error is what we call type two error. This is rejecting a good item as a defective item. So the product is actually good and the service is actually good, but it gets rejected, okay? So this type of error increases the cost of quality within an organization. So typically, these two types of errors are the errors that we’re looking for when we look at 100% inspection.

Speaker 1: (02:25)
Always we speak with data, so I go into companies and I ask them how accurate is 100% inspection and we do a little exercise. And the exercise is to demonstrate how accurate is 100% inspection. So I have this little game or this little example where I have a story and the people need to read through the story one time, so it’s 100% inspection. So one time through the story. And what they’re looking for is the letter G, the letter G is a defect, okay. And what they need to do is count the number of defects in this story. So if I just do a quick little example here, “while strolling,” and if I look at strolling, there’s one G at the end of strolling, so that’s my first defect. And then I look through and there’s another G. So what I’m going to do, is I’m just going through the story one time and counting the number of Gs.

Speaker 1: (03:20)
And then what I do, is I collect that data at the end of the session, then I give them a few minutes to go through and do the analysis. And this is the data that I get from it. And so, you see that I got variation. So we have different people that are doing the exercise and we get different results, anywhere from 72 to 83. So, first of all, I started to see that there’s variation just in the inspection process. And when we look at it, we have a range of number of defects we find, and we can take the average of this. And so when I look at the data, the average number of defects found was 77.2. The actual number of defects in the story is 83. So there’s only one person that actually found all the defects from our sample. And then, so I can calculate the inspection accuracy by taking the average and dividing by the total number of defects. And so in this case, the inspection was only 93% accurate. So that tells me that 7% defects are going to get through the system in this example where we’re using inspection-based quality.

Speaker 1: (04:41)
I also want to point out a couple things here also, is that in this simple example, the following conditions happen, which are not usually likely in most organizations. One, there’s only one characteristic being inspected. So it’s just, we’re looking for the letter G and there’s no variability, it’s either a G or it’s not a G. And third, there’s no disruption happening during the inspection process. So it’s not like you start and then get interrupted, and have to go back and continue inspecting. 93% accuracy when there’s zero variability in the decision criteria, there’s no interruptions and there’s only one characteristic. In reality, there’s typically multiple characteristics and there’s usually some variation in the decision making. So if I’m looking at a heart and I’m looking at a scratch, or I’m looking at a bur, or I’m looking at a crack, or I’m looking at a discoloration there’s variation in the decision making alone. We’re not even considering that in this example.

Speaker 1: (05:51)
So what we see is 100% inspection is not a hundred percent accurate. Depending on the situation, it can be anywhere from 80% accurate up to 93% accurate in this case. To get better results using inspection-based quality, you have to start doing 200%, 300%, 400% inspection to get 100% accurate. And of course doing those number of inspections is just going to drive cost up.

Speaker 1: (06:22)
That is our challenge for this week, 100% inspection is not 100% accurate. In next week’s session we’re going to talk about a resource standing idle, and if we have an idle resource, that’s a major waste.