One of the most often witnessed symptoms of Lean Management that I come across nowadays are whiteboards with performance indicators. These serve as one of the central elements in weekly (or when I’m lucky, daily) huddles that are so fashionable nowadays. Now of course a lot of good things can actually be said about these huddles and indicators when done correctly. However, there are some risks in using these type of representations as well, as for instance Mark Graban rightfully wrote about last August in his blog when discussing the use of charts. You can also read more on correctly interpreting data on charts using Statistical Process Control (SPC) on my own blog, while writing on climate change. But even if applied correctly, I think there is a more structural issue with concentrating too much or even solely on these type of charts.
First of all, most of these charts show us the voice-of-the-process (VOP). And surely, that is important to see whether we are dealing with a process that is stable or not. SPC charts help us understand the process, and when used well, they can help teams get their process under control. So it helps maintaining the current process. But it doesn’t necessarily show us what the performance is that the customer is experiencing. As performance is the voice-of-the-process compared to the voice-of-the-customer (VOC). More concretely, it is the data on the chart against the specification of the customer. And as Mark wrote, often this leads to a chart that indicates a stable process but that could be structurally in the red. SPC therefore is about normal and abnormal, not necessarily about good or bad or about OK or NOK. And weren’t we trying to improve?
Control Charts are Only “Mirrors”
Also, when we focus too much on SPC and the control limits in a chart, we are tempted to be happy with a stable process and accept the fact that the system – although stable – still generates defects. We consider it normal, but we should never forget that a defect is still a defect and the process, though stable, is simply not performing. Shingo saw control charts only as “mirrors” reflecting prevailing conditions. And he asked the question whether it is unreasonable and unjustifiable to want zero defects? I think it isn’t.
Furthermore, we should also understand that even when we use charts to plot our performance, we plot past performance. The charts show us what has happened, after it happened. And the less frequently we update these charts, the longer ago it even was that this performance was actually realized. This time lag is a serious drawback for its effectiveness in reducing defects. And sure, we can increase the frequency with which we update the charts, but that doesn’t change the fact that it will still be after problems actually occur. In that sense, the charts can be seen as merely inspection and as post mortems. When studied properly they can of course provide hints for process improvement, but that takes time and effort. But basically, they do not prevent defects; they don’t actually build quality into the process.
Another disadvantage of these type of charts is that they summarize data. They don’t show individual products or parts, deliveries, clients or patients. They show the average of a sum of parts, deliveries, clients over the period of time which is like a sample. Now that may come in handy in reducing the effort related to our quality control method, but it doesn’t really help in attaining our ultimate goal of zero defects.
Zero Defects is Not Impossible.
So what should and can we do?
It all starts with mindset. Embrace the idea that zero defects is not impossible. Stop accepting defects. Stop accepting products with certain restrictions. Stop thinking that as long as a product characteristic is within a certain acceptable tolerance, there is no consequence for the subsequent process or customer. Stop thinking in samples and probabilities, but instead think about individual parts, products, deliveries, customers and patients. We need breakthrough improvement, and think carefully about whether the elimination of special cause variation to stabilize a process should be called improvement or just what it is, viz. stabilization. We should start thinking like Shingo and Taguchi. Zero defects is possible, it just isn’t easy.
Zero defects requires us to move towards prevention. We should therefore never lose sight of the fact that defects are the result of our operating conditions and actions. They are the result of men using materials and machines in certain ways (i.e., methods) at a certain location and a certain time. Organizations should therefore strive to implement low-cost, 100% source inspection of all relevant conditions with immediate feedback, before jobs are started and lead to defects. I have written about this before on my blog when I wrote about the future of the quality department.
Well-known techniques in applying the principle of source inspection are of course checklists, visual controls, andons and error-proofing devices (poka yoke). And many application examples exist both on part, jig, machine and process level. They all strive for the same thing, viz., to uncover errors in operating conditions before a job is started. But I have no doubt that applying source inspection will require a great deal of learning about how our work actually works, creativity in teams to get to counter measures using the principles of source inspection and perseverance to see it through until zero defects is reached.
But in the end, zero defects is possible. And to demand zero defects is justifiable. Using control charts and Statistical Process Control may help on the road to zero defects, but it will not be sufficient. Performance charts and daily huddles therefore can only be considered a very small beginning of your Lean journey, but certainly not the end. I hope this will not frustrate you, but instead inspire you to look ahead and to start your journey towards source inspection and zero defects!