The Data Heads

Quality Magazine picked up my blog post on the Voice of the Process loop. I’m really interested in your comments on the questions posed in the post, either at the Quality Mag site, or here.

In a recent post I described the difference between the Voice of the Customer (VOC) and the Voice of the Process (VOP). I used a simple Input => Process => Output diagram to explain that both the VOC and VOP have an important role in evaluating our systems and processes. I said that it isn’t a case of one being right and the other wrong, or one being better or worse than the other. Both approaches have their place. But they are different.

It seems to me that the biggest issue comes when we confuse the two approaches.

I suspect someone is confusing the two approaches when they point to a control chart with subgrouped data plotted on it and say something like:

“When I see a point out near our limits then I know we’re getting close to going out of spec.”

On the surface this seems like a reasonable thing to say. And it must be reasonable because I hear it so often. So why do I get all worked up when I hear it?

(Because I do get worked up. I have a hard time not jumping up and down and raising my voice. )

“When I see a point out near our limits…”

The issue is that the point is the average of the data in a subgroup.

Lets look at a picture:

We’re looking at three subgroups of data.

What is your vote?
Specs are 25 plus or minus 5, we have no problem, right?
Specs are 25 plus or minus 1, we should worry, right?

Here is the actual data:
Sample A: 24.9, 25.0, 25.1
Sample B: 20.0, 25.0, 30.0
Sample C: 20.0, 20.5, 33.0

And here is what it looks like plotted on the same run chart:

Plotting subgroup averages on an X-Bar chart and then comparing those average points against the specifications is misleading. You’re blurring the use of VOC and VOP. Don’t stop using either loop – but don’t try to use them at the same time.

Have you ever seen the VOC and VOP loops confused? What issues has it raised for you? Use the ShareThis button below to mark this page, leave a comment, tweet me, schedule a conversation, or call 800-958-2709.

In the weeks leading up to American Thanksgiving and the rest of the holiday season, I found myself buried in a couple of projects related to the release of GainSeeker Suite, Version 8; and a course in Appreciative Inquiry. In the process, I got out of the blogging groove. (Version 8 is pretty cool. It includes some brand new modules that will help you get more value from your data, and make it even easier to use. More on the new version in subsequent posts. Likewise, I’ll write more about my interest in Appreciative Inquiry in the weeks and months to come.)

Now my editor at Quality Magazine is prodding me, so it is about time to get back in the rhythm, even though I still have some tasks on Version 8.

My last post at Quality Magazine on ‘Defining Quality’ (which was a revision of ‘What are they thinking…?’ published here) generated a number of thoughtful comments from readers. I really appreciate hearing other perspectives on the questions I posed in the post, especially this comment:

Only in the world of utopia are there processes with zero variation and where only Green or on target values are produced. In the real world you have to look at each process and determine the ability and cost required to reduce process variability. In some cases it may be more cost effective to use an inspection system (like a vision system) to inspect out defects, then it would be to reduce the variation that creates these defects. The rule I use is if prevention is not practical and if detection methods are effective and reliable, then the inspection method is the right choice. When detection is difficult or not reliable, then prevention efforts must be taken.

This got me thinking about a model for inspection that I’ve found helpful in recent years. Here it is:

Bear with me because this graphic still needs some explanation.

If you look at this carefully, you’ll see that it is a typical ‘Input – Process – Output’ diagram. If you look at it even more carefully you’ll see that it is there are two loops on the graphic. The question isn’t so much which loop is right and which is wrong. The question is: Which loop is primary? Which loop is emphasized?

Both loops start in the middle with an observation or a measurement. The right-hand loop is the Voice of the Customer. I’ve highlighted it here in yellow:

The right-hand loop compares the observation to the customer’s requirements and asks “Does this meet the customer’s requirements?” If it passes, you can ship the product. If it doesn’t, you have a couple of options.

If your product fails to meet specs in a manufacturing environment, your options are to Scrap, Rework or Downgrade the product. In a transactional world, your options are to Remediate, Replace, or Compensate. In either situation your options for response are always reactive and wasteful.

I see people tolerate this waste for all kinds of reasons. Perhaps there are other, more expensive issues that need to be addressed before this problem can be tackled. Perhaps the cost of getting rid of a problem seems too high. Perhaps they’re just used to it and can’t imagine any other way of doing business. Some of these reasons are probably better than others, and I’m really not here to pass judgment in this blog. The point that I want to be clear about is that the right hand loop – the Voice of the Customer Loop – captures waste and protects the customer. There is nothing wrong with that (actually there are some good things about it). But it doesn’t prevent the problems from recurring.

The left hand loop starts at the same place, but has a very different impact. This is the Voice of the Process Loop, highlighted here in green:

The Voice of the Process Loop also requires an observation or a measurement, but here is the crucial difference. Where the Voice of Customer Loop compared the observation against specifications, the Voice of the Process Loop compares the observation against what is expected.

On what do we base our expectations? Well you can guess that it isn’t a specification – or anything that is derived from a specification or a requirement.

We base our expectations on our past experience with the process. This is why we call it the Voice of the Process. The best way to tap into our past experience with the process is with the humble control chart.

The control chart tells us what we need to know about the process. If the data we observe shows no patterns, no shifts in mean, and no more variation than we’ve experienced before, then we have reason to conclude that the process is stable. Once we know the process is stable, then we can still ask ourselves “is there a way to improve (reduce) chronic variation? This can lead to improvements in the process or the inputs to the process.

If the process tells us that it isn’t stable, then we can (and should) address that. We can focus our efforts on improving the process or the inputs to the process.

Using the left-hand loop – the Voice of the Process Loop – is how you improve processes and ultimately reduce or even eliminate the need for the Voice of Customer Loop. In an ideal world our processes are well understood and stable, and we don’t need to check against specs because we know that we’ll always meet customer requirements.

In the meantime, we live in the real world. In the real world, inspections against specifications are a reality and will probably be around for a long time. They’re useful and I would be the last to advocate their complete elimination.

But they don’t lead to process or quality improvement, or to an elimination of the waste associated with failure to meet requirements. To get there you need to pay attention to the Voice of the Process. You need to stabilize your processes and then systematically reduce chronic variation.

Where is your emphasis? Which loop do you follow? What are the biggest challenges you face or have faced in shifting your emphasis to the Voice of the Process Loop? Use the ShareThis button below to mark this page, leave a comment, tweet me, schedule a conversation, or call 800-958-2709.

Last week a colleague sent me a link to a new white paper that you should take a look at. It is published by SAP and titled “Reaping the benefits of next generation dashboards.” You can download your own copy from The Dashboard Spy.

The white paper describes the problems it sees with current dashboard and business intelligence solutions (they are inflexible and too cumbersome to use). And it offers a punch list of features for what it describes as the “Next Generation Dashboard”. Here is the list:

Next Generation Dashboards must:

• Be easy to build and customize
• Provide a consolidated view from any data source
• Leverage visualization to make information easy to consume
• Offer engaging interactivity for further analysis
• Provide the information in a personalized and easy to understand format
• Allow developers to extend new features or integrate to new technology

The white paper concludes with a list of the benefits users can expect to see from these next generation dashboards.

As I read the report I wondered how this vision of the Next Generation Dashboard matched our vision of the data driven organization as defined by the Data Cost / Value Matrix. (If you haven’t already taken the Free On-line Gap Analysis you might want to do that before you read more. It only takes a few minutes.)

The Data Cost / Value Matrix  identifies four aspects of Data Cost and four aspects of Data Value. You can read more about this at the background page.

Let’s take the four aspects of Data Costs and see how the white paper approaches them:

Complete: We collect all the data we need, and no more than is necessary.

The white paper seems to begin with the assumption that we have all the data that we need, and that all data are good, reliable, and necessary.

My experience is that most organizations are smothered in data. Typically it is the wrong data. All too often organizations focus their attention on the data they CAN get, and do not spend enough energy on the data they SHOULD get.

When we make the wrong data more actionable we have gained nothing.

I think the Six Sigma Master Black Belt described in this case study from a financial services firm was right on track when she engaged in manual data collection first because she “was able to gain valuable insight into the nuances of the various operational definitions used by the process owners, and in the way the information system supported or did not support those definitions.”

This white paper overlooks this issue.

Automated: We write down very little data. In fact, we type very little data into computer systems. We use bar codes, RFID or other identification technologies. We capture data from digital equipment whenever possible. Wherever possible we have eliminated human interaction with data collection, and we are confident through data driven statistically valid measurement system analysis that the data are reliable.

The white paper is very strong on the first part of this because of its emphasis on the integration and interconnectivity of data systems. At the same time, it seems unaware of Measurement Systems Analysis and the contribution it should make to this process. It may be the MSA is too technical and therefore beyond the scope of this kind of white paper. However, the world envisioned by the white paper – where everything is fully automated – overlooks the premise that we need to be thoughtful about our data.

One of my favorite business quotes is by Peter Drucker: “Nothing is worse than making more efficient what should not be done at all.”

Integrated: We have specialized data systems to run various aspects of our business, but we don’t have silos of data that are used for only one purpose when the data can be useful to other applications. Put another way, data is never entered more than one time anywhere in our business.

This is one area where the vision of the next generation dashboard is in close alignment with the Cost / Value Matrix.

Accessible: Anyone can get to the data they need at any time. We don’t have to rely on specialists to write special queries or export data. We’ve learned that our people do not need to be programmers to make good use of data.

Clearly the white paper is aligned with this aspect of reducing the cost of data. This comes up several times in the article, as in this quote: “the next generation of dashboards empower non-IT professionals to design and connect business data to a dashboard interface.”

So the white paper endorses integration, accessibility, and automation. It seems to fall short on the issues of completeness and data reliability.

Lets turn out attention to the four aspects of Data Value.

Product Release & Control: We use data to validate that our products are acceptable for shipment. This data is primarily accept/reject type data, and may be based on either measurements or some other kind of pass/fail criteria. The pass/fail criteria is based on the voice of the customer.

The white paper jumps on this with both feet. Under the heading “Leverage visualization to make information easy to consume” it suggests a product release and control strategy as one of the fundamental ways users should consume information: “In addition to robust data visualization, next generation dashboards provide methods to visually alert a user when performance indicators are out of tolerance, then enable the exploration of details with point and click simplicity.”

This is a great example of a Product Release & Control approach to the world: Test a result against the tolerance ( tolerance = specifications = Voice of Customer) and alert the user when something fails. Clearly this is a huge advantage to companies to get this kind of information – especially if it is provided in real time so that prompt corrective action can be implemented.

Process Control: We apply statistical process control tests to key products and processes. These activities use the Voice of the Process to determine the stability of our process. We react immediately to instability and unexpected variation.

On the issue of Process Control, the white paper falls completely silent. There is no indication that the authors understand this critical point of delivering value with data.

It may be that this is an oversight. More likely it is a point of value that is not appreciated by the authors. Making this a point of value assumes that the user understands the difference between Voice of Process and Voice of Customer. In my experience this distinction is not commonly understood. Even people who have been through Six Sigma training or who are certified quality engineers sometimes confuse the issue. We readily understand “outside the specs”. We’re far less likely to embrace or distinguish the more abstract “out of control”.

This is an important short-coming in this vision of the Next Generation Dashboard.

Continuous Process Improvement: We use data to close the loop on our processes and drive continuous improvement. All of our people are trained to use this data to look for hidden sources of variation and correlation between key input and key output variables.

According to the white paper, the fundamental benefit of implementing Next Generation Dashboards is to improve processes:

Notably, C-level executives use business intelligence to improve processes, ensure compliance, optimize marketing efforts, and increase sales. And department managers can use the information to improve their operations and monitor the performance of their groups.

In another section of the white paper, the authors note that Next Generation Dashboards should “enable the exploration of details with point and click simplicity.”

Clearly these are signs of commitment to continuous process improvement. I wonder, however, if the proliferation of dashboard tools will be matched with a similar effort to make sure people use the data in meaningful ways.

My mother spent her career teaching fourth grade. I’m old enough to remember the alarm bells she rang when pocket calculators were first introduced: “But will these kids actually understand the answers they’re coming up with? Or will they just get wrong answers faster and assume that they’re right because a computer spat it out at them?”  The older I get, the more I see what she warned against. This situation is a grown up version of the same problem.

Data Visibility & Transparency: Our data is readily visible at all levels of the organization. Every stakeholder, from process owners to the leadership team, can put their fingers on the performance data that matters to them. Information is summarized in easy-to-understand dashboards that help them separate signal from noise so they don’t react to the wrong things. Furthermore, they can readily get to the underlying data to better understand the drivers of their key metrics.

This is actually a pretty good summary to the Next Generation Dashboard White Paper. Clearly the authors “get” this vision.

In summary, the white paper is in alignment on many aspects of the Data Cost / Value Matrix. There are a few points where it falls short. Most notably if falls short in its vision of the importance (or the real cost) of complete and accurate data, and the value it places on the Voice of Process.

The fundamental assumption of this white paper is that business processes can be characterized by data. If we set aside the (very important) questions about the reliability and repeatability of data for just a minute, the question that comes to the forefront is “what theory shall we apply to the data that characterizes this business process?” Or, in the words of Dr. Deming, “By what method?” shall we reach our goals?

If we fail to attend to the Voice of the Process, our efforts will certainly be suboptimized. GainSeeker Suite and GainSeeker’s Enterprise Dashboard implement all of the requirements outlined by the white paper for the Next Generation Dashboard and they make it very easy to pay attention to the voice of the process.

What do you think? How important is the Voice of the Process in your dashboard? What are you doing today to build dashboards for your business?  Use the ShareThis button below to mark this page, leave a comment, tweet me, schedule a conversation, or call 800-958-2709.

If you’re on our mailing list you’ll get an email from me later today that invites you take a survey about your use of social media. If you’re not on our mailing list, consider this your invitation to take the survey now.  Here is what I wrote to my customers:

In the last few years social media has exploded on the web. Perhaps you already are Linked-In and Twittering – or maybe you’re on Facebook, and bookmarking websites with Delicious.

For the last several months I’ve been dabbling in blogs and twitter.  I can see some great potential for how we communicate with and support our customers, and how we might help our customers connect with each other. Together we can solve more problems and make the world – at least our little corners of it – a better place.

But I need a reality check. Being a data guy, I’d like to get some hard data about what you’re doing with social media today. I invite you click on this link and take a short survey. I’ll use this data to see if we can create better ways to share and discuss information among our customers. Here is the survey.

Once you’ve completed the survey, I invite you to join one or more of my networks. If you’re already on Linked-In, please review my profile and join my personal network using my email address (ejmiller at hertzler dot com). You can also sign up to follow me on twitter. Finally, you can subscribe to my blog in a reader through Feedburner.

Of course I’ll publish the results in my blog, so you can stay tuned there.

Thanks in advance for sharing your thoughts.

Interested in the survey and the results? Use the ShareThis button below to mark this page, or leave a comment, tweet me, schedule a conversation, or call 800-958-2709.