[FREE PREVIEW: Go-Getter Membership Webinar] Did you know that a Control Chart is just a Line Chart that went to college? But it’s that extra bit of education that makes the difference.
Using Control Charts can help you understand the trends and shifts in your process performance. They can help you avoid wasting time overreacting to single data points. Modern organizational information systems can leave us data rich and information poor, but Control Charts can put you back in the data driver’s seat. These charts are coming up on their 100th birthday so come celebrate and add another powerful tool to your toolkit!
- Intermediate [Webinar recording will be available for Go-Getter Members]
In this 1-hour webinar, we will cover the following:
- What are Control Charts?
- Why should you use them?
- How do you put them to good use?
- What are some good examples?
- What are the tips and tricks of Control Charts?
Tracy O’Rourke: Hi, everyone. Welcome to GoLeanSixSigma.com’s webinar. And thanks for spending some quality time with us today. Almost six hundred people have registered for this webinar. I never realized how many people really want to know about control charts. So we are really excited that you’re here.
Lean and Six Sigma are the go-to improvement methods used by leading organizations all over the world to delight customers, minimize cost and develop better teams. And every month that we craft webinars just for you so Elisabeth, good job on picking this because obviously people are interested in it. And we love simplifying concepts and tools for Lean Six Sigma, of Lean Six Sigma, so you can understand and apply them more easily and be more successful.
About Our Presenter
So our webinar for today, How to Use Control Charts is going to be presented by Elisabeth Swan. I’m Tracy O’Rourke. I’m a Managing Partner at GoLeanSixSigma.com. And today, Elisabeth, she is also a Managing Partner at GoLeanSixSigma.com, my colleague, the wonderfully talented, innovative, and consummately passionate about learning, Elisabeth.
How are you today, Elisabeth?
Elisabeth Swan: I’m awesome, Tracy. You always make me feel good with your introductions.[Laughter]
Tracy O’Rourke: Just a little background on Elisabeth. She is also my co-host for the Just-In-Time Café Podcast and co-author of The Problem-Solver’s Toolkit, which is our book that came out last year. She is also a long-time Lean Six Sigma Consultant, Master Black Belt, Coach, and Trainer for over 25 years. Aside from almost three decades in the business, Elisabeth has also performed with Improv Boston. So you know that she makes stuff funny too. Since half of her process improvement is in improvisation, that’s a great combo.
So welcome, Elisabeth.
Elisabeth Swan: Thanks, Tracy. I usually keep the improv for on stage. I’m going to stick to control charts today.[Laughter]
How to Interact
Tracy O’Rourke: OK. Good. So here are a few housekeeping notes before we begin. At the end of the presentation, we will have question-and-answer session where Elisabeth can answer questions for you about this webinar but please feel free to ask any questions at any time and just type them into the chat window.
We also welcome you to participate. So we’re going to ask you to vote with some polls. And if we don’t answer all of your questions during the webinar, all of the answers will get posted in the Go-Getter Membership forum. A copy of the slides will also be available in the respective webinar post on our website. It’s free for introductory webinars and it is exclusive for Go-Getter members for intermediate, advanced, and leadership webinars. So if you’d like to become a member of the Go-Getter Membership, we will share more information with you about that at the end of the webinar.
Where Are You From?
Please join us for our first activity sharing where are you from? So we’ve got hundreds of people registered for this webinar as I said earlier from all of over the world and we would love to share where are you from? Where are you located? So, please locate the chat section and type in where you’re joining us from today.
Geez! Wow! Look at that! OK. So wow! This is – I can’t even read them. They are moving so fast. OK. We got Malta, Vancouver, British Columbia, Rome, Georgia, Hungary, San Diego, California. I wonder who that is. We have a lot of people from San Diego. Upstate New York, Ivory Coast, Africa, Advance, North Carolina, Hollywood, Florida, Peru, Nashville, Warwick, UK.
We have lots of people and lots of places, Elisabeth, excited to hear about control charts because they are such a great tool. So thank you everyone for joining us today. I’m going to hand it over to Elisabeth.
Our Mission and Core Values
Elisabeth Swan: Thank you, Tracy. So yeah, thank you guys. And I’m happy to be here with all of you from all over the world. Thanks for joining. As Tracy said earlier, we’ve both been with GoLeanSixSigma.com since its inception but we founded this company with a clear purpose. We want to revolutionize the way people learn about process improvement, making it easy for everyone everywhere to build their problem-solving muscles.
And we’ve got three core values. We believe in cultivating community, and that’s you. You are our community.
We also believe in having a servant’s heart. We are here to be of service to you with these free webinars.
We believe in having a trailblazing spirit. I don’t think there’s a lot of Lean Six Sigma webinars out here like ours.
We’ve Helped People From…
So thankfully, there are lots of organizations who agree with our philosophy and here are some of the organizations that we have helped.
As you can see, lots of diverse organizations. We’ve got brick and mortar, online companies coming to us for training. There are diverse industries such as health care, financial services, manufacturing, and state governments. And the reason is because Lean Six Sigma is about problem-solving and every organization has problems to solve. And all organizations need people that are good at problem-solving. So for anyone looking to strengthen and build problem-solving muscles, Lean Six Sigma can help you.
So more on benefits later but first, let’s review our objective for today. So after attending this webinar, you’re going to be able to read control charts, you’re going to be able to use control charts to understand your process, you’re going to be able to separate noise from signals in your data, and you’re going to be able to detect if you have improved a process, improved process performance.
So our agenda is what are control charts? Why should you use them? How do you put them to good use? What are some good examples? What are the tips and tricks of control charts?
What Are Control Charts?
So one way to think about control charts, as if they are run charts and they went to college. So like run charts, they track data over time but they provide a lot more information about your process. One difference is the addition of a central line, which may or may not on a run chart. That’s the blue line here with the label X bar and that means average.
The biggest differences are the pink lines that are listed as upper or UCL and LCL which stands for upper and lower control limits.
As a side note, you may or may not know, I got inspired to do this webinar after reading Mark Graban’s new book, which is called Measures of Success, and it’s all about control charts and he talked about it on one of our recent podcast episodes. I wrote a review of his book and you can find both those on our website.
He also refers to control charts as process behavior charts. And what I like about that label is that it’s an accurate description of control charts. They display the behavior of your process. He also refers to the upper and lower control limits as natural process boundaries. And that’s another great description since people can get confused by the word control. So these upper and lower control limits or process boundaries tell you what to expect of your data.
For example, almost all your data should fit within those two pink lines. And we will explore the construction of control charts in more detail in a moment. But first, I want to hear about your experience with control charts.
So we’re going to go to this poll. And the question is, what is your understanding of control charts? Either it’s the first time you’ve heard of them, you’re aware of them, you’ve used them to understand your process, or they are your go-to charts. So I’m going to go over and launch this poll for you. So it’s launched and I’ll let you guys start chiming in. OK?
How about you, Tracy? What do you experience out there when you’re consulting and working with clients? What do you see in terms of control charts usage?
Tracy O’Rourke: So control charts and what I’ve seen is people are not really familiar with them if they haven’t gone through some sort of DMAIC training or Green Belt training, some level of that. And so they aren’t really using them but once they get to learn them, they really do like them and they start to use them a lot more.
So obviously with education, comes more use, which is a wonderful thing because you can learn about stuff all the time and not use it. But I find people do use them after they learn about them and know them.
Elisabeth Swan: OK. Well, let’s see what these guys have as experience. So, I’m going to end that and share it.
Tracy O’Rourke: So it looks like we’ve got 46% saying I’m aware of control charts. So that’s great. Almost half the people, close to half the people have said yes.
That’s followed by I have used them to understand my process, 28%. So that’s wonderful too. That means that we’ve got a lot of familiarity on this call with control charts at least.
20% said first time I’ve heard of them. And so, we’ve got a lot of newbies on the call so that’s great.
And they are my go-to time charts, 7%. So there are some lovers out there.
Elisabeth Swan: Well, that’s cool. So our goal is to convert them all into D, to make this their go-to time chart.
Why Use Control Charts?
Let’s about why should we use control charts? All right. We’re going to unpack what’s special about control charts by covering these top three reasons to use them.
One, they let you see your data in context. An example would be if you went to the doctor and she told you you were 10 pounds overweight but you had just spent the last six months and you’ve lost 50 pounds, you’d wish she discuss your weight loss within the context of your recent weight loss.
So another thing is they let you separate signals from noise. So imagine seeing a stock you just bought dropped 10 points and thinking, “I got to sell!” But in reality, it goes up and down 10 points all the time. It’s not out of the ordinary so there’s no reason to sell.
Number three, they let you detect improvement. So imagine studying and trying to – a student trying to pull up their grades from a C average, they finally get a B and then thinking, “Wow! I’m getting better at this school work thing.” But the next few semesters, they got nothing but C’s. So it was just a blip. So we want to be able to detect improvement, and control charts can help us do that.
So we will cover all these reasons to use control charts but let’s start with the fact that they let you see your data in context.
1. See the Data in Context
So why do we care about seeing data in context? The first example is something that we see lot. It’s an average of the type of customer satisfaction rating we get from surveys and comic cards. And averages can be misleading since they don’t necessarily capture the true variation of what the customer experiences.
So a hilarious example you may have heard that the average diaper wear is 40 years old. Now, that’s silly. But you get that when you average newborns and 80-year-olds.
So same thing here, if we use a histogram on this customer satisfaction data, we discover that the average of 4 rarely happens. This display uncovers that people either have a great experience or horrible experience. So they give us 1’s or 7’s.
Again, it’s better than an average but histograms only provide a snapshot of our data. They don’t tell you when each of those data points happens.
Look at Data Over Time
For that, let’s go to a control chart. So looking at the data over time, this quality data, we can see that something, we’ve got customer satisfaction scores between 4 and 7 but something happened at the end of June. They got much worse in July. So if this was your process, you would have a much different reaction to this chart than if you just were hearing that the average customer satisfaction rating was 4. So that’s key. We want to understand our data in context.
Single Points of Data
So this is something else we also see a lot of. A display like this bar chart shows that the process has changed. In this case, the bar chart on top indicates that quality has dropped since last week. Now, given that visual, a leader may ask, “Hey, what happened? We were doing so well last week.” The resulting discussion might end with finger pointing, changes in plans, extra over time, or other directives to address the problem.
Now, another common is this dashboard. So here we see color coding. And we are huge proponents of color coding and visuals and using those to help understand the process. But here, if red is bad and green is good, it’s only reacting again to a single point of data. So leaders might see this visual and ask, “What’s wrong? Why aren’t we hitting our quality target?” The people responsible might feel pressured to add staff or worse, blame someone and cut staff.
So just seeing that quality is below target doesn’t tell us that the process is getting worse. For that, we need a control chart.
What Does Data Reveal Over Time?
So this week and last week looked pretty common for this dataset with an average of 94, 95 is actually above average. So it looks like 95% for this quality level is what a leader could expect. It would be reasonable to expect that of this process. So even if it dropped all the way down to 86%, it’s still within these upper and lower control limits. This is a stable process.
If this isn’t good enough, then it’s time address system. The difference is looking at the process as a whole and then working to improve it as opposed to just looking at what change from last week to this week, which is nothing.
So reacting to the dashboard or the bar chart to the drop since the previous week would have been a waste of time. Leaders making decisions on single points of data are often doing just that, wasting their time and effort. They need to look at the data in context.
Recap: See the Data in Context
So let’s review, one of the reasons that’s really good to use control charts is that they tell you, they let you see your data in context. So, one data point cannot tell you if the process is getting worse. Dashboards present a limited view of your data. It’s important to look at your data over time. And looking at your data over time provides you with the context you need to determine if your process is really getting worse or getting better or staying the same. So that’s the first reason to use a control chart.
So now, let’s take another poll from you guys which is, how is data commonly treated in your organizations? Now, we want to understand. Do your leaders react to single points of data? Do your leaders make decisions based on averages? Do your leaders look at data in context? Or is it a really a mixed bag and you got some people using data in context, some using and responding to single points of data.
So let’s go to our next poll. So Tracy, what’s your experience?
Tracy O’Rourke: So the most common thing I’ve seen is that leaders aren’t tracking anything but an average. They are making decisions based on averages. They get a report that shows an average but doesn’t show a graph. It just says, “Here’s your average for this week.” And they do assume everything is OK. They don’t even look at standard deviation sometimes to see if their variation in the process is changing or adjusting. They just look at the average. So only seeing really less than half of the picture is the most common thing I see.
Elisabeth Swan: Yeah, averages smoothed it all out and make it look a lot better than it really is. It really takes away what your customers are actually experiencing.
OK. So let’s see what these guys experience.
Tracy O’Rourke: OK. So 55%, it depends on the person, it’s mixed, followed by 33%, leaders make decisions based on averages, and then we’ve got a tie for the last place which is leaders react to single data points, 12%, as well as leaders look at data in context. So I guess that’s good.
Elisabeth Swan: Yes. Aside from that it’s mixed, they had the same experience as you that people are basing it on averages which is also can be problematic. OK. Thanks, guys. That’s helpful.
2. Separate Signals From Noise
OK. Next piece is separating signals from noise. So the second reason it’s important to use control chart is that they help us separate what we call signals from noise in your process. So let’s take a look at example of both of these.
So we arrive to work every day, those of us that actually go somewhere when we go to work, few minutes early, few minutes late. So what accounts for that kind of variation? If you got kids, maybe it’s who is taking up time in the bathroom. If you got a dog to walk, that might take more time depending on the dog. If there’s traffic, the line at Starbucks, that’s all common cause variation. It happens all the time. It’s normal stuff and it’s often referred to as the predictable noise in a process. That’s all the things that we experience everyday that are not surprising and we’re a little early, we’re a little late.
Now, the second reason why it’s important to use control charts is that it helps us see what signals in our data. So what if we are three hours late? The list we just went through still applies. That stuff all happens. But clearly, something significant happened like a major road closing, an accident, severe, weather, a really, really, really big night. This is known as special cause variation. It’s unpredictable. On a chart, this would create what we would call a signal. So these are two different kinds of variation.
A Way to See Signals
And let’s take a look at what does a signal look like on one of our control charts. This is common control chart called an X or an Individual’s Chart that someone were using throughout this webinar. It’s a great all-purpose control chart. There are lots of control charts that are very specific for different types of data but we see this one most commonly.
So once again, we are looking at the percentage of quality over the past 25 weeks. You can see weeks. Time is always across the X axis. And in this data set, there is a big drop to an 80% quality level during the week of September 17th. And the Individual’s Chart has the ability to show us when a data point is worth reacting to. This is one of those points. It’s a signal of special cause variation because it’s outside that lower control limit. And we know that 99% or the vast majority of our data should be inside those control limits.
So, this one is an outlier. And how did we come up with this idea of control chart? How did we come up with these limits? For that, we will go to the inventor of control chart.
Shewhart & Deming
So Walter Shewhart was a physicist and an engineer and a statistician and he is known primarily as the Father of Statistical Quality Control. And he worked at a place called the Hawthorne Works. And this was the predecessor to Bill Labs. And he was also the mentor to Dr. Edwards Deming who is known as the Father of Quality. He worked there.
So Dr. Deming championed Walter Shewhart’s control charts and he brought them into the greater quality world. So a tip of the hat to both these guys who did great work.
And Walter Shewhart figured out years ago that there were two types of variation and used these charts to understand which types were at play in the processes that he was working on. So one of them represents the noise in the system and the other one indicates the signal.
So his job was to improve voice quality over the phone lines. So instead of getting on a call and saying, “Can you hear me now?” he separated results of the process into two categories. Some data was predictable, things that happened on calls all the time. That’s like low voices are heard to hear, the longer the distance, the lower the quality, or the thickness of the wires made a difference. That was all the common stuff.
Now, the unpredictable things didn’t happen that often, a heavy ice storm, snow storm might cut out voices entirely. If lines were too near the bridges or a bridge, it might turn the call to static. This was special stuff. It did not happen all the time. It was unpredictable.
So he determined that if a process could remove or prevent the unpredictable stuff then it could be controlled. And the only way to consistently meet customer requirements is to have a predictable, controlled process. So hence, that word control. He was looking for predictable, controlled processes.
Distribution of Data
To understand the structure of controlled charts, we will go back to the bell curve. So based on a normal distribution of 99%, the data falls within plus or minus three standard deviations of the mean. So that means we’ve got most of our data in between those two standard deviations, and that’s where we get those three standard deviations on a control chart.
Natural Process Limits
So now, let’s look at the structure of the control chart. Dr. Shewhart used those three standard deviations to calculate the upper and the lower control limits. It comes directly from the data. So the data can be – you can calculate two standard deviations. You can add it to get that top line and subtract it from the mean to get that lower line and then 99% of the data should appear within those two lines just like the bell curve.
But he experimented on which limit should go on these charts. So he lands it on three standard deviations because that’s when he got results. So he was chasing down data points just like we talk about leaders chasing down data points. And he found that when the points fell inside those limits, he found nothing. Nothing was going on in the process. And he called it chasing ghosts. And he realized it costs money to chase down ghosts. Research data that was just random variation took a lot of time and effort to go research.
So he developed control charts to conserve precious resources. There’s no need to go try and find out what happened if the data didn’t fall outside those control limits. So what Dr. Shewhart discovered was that there was common cause variation, the expected factors that regularly influence the process and there’s special cause variation, the rare sporadic factors that impact the process in unexpected ways. The key is to understand the kind of variation you’re dealing with before reacting to it.
3 Control Chart Signals
So let’s go and look at some other signals that can appear uncontrolled charts. So we’ve got lots of stats packages that can let you know when you got special cause but here are three classic signals of special cause variation.
An outlier, so one point more than three standard deviations from the center line is called an outlier. And we’ve just seen an example of that with that quality chart we were looking at.
Next, we’ve got a trend. So this is 7 points in a row, all increasing or decreasing. And once again, the odds of this happening randomly are less than 1%. And the way I would have you think about is if you had a – if you were flipping a coin and you were going to flip it 50 times in a row or flip 100 times in a row, you could expect roughly 50-50. Would it be exactly 50-50? No. It might be 54-46. It would not be exact. But you know what you’re expecting.
Now, what they’re saying and what Walter Shewhart and statisticians are saying is that if you flip that coin and it goes heads then heads then heads then heads seven times in a row, well, something is off. That is special cause variation. The odds of that happening due to random chance are less than 1%.
So we got an outlier, we got a trend, and this last one is a shift, and that is 7 points in a row on the same side of the centerline. And that once again, flip that coin, by the 7th time you get heads in a row, something has happened, something has shifted in the data.
So if it’s good, if the shift is good or if any of these special cause signals indicate something good for your process then you want to replicate it. Like how did we do that? Let’s do that again. And if it’s bad, you want to remove the cause or you want to plan for it.
And you’ve got stats software like Sigma Excel, QI Macros, Minitab, and they have 8 rules like this in general. And you can opt to have each signal labeled with the number of the rule that is being flagged. So we will just cover these three today for simplicity but the same stat packages will flag all these signals for you.
How to React
So as Walter Shewhart discovered, the great advantages of control charts is knowing what to do next. If there is a signal of special cause variation, the next step is to investigate, well, what caused that? You want a stable, predictable process. And the goal is to eliminate the root causes of special cause variation unless of course it indicates something good. So if it was a spike in sales and sort of defects then you want to investigate the cause so you can replicate it.
So that’s how you react to special cause variation. You investigate that point, that shift, or that trend. What happened in that point of time? What’s different about that point in time and every other point in time on this chart? So that’s what you do when it’s special cause.
Now, if a process is stable and nothing but common cause variation is going on, then there is nothing to investigate. But if it’s stable and it is not meeting customer expectations then it’s time for some process improvement. You’re going to have to make fundamental changes to the system in order to improve the process. It’s time to apply DMAIC, PDCA. You’ve got to do this with rigor. You have to approach the whole system, not the one data point.
Recap: Separate Signals From Noise
Let’s recap. A great reason for using control charts is to separate signals from noise. All processes have random predictable common cause variation. They all do. Every process is going to have that.
If a control chart flags a signal of special cause variation like a trend or shift or an outlier then it’s important to investigate the root cause of that signal. If it’s a good thing, you want to replicate it and if it’s a bad thing, you want to get rid of it.
Processes that are stable and show only common cause variations are great but if they don’t meet customer expectations then your next step is to initiate some process improvement. You have to address the process as a whole using something like DMAIC or PDCA. It’s time to conduct process walks and breakout the fishbone diagrams and do the real work of improvement.
3. Detect Improvement
Let’s move on to our third one. So another reason why it’s great to use control charts is because they can help you determine if a process has truly improved. So if we look at this dashboard again and we see the quality metric and we see it turned from red to green because we hit the target, might be cause for celebration. But now, you know too much so you’re too smart to assume that the process is truly improved based on a single point of data.
Did something happen? Is the process really improving? Well, let’s go look at our control chart.
Was It Common Cause Variation?
So this control chart shows a lot of variation but it’s all random. We’ve the target a number of times but we’ve also dipped below 95%. So since the process is stable, it’s not getting any better, it’s not getting any worse. If we don’t do anything, we can expect the process to continue with this much random variation.
Make a Change to the System
If we’re not happy with this, then we need to address the whole system. And we do that with DMAIC. We need to apply some process improvement but we will address all the variation in the system. We have to build profound knowledge of the process. And once we’ve addressed the system as a whole then we can check to see if we’ve made an improvement.
Was It a “Blip?”
So let’s look again. After we apply improvement, we might see something like this. So on this quality chart, again, we see we’ve hit 100 but now we know that’s an outlier. It is special cause variation. And we are hoping we are the reasons, right? If we make an improvement to a process, we want to see special cause. We want to see that what we did caused a change to the process. So looking at this control chart, it looks like it’s a one-time event so we’re going to need to collect some more data to ensure that it’s not a one-time event.
Has the Process Shifted?
So now, it looks like the process has shifted. We collected a few more weeks to data and we’ve got another signal of special cause. The process signal is a shift which in this case is good, right? Quality shifted up. So this is what we wanted to see after we applied our countermeasures. And this means the process has truly changed, which means the average is no longer 92% and the upper and the lower control limits have also changed.
So when this happens, it’s time to display the old process separately from the new process, right? It isn’t really the same process anymore. So looking backwards, we can see, well, those control limits made sense back when the process from June through August. But now, they don’t really make sense anymore.
Display the Old vs. New Process
So how do we deal with that? Well, I used Sigma Excel on this one but it works in Minitab, QI Macros or any stat programs will do the same thing.
And now, we calculate the new average and the new upper control limits of the improved process. And that’s what you want as you apply process improvement. That’s the goal. So this is a good time to talk about the data collection that’s behind all this.
Control charts require a minimum of 20 data points. And that’s great. As soon as you have 20, you’ve got a decent control chart. You can even make them when there’s less than 20. We’re trying to get some idea of what’s going on there but that’s kind of an official rule around when they become more valid.
And we’ve seen improvement efforts run into trouble when data is collected rarely. Even some of your metrics may come monthly and no one thinks about. They just are doing some improvement work. But then how many months of data? How long do you have to wait to see if things have actually changed? It takes – it will take 20 months to build that control chart. So collecting weekly is better than monthly and collecting data daily is better than weekly.
So the more granular your data collection, the better, but will acknowledge across the board data collection takes time and effort. So you can only collect what you can afford. So just keeping that in mind. Obviously, more frequent, better. But you got to decide for your process and for your resources what’s possible. But keep that in mind.
Recap: Detect Improvement
So let’s recap. Another great reason to use control charts is because they let you detect if your process has truly improved. Don’t be fooled by a single data point. You need to see a shift in the process.
Control charts require a minimum of 20 data points.
And collecting data weekly is better than monthly and daily is better than weekly.
And once you’ve seen a true process shift then you can show the before process as separate from the after process. And STAT packages can do that for you or you can create two separate charts. There’s a low tech way to do it so don’t be intimidated.
Question for You
All right. So now, I got a question for you. So your process can be in only one of these three states. It could be stable, not improving or getting worse, or it could be improving or it could be getting worse. And my question for you is how do you detect this now? And just think about for a second. Pop that in the chat window so we can hear how you’re detecting processes improve or get worse or stay stable now.
And then Tracy, let me ask you. What do you see happening?
Tracy O’Rourke: So what I see happening out there is, well, this is kind of sad I think, but the way – they really aren’t necessarily measuring stability. And improvement is if they have time and they monitor getting worse by customer complaints. So this is after the fact, it’s already affecting the customer. It’s very reactive in my opinion to just wait until the customer starts complaining. But unfortunately, that’s typically for folks that are not process disciplined and not tracking it. It’s really customers that are alerting them that now you have a problem.
Elisabeth Swan: That’s actually a great example of a lagging indicator. And as you say, it’s too late. It has already happened. And as we know, by the time it hits the customer, the impact to your reputation is much higher than if you catch it internally before it hits the customer.
The other thing that it points out is that these control charts are leading indicators. They can tell you that things are getting worse before you actually find out from a customer.
Tracy O’Rourke: Yes.
Elisabeth Swan: So let’s see what have you gotten from folks as examples?
Tracy O’Rourke: So the good news is people are saying, well, from a chart, they might be using a run chart and they are indicating trends. We have control charts. People say the control charts. So that’s great too.
Also, people are mentioning, let’s see, we have Mira Rodrigues is saying that data collection of the customer.
David Nickels is saying, “Weekly KPIs define our direction.” So that’s great. They’re checking it. They’ve got a process to look at some of these important measures.
Debra writes, “Research process. Collect data. Verify data accuracy to do a chart over a period of time.” And they do that quarterly.
Monique says, “Based on velocity but it’s so volatile and unreliable. I decided that Lean Six Sigma and control charts might be the better solution so here I am.”
Elisabeth Swan: Way to go! Well-spoken.
Tracy O’Rourke: Yup. Rick mentions, “Personally, I use control charts for everything I can. Organizations typically are reactive.” So that’s great.
Elisabeth Swan: So he has already seen the light. So that’s a great technique he is already using.
Tracy O’Rourke: Good. And Glenn says, “Be aware of potential shifting in trends. It may or may not be a signal but add more analysis if the trend continues.”
Elisabeth Swan: That’s good. Skeptic.
Tracy O’Rourke: Yeah. Yup. Thank you for those. Those are great.
Great Reference Blog
Elisabeth Swan: Yeah. Thanks for that, guys. Those are great examples. OK. So one thing that might be helpful to you is this great reference blog written by a colleague of Tracy’s and mine, Master Black Belt Craig Tickle. And it says there are basically 5 Ways to Improve a Process. That’s a really nice blog that just sort of lays out here’s what you can do, here’s how you can make things better. And two of them are related to what we’re talking about today.
So one of the methods is to reduce common cause variation and another method he talks about is reducing special cause variation. So it’s a great add-on once you’re out of this webinar. I highly recommend. We’ve got a link here or just look up 5 Ways to Improve a Process or look up under Craig Tickle.
Control Chart Tips & tricks
So now, let’s come back to some tips and tricks about control charts. So always include a control chart. Follow Rick’s advice, one of our webinar attendees, when you’re displaying data. Individual’s is fine. That’s the chart we’ve been using. It’s called an X or an Individual’s Chart. There are specific control charts for every type of data. The other big common chart is called a P Chart. That’s usually for proportion-defective. But I’ve seen people use Individual’s for those. So I’m probably going to get my hands on that set. But it’s just a good all-purpose chart in my opinion.
You want to differentiate special cause variation from common cause variation. So that’s back to really watch that data until it’s an actual trend or it’s an actual shift or it’s an outlier. So differentiate. Is this just random variation? Are process goes up and down? They all do. Or is this something special going on that I should react to? So investigate when it’s special cause.
If it’s common cause variation, you want to understand, is that – it’s stable but is it good? So you can have stable and good. You can stable and not good. So if it’s not good then you’re going to want to tackle the process and use some method like DMAIC to approach the process and improve it.
And then collect data as frequently as reasonable. So what can you afford? That’s how often you want to collect the data. So those are some tips.
Call to Action
Now, this is what I would suggest coming out of here. We’re going to have by tomorrow morning, we will upload a control chart template and this is new. And it’s very simple, again, an X or Individual’s Chart. And our templates are always free but this one has a space for you to enter data. You can see the image there of the data already having been entered in here. So it will calculate your limits for you and then it will create a chart for you.
And we’ve got the definition of the chart in there and the questions that it will answer. So it’s a nice, easy to use. It creates a chart for you. You can create as many charts as you want. Just keep creating separate tabs for all your separate charts. So take that and use it to properly react to your process data so you can get started whether you got a stat package or not. These are not super sophisticated. We talked about how we calculate those upper and lower control limits. So that’s what’s going on in these and the macros involved.
Today’s We Covered
So let’s look back. What are control charts? We covered, why should you use them? We gave you three big reasons. Why – how do you put them to good use? What are some good examples? And we covered some tips and tricks of control charts.
So that brings us to some Q&A with all of you. So take your time to enter questions you got for Tracy and I in the chat window again. And while you guys are taking some time to enter your questions, Tracy and I are going to cover a few bits of information for you.
So one, you can learn about control charts in more detail by going after your Green Belt Certification or taking that to another level and become Black Belts. So get more training. There are lots of training here. The White Belt, Yellow Belt, always free so you can pursue more training.
Another thing that’s new if you have not heard of it is we now have this thing called a Go-Getter Membership. So there is a forum you guys can network with your peers, you can connect to our Master Black Belts. You get access to exclusive members-only content, which this webinar is one of those. So you will get access to that. You get access to all the single modules. That’s almost 300 bucks right there. So you make your money back immediately. And then you get free access to intermediate, advanced, and leadership webinars. You get books, guides for free. You can take 20% off anything. We’re not just handing over to you. You definitely want to check that out.
Tracy, did I miss anything?
Tracy O’Rourke: No. I think you’ve covered it all.
Elisabeth Swan: OK.
Tracy O’Rourke: So I would say get going on the Go-Getter Membership.
Elisabeth Swan: You’re good. You should be in sales.
Tracy O’Rourke: Right.
The Problem-Solver’s Toolkit
Elisabeth Swan: Also, Tracy mentions, our book is available now in paperback. We provide a basic tool set of 35 tools to get you from bad to better process performance. We give you examples, instructions, short pass around sort of common mistakes people make with the tools and with the process, and a little sightseeing options, links to send you to other webinars and blogs and things that are related to the topic if you want to learn more.
There is also a downloadable ebook which is packed with colorful charts, constructive infographics, and the familiar faces of the staff at the Bahama Bistro with their own case study throughout just like in our training. I think it’s awesome, Tracy. And it’s now being handed out at universities, isn’t it?
Tracy O’Rourke: Yes, it is, a couple of them.
Upcoming Webinar: Feb. 28, 11am PST
Elisabeth Swan: Yeah. OK. And then we’ve got an upcoming webinar. Tracy, why don’t you describe what’s going to happen on the next webinar, February 28th at 11AM Pacific Standard Time?
Tracy O’Rourke: Sure. So this is really around the soft side you could say of process improvement. It’s about culture and what leader’s role is in building a culture of process improvements. So a lot of the belt level trainings are focused on tools and practitioners, people applying those skills and tools. But that we all know that that’s not the only thing that creates and helps an organization be successful with applying process improvement.
And leaders also have piece and a role, a very important role in promoting a culture that will allow people to use these tools and to actually improve a process. I don’t know if any of you have ever had the experience of learning how to do process improvement and then basically you weren’t really allowed to use the tools or allowed to – because your leader didn’t want you to or they were afraid.
And so, it’s really hard to apply process improvement in that environment. So we’re going to really spend some time talking about what leaders can do and what their roles are in building a culture of continuous improvement. So come join us next month.
Elisabeth Swan: Yeah, absolutely, Tracy. And it’s a great counterpoint to a webinar like this one where we’re getting into the nitty-gritty of reading your data and then to go to the flipside and say, “Yeah, but who is supporting this and who is helping to build that culture?” So really great, great combo.
Just-In-Time Café Podcast
The latest episode of our Just-In-Time Podcast includes an interview with Mohamed Saleh. He is a Lean Sensei from Hartford Healthcare in Connecticut and he is just full of fantastic wisdom about again, how leaders can help support Lean and a lot of what not to do. He has got a lot of cautionary tales and he is great. He is funny and got great experience. So that’s a good one. Please join in or download that one on iTunes.
Success Story Webinar
Another great success story you can look at, this is a local nonprofit. This was a group I worked with, local child care organization. And this was – she actually then became the CEO but she reduced provider payment process rework from 25% to 5%. It’s a great story. She discovered all kinds of great nuggets in this process improvement effort where something like – one of the fun discoveries was they had to change the paper to pink paper every time they printed an invoice. And this was based on something ancient, somebody wanted from finance years ago and no one knew why and it caused the kind of problems you would guess it was. So it’s a great story, good one. Listen in to that.
Wonder Women of Quality
And please check out our infographic on our latest Wonder Women of Quality. This is a great series. And Tracey Richardson is a great add. We also have a podcast with her and a book review of her new book, The Toyota Engagement Equation. She worked at the Kentucky Plant, first North American Toyota plants in Kentucky and has some great stories of what she learned along the way. So that’s also a wonderful, fun infographics. These I find so inspiring. I always recommend checking those out.
And that brings us back, Tracy, to some questions. Remember to put your questions in the chat window. That’s where we can see them. Go ahead, Tracy.
Tracy O’Rourke: OK. So this one came during the webinar from Mr. Al Hamazi. I’m sorry if I said your name wrong. I apologize. His question is, “What’s different between run and control charts in a nutshell?”
Elisabeth Swan: Great question. Yeah, great question. I would say the biggest difference is that a control chart has these calculated upper and lower control limits and run charts don’t. They both might have a center line but the control chart has upper and lower control limits that are plus or minus three standard deviations from the mean and we expect 99.3 something – 99.97% of your data to be inside those control limits. And if they are out, that’s a sign of special cause. They help us – they allow us to see signs of special cause. If you see a stat package, you can have it flagged one of eight different special causes to see if there’s undue influence on your process or if it’s all just common cause variations. Those are the big differences.
Tracy O’Rourke: Cool. Thank you. Very helpful. And Carlos wants to know, “Is it still common to show both the X or Individual’s Chart together with the Range Chart.
Elisabeth Swan: Yeah. I didn’t show the moving Range Chart. That is a great combo effort with the control chart with the Individual’s Chart. And that one, it just requires a little more explanation. It’s looking at the point to point data difference as opposed to just mapping the data itself. It looks at, well, what’s the range between the data points.
And the underlying assumption if it’s the same process then point to point differences in data shouldn’t be that great. It’s just another good measure of the variation in your process so you would look there for a special cause variation just like you would look in the Individual’s Chart. But great question, Carlos. Yes, those are often used together.
Tracy O’Rourke: Good. Thank you. Bettina has a great question as well. She writes and her question is when can you stop using a control chart?
Elisabeth Swan: Never! Never! Why would you want to stop? They are so handy. They are so good. We should always use control charts. We should make it your habit.
Tracy O’Rourke: Yes. And I think sometimes we look at them a little less. So once a process is stable, you can look at a control – if you’re looking at a daily to see something, you might look at it weekly or monthly. So sometimes the frequency of looking at it may change depending on where the process is.
Her second follow-up question, Elisabeth, was, can you follow more than one parameter on a control chart?
Elisabeth Swan: No. And I think that that’s something that you could do with run charts and you could do with line charts is say, “I want to look at the change over time for this year and last year.” And that would be multiple line charts. Whereas a control chart can only calculate upper and lower control limits for one dataset so that excludes the possibility of having a second line on that chart because it would be a different dataset and would have a different upper and lower control limits so it gets a little messy. So it can only do its magic if it’s on one dataset.
Tracy O’Rourke: OK. Wonderful. Thank you. OK. Another question we have is Alicia would really like you to briefly review how to calculate those control limits again.
Elisabeth Swan: Good question. Yeah, you could create your own control charts. You take the average. You find out the average of your dataset. So that’s pretty easy. Excel will give you the sum of your column of data. And then you take that same dataset and you calculate the standard deviation of that data. So again, Excel has, I think it’s STDEVP. So you can use and Excel formula to calculate the standard deviation of that dataset.
And then you take the average and subtract the standard deviation to get the lower control limit and then you take that average and you add the standard deviation to get the upper control limit. And the only time this is different and I showed it to you today is when we had a quality measure that could not go any higher than 100%. So I wrote in my own upper limit of 100 because it was going to calculate – if I add the standard deviation, it would have gone above 100 and the quality score can’t go above 100.
So I hope that – does that sound clear, Tracy?
Tracy O’Rourke: Yes, that was very helpful. Thank you, Elisabeth. Good job. I think it helps to see it on the slide but I think for the Q&A, I think that was really good.
OK. Rick has a question somewhat related, not really, “How do you account for lack of normal distribution in Individual Charts like percentages that can exceed zero or 100 that are just used to gain insight and ignore the stats? Curious about your thoughts.”
Elisabeth Swan: Say that one again. I have to absorb it.
Tracy O’Rourke: Sure. How do you account for lack of normal distribution in the Individual Charts percentages that can exceed zero or 100 that are just used to gain insight and ignore the stats? Curious about your thoughts.
Elisabeth Swan: They have to look at the charts to answer that with, can be helpful. I wanted to let anyone out there know, if you are using Q&A to enter questions, I don’t think Tracy can see them.
Tracy O’Rourke: No, I can. I can.
Elisabeth Swan: You can. You can. All right.
Tracy O’Rourke: Sorry about that.
Elisabeth Swan: That’s all right.
Tracy O’Rourke: OK. Another question. Do you have any tips for implementing control charts in a daily management board setting where charts maybe manually drawn on a daily basis?
Elisabeth Swan: That’s a great question too. And it’s a cool thing to do. I’ve seen this in huddles and when teams have a control chart, as you say, manually drawn in their area with room so that every day they can plot the latest point and leave the historic calculated average and the historically calculated upper and lower control limits, and keep plotting and watch if the process shows any outlier or if there’s an outlier or a shift or a trend then they plot it manually.
I have a really interesting story. I have a friend who did a lot of work with US school systems and they use control charts in the classrooms and they had even young kids plotting both attendance for the day and also test scores. And they had them on control charts. So they were doing just what you described and they had them on the wall. And the kids were the ones that had to enter the daily numbers. So they had to enter the attendance and they had to enter the test scores if there was a test.
And she told the story where the two kids missed the bus and they ran all the way to school. And this is Texas. These kids arrived covered in dust, having run along the side of the road, and hold on to the principal’s office because very dangerous. They were very upset these kids had run alone into the school. And the principal asked them, “Why didn’t you get – call someone and get a ride in?” And they said, “Well, we don’t want to be late. We don’t want to be late. We don’t want to be the reason the chart goes down.”
They knew. They were so attuned to it. So what you just described is a really good practice where you have it front and center. People see it. They are watching their data. They are watching it go up. They are watching it go down. They are watching if it’s random and common variation. They are watching if something special is happening. It’s a really cool practice. And thanks for bringing that up.
Tracy O’Rourke: Yup. And I would even say, have the people on the team that attend the meeting update the chart manually. That’s engagement for sure.
Elisabeth Swan: For sure.
Tracy O’Rourke: OK. I think we have one – Simon had a question, “I hope to hear something about the different types of control charts. Could you explain something about the different data types and which type of control chart is subsequently indicated to be used?
Elisabeth Swan: So for that, and correct me if I’m wrong. I’m pretty sure we stuck our control chart tree.
Tracy O’Rourke: Yup, on our website.
Elisabeth Swan: On our website. And that’s probably a good thing to stick on these on the template. We could put the template in there. But anyway, yeah, there’s a great tree that we have that will tell you exactly what you just asked and you can just follow the – it’s a discreet data which is count data, how many mistakes were made or is it continuous data, something like time, how many minutes. Then you got choose the different families of charts, and it goes from there.
So great question and there is a tree for you.
Tracy O’Rourke: OK. William has a question, “Do you have any special training that we can have access to?” I think we might have some training.
Elisabeth Swan: Tracy, what have we got? We’ve got Green Belt training. It has control charts in it. We’ve got Black Belt training. Also keeps rolling with control charts and gives lots of instruction. We’ve actually got explanations online and datasets to use with either Sigma Excel if you use that or Minitab. We sort of went with – we try to stay platform-agnostic but Minitab works on PC and Sigma Excel works on both so we got data and instructions for you. So we got both training and some stuff to play with.
Tracy O’Rourke: OK. All right. That’s all the questions we have time for today.
Elisabeth Swan: Awesome. Thank you all. Those were great questions. Thanks for joining us and look forward to seeing you guys next month.
Tracy O’Rourke: Thank you, everybody. See you next month.
View our upcoming webinars and join live so you can ask questions and let us know what you’d like to us to cover next. We’re busy building new webinars all the time. And we’re happy to know you’re busy too – building your problem-solving muscles – keep it up!