GA 492 | Better Metrics for Better Management with Mark Graban

You shouldn't be measuring something like quality, you should be improving it. How do you know if it's improved if you're not measuring it? Now this is kind of like, well, wait a minute. Again, measurement is important, but it's not the only thing. Measuring something doesn't guarantee that it's going to improve.

Today I'm excited to welcome Mark Graben back to the show. Now, Mark was on not too long ago, but he was back in town. And so we decided to get together to do another in person interview here at the Gemba Academy Academy Studios. And so today we talk all about a workshop, actually that Mark's going to be running for Gemba Academy in October, October 10 and 11, called Better Metrics for Better Management. And so really what we talk about is the whole idea of using process behavior charts, some people call them control charts, and just how to, how to measure and manage our business. So it's a really interesting interview. So it's not just all about control charts and whatnot. It's really about how do you, how do you pick metrics to measure? And then when you have them, you know, you have those metrics picked, like, how do you go about it? What should you be looking at? And so Mark is really an expert on this topic. And yeah, I think you're just going to find a lot of value in this episode. So show notes can be found over@gembapodcast.com just look for episode 492 and also go to gembaevents.com and that's where you're going to be able to see more information about this workshop that we're going to be doing in October with Mark. October, October 10th and 11th here in, in the Gimba Academy Training center here. And if you come, you get to see where we do the, all of our videos, you get to see where we do our podcasts and all that kind of cool stuff. So it's going to be a good time. Okay, enough for me. Let's get to the show. Mark, welcome back to the show.

How's it going, Ron? It's great to be back. It hasn't been too long this time.

I know, I know, I know. We always just have so much to talk about, man.

I'm glad we have the chance again. It's good to see you.

Yeah, yeah, yeah. So. Well, hey, man, I, I'm excited to talk about Better Metrics for Better Management. It's, it's. Obviously we got a workshop coming up here in October, but we'll, we'll Talk about that in a little bit. But before we get into all that, you know, you know the game here, we like to start with a quote. What do you got?

So I'm going to, I'm going to read a quote and it's a misquote that's commonly quoted. But then let me explain, I think what the, the misquote is. So it's often attributed to Dr. Deming. According to the Deming Institute, the quote that's often thrown around is a misquote as they spell out on their website. So here's the misquote. If you can't measure it, you can't manage it. And they'll often say in quotes, period. W. Edwards Deming, 1900-1993. It's taken out of context the full quote from one of his books and what he taught. This is from the book the New Economics. Here's the full quote. It is wrong to suppose that if you can't measure it, you can't manage it. A costly myth like that's as close to a completely 180 degrees off misquoting of.

I didn't know this. I mean I've heard the misquote a million times.

But okay, so yeah, so there's the, I don't know if that counts as a so called truth bomb or a devil.

So what's that all about? What was he getting at, you think?

Well, I mean can paraphrase what I kind of did the refresher course of what the Deming Institute says and people there. Deming of course isn't. I mean he's not saying don't use data. I mean Dr. Deming talked about measurement and he was an advocate for statistical process control that he learned from Walter Shewhart and he worked with and actually in a lot of ways Don Wheeler, the statistician who's still active today. Wheeler was both learning from and teaching Deming. Good example for Dr. Deming to continue learning these things. Dr. Deming was an advocate for a lot of these methods. But I think if you start looking at the nuance of the statement. Deming also talked about how sometimes the most important things are immeasurable, but you have to try to manage them anyway or we have imperfect measures. So like if we were to think about here's an important concept we talked about last time, psychological safety. There's not really a direct measure of the level of psychological safety in an organization, but it's important. You have to manage it. Leaders behaving in certain ways can strengthen people's feeling of psychological safety or not. You can survey people, you can put numbers to it, you can try to quantify it, but that's maybe at best, an imperfect measure. And I think there's another piece of it, of there's a different expression thrown around, like if you don't measure it, you can't improve it, or there's almost implication that measurement is one of the most important things. But just measuring and monitoring and looking at a metric and telling people to try harder probably isn't going to move the needle. So I think there's a time and a place for realizing, sure, metrics are important, but we've got to use them in a way that's most helpful or less dysfunctional, and to think through what are the things that are hard or impossible to measure and how do we have to manage them anyway.

Yeah, yeah, got it. All right.

So if people want to say if you can't measure it, you can't manage it, they can advocate for that. But I wouldn't attribute that to Dr. Deming as such.

Yeah, they need to call it Mark Graben, etc.

Well, don't cite me. I'm. I mean, I'm. I'm. I'm. I think, in agreement. Yeah.

With him.

With Dr. Deming. Yeah. You know, and there's other things that, I mean, like, you have to be careful when. I mean, I. I don't know if someday or even now if somebody's. I don't know, in a way, misquoting me from the book Lean Hospital. Who knows?

Yeah, well, you know, it's. It's interesting, just the. The importance of. Of balance in all things. You know, when you get too far to one side on anything, you're probably headed for trouble. Right, sure. And, yeah, so I think that definitely applies to what we're doing in continuous improvement. You go to a value stream mapping workshop, and now you come back and we're gonna map every single value stream. Or you're.

You're.

I think you popularized lame in our movement. Lean as misguidedly executed. You know, so, like, you taping out where the mouse goes on your desk because you learned about 5s, that's a little over the top, right?

Yeah. I mean, people might use tools in a way that's unhelpful. Whether that's 5s and look, I mean, I think a sign of it being unhelpful is if you're having to force people into compliance through audits. If something is helpful and people have time to use it, they would probably tend to Use it. It's a little more complicated than that because people sometimes just get very comfortable with the status quo. But if you're forcing a tool, that's probably a bad sign or the risk of misusing a tool. You can go and current state map all of your value streams. But so what are you identifying the gaps to the ideal state? Are you sketching out a future state map and are you working toward it? Mapping the current state in and of itself does nothing. Measuring performance in and of itself does nothing. If again, you're not trying to connect that to improvement.

Yeah. Love it. Love it. All right, Mark, you've been on the show a number of times. Like you said, not even that long ago. But just in case for some reason people don't know who you are, give us a quick update on who's Mark Raven.

So I'll give a little bit summary of what I do. And it's funny, people even like people who know me will ask this question and sometimes there's different tones of voice. So it's like what do you do? And then there's the what do you do? What do you do? I have these different categories. I have my own business. So author and publisher, speaker, coach, consultant, podcaster. Even though that's some not we're on a podcast. So like why am I about to go on a detour? Like, well, podcasting. Why would I. Yeah, yeah, why would I tout that? But yeah, I get to do a lot of different things. I have a part time role with Kynex's software company. I'm an industrial engineer with an MBA who thought his career was going to be in manufacturing. And then in 2005 that took kind of a detour when my wife and I moved here to Keller, Texas at the time, real close to you, Ron, and and had an opportunity to start doing consulting work in healthcare, originally through Johnson and Johnson and you know, nowadays and over the past five years with the book Measures of Success and then with the new one, the Mistakes that Make Us those, those are both intentionally not framed as healthcare books. I'm trying to reach a broader audience and just as if, just as like I wasn't running away from manufacturing in 2005, I'm not running away from healthcare, but sometimes there's just opportunities to try to work on interesting problems with motivated people in different industries. So I enjoy doing more of that again these days too.

Yeah, I see you, Mark. As you like to help people.

I do. It can be helpful.

Yeah. And there's people in all walks of Life. And that's how I've always seen you. So wonderful, wonderful.

Thank you.

All right. Well, yeah, man, like I said, October 10th and 11th, we're doing an in person workshop here in Keller, Texas. Better Metrics for Better Management. I'm very, very excited if for those that are kind of been around the quality movement, you know, you might have heard about this red bead experiment and maybe we'll talk about that in a little bit. But Better Metrics for Better Management, obviously you named it so and it's going to be largely based on, on some of your writing. So what does it mean? Better Metrics for Better Management. Like where does that phrasing come from?

Yeah, I'm trying to think, where did that come from? I know it popped in my brain at some point, but I think that's part of the goal. It's better management, better performance, and how can we use metrics in a way that moves us in that direction? And just real quick on the workshop, I'm real excited to be doing this in person, this kind of evolved, trying to continuously improve. So hopefully this version of the workshop is the best the workshop has ever been. This will be the first in person workshop I've been able to do since 2019.

Wow.

So I'm real excited about that and looking forward to having people come together. But when you think about better metrics, better management, like to me, better metrics, there, there are two pieces to that equation, if you will. You know, first off is choosing your metrics. What are we measuring and why? How are we trying to create alignment through the organization? That's not really the scope of the workshop. That would be a different workshop, I think, on strategy deployment or whatever term you want to use. Critically important discussion around what are you measuring? Maybe not trying to measure too many things, making sure that you're not measuring in a way that encourages people to suboptimize they're part of the organization. But then there's the question of what do you do over time with those metrics? And that's really what the book Measures of Success and what this workshop is about. We'll assume the things your organization is measuring are the right things to measure. I mean, that could be a dangerous assumption. Or let's just roll with it. But there's a book Measure what Matters, John Doerr. It's been really popular methodology called OKR Objectives and Key Results. Really popular in Silicon Valley and tech circles. I've seen other people even kind of adjacent to or in context of Lean adopting that approach. It's in a lot of ways similar to strategy deployment or Hoshin Connery as some people might call it. But when I read Dorr's book and like, okay, well, it's great, again, important discussion of what are we measuring? But then his book doesn't really continue the discussion of, well, what do we do with the metrics over time? Like when, if we're monitoring a metric, when does the most recent data point or a group of data points tell us that something important or meaningful is happening? And you know, there's this important lesson and we'll explore this in the workshop in a lot of different ways. Separating quote, unquote signal from noise in your performance measures. There's this trap a lot of organizations fall into of like saying, well, okay, here we are in August, we just got our July metrics put together and July's number, whatever this number is, July's number is worse than June's. Please explain. Let's talk about that. And that could be just noise. It could be random fluctuation, basically common cause variation. And when we react to that, we basically end up like wasting time. And nobody has time to waste. Right. So let's say if you've got 20 metrics and maybe 10 of them are worse than the month before and 10 are better than the month before, all that time spent explaining all those metrics that are a little bit worse, like, what is that really accomplishing? Because we might talk about it a bunch and we might say, okay, well we're going to go try to do better and then we get our August metrics and things have bounced back up a little bit. Yeah, maybe that number is just fluctuating. Yeah, sometimes people are only looking at a couple of data points and, and if you look back at 10 or 12 or 24 months worth of data, you can start seeing patterns like, is that number just fluctuating around an average? If so, let's stop micromanaging every up and down. And then there's the question of if we're doing some sort of project, Kaizen event, Six Sigma project, what have you. We have baseline performance that when I think of projects I've worked on, metrics probably fluctuating around a stable average. And now we're doing an improvement project, we want to be careful and make sure we don't declare victory of say, oh, the first data point after a project is better than average. Hooray. And like, well, maybe that's part of the past fluctuation. So, you know, we want to look and see are we you know, if we're driving statistically meaningful improvement, are we seeing like a step function shift up in performance? If we're trying to move the metric up? You know, we can use this method, process behavior charts, a specific form of statistical process control, to help us answer both those questions. When do we react to a data point or not? And then make sure we're doing a good job of evaluating improvement projects or efforts.

Yeah. So I'm guessing many of the folks listening to this right now are familiar with process behavior charts, or maybe they call it control charts or something like that. But maybe just quick lesson, explain the control limits and just high level, like, what is a process behavior chart?

Right. So here we have the challenge of explaining a very visual tool.

Yeah, yeah.

In an audio format.

Everybody close your eyes and imagine a graph.

So let me compare and contrast a little bit, though, to what I think is really common practice. I'm not calling it best practice, but common practice is to see a list of numbers, basically a spreadsheet grid of numbers. So, okay, fine, we're not just looking at two or three data points. We've got the last 12 months list of numbers. There may or may not be red and green color coding in these numbers. It's really hard for people, I think, to make sense of a list of numbers. We have this technology. I don't know when this was invented. There's probably an example drawn on a cave wall somewhere, what we call a run chart or a line chart of just plot those data points, connect them with lines if there's a trend or fluctuation or a shift in performance. My gosh, it's so much easier to see that. Just use a simple line chart or a run chart.

Yeah.

So much more helpful than a list of numbers. But then what makes that a process behavior chart or a control chart is that we add three lines so we can calculate an average, in some cases, maybe a median. But we're going to calculate an average from a baseline set of data points, plot that average on the chart. Then we calculate what are called the lower limits and the upper limit, two more horizontal lines. We can think of them as like guardrails. These are not the targets or the specifications they're called. This is a process behavior chart. That phrase coined by Don Wheeler, the statistician. It's a chart that describes the behavior of your process, of the quote unquote voice of the process. A target or a specification might be the voice of the customer. Yeah, that's where we get into the red, green thing. But if we're looking again at a process behavior chart, we calculate, we do not get to choose. We calculate this lower and upper limit. We'll go through the formula in the workshop. There's some relatively simple math to look at this baseline data, not just what's the average, but how much point to point variation is typical. If there's more point to point fluctuation, those limits are going to be calculated to be wider around the average. So you calculate that lower limit and the upper limit, you throw those on the graph, then you can evaluate the chart visually, both retrospectively and as you add more data points going forward. There's a couple of different rules that we're looking for that would define a data point or data points that are a signal that would tell us the system has changed.

Yeah.

So if the system's been stable and that number is just kind of fluctuating around the average and all of the data points are within the limits, we'd say like whether we like this level of performance or not, at least. The good news is that it seems to be predictable that the next month the data point, unless something changes, is likely to fall within the lower and upper limit. So, you know, the first rule for finding signals is looking for a data point outside of those limits. Then we would ask what changed?

What happened?

Yeah. And if we're looking back over the last 24 months and we see one data point last November that was above the upper limit, we'd want to ask if anyone really knows the answer. What was different about November? Was this something we were doing differently? Was it some outside factor? It was a strong effect, but apparently temporary because the chart went back to fluctuating within those limits. But then if you see a shift of eight or more consecutive data points that are all above the average or below the average, there is a sustained change in performance. Where again, we would want to try to ask and understand, well, what was different in that system. So the process behavior chart methodology with a little bit of math and a lot of visual.

Visuality.

Oh, thank you. That's a Gwendolyn Galsworth word, Visuality. Yeah. Third time's the charm. You know, it helps us. It eliminates the guesswork of, well, is that data point. Are those data points worth reacting to to detect? Has the system changed whether it's in a good direction or bad?

Yeah. I'm glad that you mentioned earlier on the difference between these. What some call what these. What would some. The control limits. If you plus or minus three standard deviation or approximately. But from specification limits because there's a lot of confusion. So, like the Six Sigma people out there are all about their cp, CPK and all this kind of stuff. And these are using your upper specification limit and your lower specification. This is from your customer who says, thou shall not send me any part greater than this number or less than this number. If you do, that's a defect. Right, right. And so, but there's a lot of confusion out there on the difference between specification limits from our customers and control limits, which are calculated based on the variation of your data points. So I think that's a really important differentiation.

And you would ask, is your process predictable and is it capable? Yeah. Right. So you compare the spread of those control limits to the spread of the specifications or, you know, so it's one thing to use control charts like manufactured parts and the diameter of a hole that you're cutting in metal or started my career at General Motors. That was the type of thing we'd be measuring. But you could also be measuring output of a system. You could be measuring something like a throughput rate or a defect rate. And you might have a goal or a target. And I think there's this sticky situation where let's say if it's a bowling chart, list of numbers, or even with a control chart, I've seen this a lot in healthcare. People set goals that aren't really a great stretch. They might look at last year's average and say, well, let's try to be 3% better. So now you've got this goal set kind of near the average. And if you have a red and green threshold, that metric is going to kind of bounce back and forth between red and green.

Yeah.

And there are some of these rules of thumb that just aren't real helpful. Again, like if you're trying to root cause analyze every red data point, if those red data points are noise, it's just not a good use of time. And then I've seen one organization in lean land that teaches. They'll say, okay, well, we're not going to react to one red. The rule of thumb they teach is, well, if there's two consecutive Reds, now we're going to do an A3, we're going to do a root cause analysis. And that could be just as statistically invalid as reacting to one red data points. So there's a lot of old habits to try to unlearn. Well, sure, our target and our goal is important, but if we're fluctuating between red and green, the pathway to improvement is not understanding that Last data point. This is where we need to use some systems thinking or improve less reactively and more systematically. How do we shift performance up? Because the same system that generated the red data point last month also generated green the month before. It's not a matter of what went wrong last month to make it red, but we have to step back and ask that broader question of why is our system not capable currently of delivering better performance? And I think we spend all this time chasing the reds, doing this sort of firefighting that just, that's time that we should really spend on more systemic, systematic improvement.

Hey, Mark, what is your thought on frequency of measurement? When we're coaching people through our certification programs here, and they'll often have some sort of control chart or process behavior chart where they're monitoring with a goal. But we typically, and I'm curious if you agree, we typically tell people to shy away from monthly data whenever possible. Because if you make a, implement a countermeasure to try to make it better, you're gonna have to wait three, four months just to see if you made it any better. Whereas.

Or maybe eight months, that's very slow.

Right. So whereas if you're able to get, you know, weekly data, which is probably the norm for us, or even depending on the process, daily might even be appropriate or possible. What are your thoughts on that?

I mean, I think generally speaking, a faster cycle, shorter cycle metric is better. There might be a bit of a trade off around the cost of compiling the metric a certain way. But I can use an example. This is the last project I was working on on site back in 2019. Hospital Surgery Center. They were looking at kind of like the end result metric they were looking at was patient experience scores. Okay. Very. Traditionally those are compiled and summarized as a monthly metric. And that's kind of like force of habit. The data is available to break it down into a weekly metric if you like. But then what you see in the process behavior chart methodology helps us with this. Actually, there tends to be more week to week variation in a metric than there would be month to month variation. Sure. You know, the longer time of a month tends to sort of smooth out some of that variation. So it's very important. If we are converting a chart from monthly to weekly, we do the calculation of the limits based on the weekly numbers. How much is that fluctuating? Because you have fewer data points, smaller number of patients, there's going to be a more inherent variation. But again, the chart shows us that. So we were trying to shift to say at least For a period of time here, we've got to look at the patient experience scores weekly. So to your point, Ron, you're totally right. You want to see the needle move more quickly. And then there were some leading indicators, the amount of time patients were on site from arrival to the start of their procedure. We knew from patient feedback that patients felt like they were waiting too long. And there were some things baked into the process of, like, well, if you tell every patient to show up at 6am Regardless of their procedure time, some of them are going to wait a long time and it doesn't have to be that way. So we could challenge that habit. And we put a bunch of process improvements in place as sort of a package that we went and go and we went to pilot. So we were looking on a daily basis the average patient waiting time and then how that would contribute then to patient experience scores. And we could see very quickly on the daily waiting time number that that had shifted down quite a bit. Not ideal, but much better, out of the range of just fluctuating around the old average. And then we had to wait, though, to see if this hypothesis played out. We knew there was a connection. It seems, I think, reasonable that shorter waiting times equal higher patient experience scores. But we didn't know how much better the waiting time needed to be to really adjust experience scores. Because there's a perception thing.

Yeah.

So you could say like, well, rationally, the waiting time is 42 minutes less. You should be much happier. Well, not if it's still longer than they expected. So we did see a shift upward where that first patient experience score data point was above the old upper limit of the fluctuation. And it was perfectly fluctuating around an average in the past. And so that gave us confidence that we weren't just nibbling around the edges of a process and hoping to make it better. I think we had actually gotten to some of the core of the waiting time in a way that reflected in really bumping up that satisfaction score in a way that wasn't just better than the old average. It was a number that wouldn't have been generated by the old system.

Wow.

Yeah.

Super cool. Red bead experiment. I know you're going to do this in the workshop and it's famous, been around a long, long time. For those that aren't familiar with it, what is it and why should anyone care?

Yeah. So the red bead experiment, also known as the red bead game, maybe that sounds more fun game, the red bead game. But if people are like, oh, we can't play games. This is a training class. We'll call it an experiment. This is associated with again, Dr. Deming and you can find, thankfully, even though Dr. Deming passed away 30 years ago, video of him on YouTube talking about and facilitating some of the Red Bead game. It's fun to go watch that. The game was actually created as a gift to Dr. Deming is the way I understand the history. Somebody who worked at hewlett Packard and Dr. Deming had been working with them and teaching systems thinking and continual improvement. And I'm sure control charts. SPC was part of the discussion. Some person from Hewlett Packard. What a delightfully nerdy gift. Like here I've created a learning simulation for you.

I never knew that.

Yeah. Wow. And Dr. Deming used that. I think it was created in the 80s and Dr. Deming was teaching right up until he passed away. So it's associated with Dr. Deming and it's something. Other people still use it today. I'm not the only one. But it's a hands on experiential way of trying to build some new intuition and sort of poking holes in your old habits and old intuition. So it's one thing to hear about it, it's one thing to read about it. There's a chapter in Measures of Success. It's kind of a narrative of what tends to happen. There's no substitute for actually playing this game. So we have six production workers from the class and there's a bucket full of beads. 20% of them are red, representing defects, and the rest of them are white. And you dip this paddle in and you kind of obnoxiously give this overly specific engineered standard work of how to do this. And then you dip in the paddle. And out of the 50 beads, you would predict on average about 10 of them would be red. 20% of 50, 10. But it's not always, it's very rarely actually 10. That number generally ranges from 3 to 17. It's easier to predict the average than it is to predict the variation. And so then through the course of the game, you know, people are kind of in on the joke. It's like it's kind of a analog random number generator with a little more nuance than what a computer spreadsheet random function would generate. But you know, people are trying really hard and like, as a facilitator, I'm role playing of like kind of, you know, demanding performance. Try harder.

What's wrong with you?

Quality. You know, we put up posters and slogans and I, you know, in the first round, give somebody an employee of the week award, and I put somebody else on probation.

Oh, don't give it all away, Mark.

You know, but kind of poking holes in, you know, these ways that people generally try to manage, and people are in on the joke, but they do get really competitive. Like, they can. I don't know if it's, you know, PTSD from other jobs. Like, you know, I don't think I've. I don't think I've made anybody cry. Yeah. But people do get really into it, and then we build upon that. Not just to teach how to create and use a process behavior chart, but we can talk through some of the management piece. So, again, back to the workshop. It's not just better metrics. It's better metrics for better management. The red bead game is such a great way to open people's eyes, and I think of doing this class in a healthcare setting. I was the chief medical officer from a big academic medical center who was a participant. And during the debrief, he was very thoughtful, and he said, you know what? I think all of our patient safety metrics are red beads. The number of falls, the number of infections, the number. These numbers tend to fluctuate around an average, but yet hospitals tend to be very reactive, and there's a better way. So can we try to convince anybody that there's a problem with the current state management method? That's one of the big challenges here. Right there is this different playbook that some organizations are using. But a lot of times people are just comfortable with the status quo. They don't think that's where the problem is in the organization. Or I've had people say quite literally, like, well, we can't move away from these bowling charts because that's what our lean sensei taught us to do last year, and we just implemented that, and we can't admit that there's something better. So what do you do?

Yeah. Yeah.

Wow. Yeah.

Super excited for it. The last question I have, Mark, and I don't mean to end it on a downer, but maybe we could spin it around, but.

All right, go ahead.

You know, have you experienced any sort of pushback to just the whole idea of process behavior charts within our. Our community of improvers?

Well, yes, and I mean, I sort of took it in that little direction around the change management challenge of not just trying to throw solutions at people. But can we come to agreement on the problem statement? And is there a problem that's worth working on that's actually worth changing. I mean, I've heard a number of things that were either kind of offhanded, offhand comments, or more serious. I remember once somebody asked me, oh, so now you're going Six Sigma? I'm like, well, no, I mean process behavior charts, statistical process control. Long way before Six Sigma, right? I mean, someone could say, oh, what are you going tqm, Honestly? I mean, control charts as a method are kind of a standalone thing. Sure. It gets taught and used in the context of Six Sigma. And I'm not ripping Six Sigma here, but my point to that person was that, well, we can incorporate, I think we should incorporate process behavior charts into our lean management. Toyota uses control charts as part of the six basic QI tools that have been around for a very long time. Toyota's not afraid of. I mean, they incorporate this method and I think other organizations could as well. I got a strange piece of, of pushback once somebody read the book and said something like, well, at Toyota, you shouldn't be measuring something like quality, you should be improving it. I'm like, well, how do you know it's. And I don't think I'm misquoting this person. It just didn't make sense. I'm like, I don't even know how to continue that discussion. How do you know if it's improved if you're not measuring it? Now this kind of like, well, wait a minute, it's a dumbing thing. You started off poking at that. I mean, again, measurement is important, but it's not the only thing. Measuring something doesn't guarantee that it's going to improve. So I'm not saying, well, just measure it and watch the chart. It's got to be connected to our improvement. So that was another piece of pushback. And then I think again, I think a lot of times people are like, oh, well, we're comfortable with bowling charts. Our leaders like bowling charts. I mean, if leaders like reacting and lighting a fire under people, they get more opportunity to do that. With the bowling charts, I don't think it's really all that helpful. And then there's one other piece of pushback I can think of. You calculate the three sigma limits plus or minus around the average, and people look and say, well, there's no signals. I don't have anything to react to. I'm like, well, exactly. Good. Okay, now let's go improve and try to shift performance. And sometimes people will say, or I've seen people do it. Oh, we set the limits at two sigma. So we had more to react to.

Oh, wow, I never heard that. Goodness.

That's just flying in the face of what the methodology is supposed to be. I mean, I would say, I mean, I try not to take too hard of a line, but I would say, please don't do that.

Yeah, I mean, just. And I might have this wrong. So the statistics expert come flying at me here after I say this. But if memory serves me correct, the. The reason we go with plus or minus three standard deviations, that's not exactly three, but three point, whatever it works out to be is what is it, 99.73% of all the data points should fall between those. It's something like that, roughly.

Yeah. And you're trying to strike a balance between not having too many false positives and also not having any false negatives. So plus or minus three sigma. And we can calculate estimate sigma in different ways with the different control chart methodologies. But yeah, I mean, you know, let's. I mean if. Yeah, we don't want to artificially invent more reasons to react. Now it might be, you know, reacting to two sigma variation might be slightly better than reacting to every up and down.

Yeah.

But it's. Well, it's just not really the way to do it.

It's just like, you know how its specification limits can be unusually tight. It's like, like you mentioned a machining thing. It's like, I've seen it before to where it's like, okay, the engineers say that we've got to have plus or minus this for our specification. And I'm like, is the customer asking for that?

Right.

Well, no customer's asking for this, but we're going to be. I was like, well, what the heck, man? That's over processing. Right? One of the seven deadly wastes.

And that's why I said earlier, you know, the target, target or the specification limits may be the voice of the customer. It might be the voice of management, the voice of a quote unquote arbitrary target. So like in healthcare, the organization might have a goal that 90% of patients have a quote unquote top box experience score of 90% of patients will give a 9 or a 10. Like what? That's just kind of all. I mean, like, so we're going to be happy that 10% of patients didn't have. Only 9% of patients had a kind of crummy experience. Like, that's far from perfection. Yeah. Yeah. Wow.

Good stuff. Well, thanks for coming on again, Mark. Thanks, I appreciate it, man. And I love doing these in person. I'M so thankful that you're able to come in from time to time when you're. Yeah, when you're around. So how do people connect with you? What's the best way? LinkedIn.

Where are you at LinkedIn? My website's markgraven.com the new book, if I may, is a mistakesbook.com audiobook now available now. I have a hypothesis that podcast listeners also enjoy audiobooks. So that's available now through Audible, Amazon and Apple Books.

Do you read it? I haven't listened to it.

I read it. So nice. Yeah, I love it.

It's about five when the author reads it.

Yeah, it's about five. Went into a studio, it was about. It ended up being about five hours of listening. It took about, I mean there were breaks involved. It took about 10 hours to record what ended up being about five hours because I slipped up and yeah, the engineer caught most of them or I would catch them. And I did have to come back in and do a little bit of rework after he did editing. Sometimes you just say the wrong word as you're trying to read smoothly and looking ahead, but you know, looking ahead. Really excited to be able to do the workshop here at the Gemba Academy building headquarters facility. Yeah. October 10th and 11th. I hope people will come. Sign up.

Yeah. Gemba events.com and we'll have everything linked below. We've been talking about a lot, so. All right, well, hey man, I appreciate you coming on and yeah, until October, we'll see you.

All right. Thanks Ryan.

Thanks for listening. Whether you've been on the continuous improvement journey for many years or perhaps you're just getting started, Gemba Academy is here to support you. And while we're best known for our more than 1500 Lean and Six Sigma teaching and virtual tour videos, we also have a team of experienced Lean and Six Sigma practitioners available for one on one coaching as well as a variety of Lean and Six Sigma certification options. To learn more and to schedule a demo, head on over to gembaacademy.com.