An improvement project, delivered...

Improvement Insights Blog

An improvement project, delivered…


pizza-4457006_640 560x371.jpg

I can’t recall a time when I’ve done so many improvement projects in such a short time.

In just three short weeks, I’ve rearranged my kitchen to a more efficient layout that saves steps when I’m preparing a meal (after creating a spaghetti diagram of the food preparation process), I’ve reorganized the spices located on the rack on the counter as well as in the cupboard (moving away from a strict alphabetization plan to a two-tiered system based on usage, where the most-used spices are stored in the rack on the counter and the less-frequently-used spices are stored in the cupboard). I’ve even restructured my clothes closet to make getting dressed more efficient, grouping the more formal items in one space (three items on hangers in the back of the rack) and grouping more casual items together in another space (the entire rest of the rack).

In short, home quarantining is driving me up a wall… and it’s only been three weeks.

I can’t complain too badly. I’ve been working from home, and I’ve been able to maintain contact with my Quality Improvement clients via email and video meeting. I’ve been in contact with my friend Annie, a nurse at a local hospital, quite a few times. (Remember Annie from last month’s story?) I helped her hospital design a contingency plan to prepare for the case load that’s only now beginning to come in, and helped her ensure that patient flow was as efficient as it could be and the layout of the facilities would allow for smooth and rapid movement of the healthcare providers as they cared for their patients. (I can’t say enough about Annie and her team… really, about all the medical professionals out there. They’re just incredible.)

However, when an improvement project showed up on my doorstep – and I mean that in the most literal way possible – I jumped at the chance.

My own cooking was getting boring, so I had called in a pizza delivery order to Pareto’s Big Bar, my favorite statisticians’ watering hole. The governor of our state shut down in-person dining for bars and restaurants, but takeout and delivery were still available. I saw the driver leave the order on my front porch and ring the doorbell, but when I opened the door I saw something I didn’t expect. In addition to the pizza was a six pack of my favorite frosty beverage with a note taped to it. The note read, “Consider this an initial payment for your services… or a bribe, whichever you prefer. I need help. Call me at 9 p.m. – John”

Now sharp-eyed readers have already guessed that the “John” who sent the beer and the note was none other than John the Bartender, who ran Pareto’s Big Bar. I’d seen John regularly at Pareto’s ever since I moved to town many years ago. (Some would say that I’d been in that bar TOO regularly… those folks would be wrong, but that rarely stops such folks from saying such things.) Needless to say, at 9 p.m., I dialed his number. He answered on the first ring.

“I see that you found my note,” John said in his heavy Boston accent.

“Hey, John,” I laughed. “I applaud your style. Taping the note to my favorite six pack is a heck of a way to make sure it gets noticed.”

“I’m a fan of taking all necessary steps,” he replied. “In this case, getting your help is necessary. I don’t know what I’m doing here.”

“Let me email you a link to connect to your computer so we can both see what’s going on,” I offered, and he readily accepted. Once connected, I asked him to elaborate what the issue was.

“Well, I had to take all of my wait staff and bartenders and put some of them on the phones taking orders, some of them into the kitchen helping out the cooks, and some of them are delivery drivers. That’s all well and good, but I’ve got to show the owner that we’re still operating efficiently.”

“Got it,” I said. “So I assume you’ve been tracking data?”

“I’ve been listening to you go on and on for years about tracking data, so yeah, you could say that a thing or two has penetrated through this Red Sox hat,” he joked. “I track the time the order gets entered into the system until the time it’s ready and the driver heads out the door. I even put it in a chart like you talk about… but I don’t know what that tells me.” John pulled up this chart:

0420 XmR chart 560x441.png

“All right, John. Congrats on tracking the right data. From this chart, it looks like you’ve got a really stable process. However, we’re going to need to use a different chart in order to analyze this.”

“Uh huh, I knew you were gonna say that, and I bet I know what you’re going to say next: You’re going to tell me that I need one’a them History Grandmas,” John interjected.

“Uh… if you mean ‘histogram,’ then yes, that’s exactly what I was going to recommend,” I replied. I could see his computer screen changing as he brought up another chart.

“Same diff’. I ran one’a them too. Now you told me once that the bigger those Cp and Cpk numbers on the right were, the better it is, right?” I could hear the frustration in his voice. “Well, these numbers don’t look so good.”

0420 Two Sided Histogram 560x408.png

I examined the chart for a moment. “John, I noticed you entered 0 and 45 for the spec limits. Why is that?”

John replied, “Yeah… I figured that 45 minutes is about the most that people are willing to wait for their order to be ready, and it just makes sense that 0 would the the other one.”

“Yes, that sounds like it makes sense… but it isn’t exactly right. What you’ve got here is a two-sided histogram. I think what you need is a one-sided histogram, and maybe a one-sided frequency histogram.” I offered.

“Okay… what does all that matter?” John asked, puzzled. “The numbers are still the same, no matter what chart we use to show them, right?”

“Not necessarily,” I explained. “These charts are tools. If you use the wrong tool, you can get a result that doesn’t look right. You can start with two identical screws, but if you pound one with a hammer and drive the other with a screwdriver, the end result looks very different, right?

“Let’s say you were measuring the beef patties that your cook pressed out. Each one should be .25 lb, but as long as they’re not less than .2 lb or more than .3 lb, that’s okay. In that case, your Lower Spec Limit would be .2 and your Upper Spec Limit would be .3, right?” John mumbled something so I continued. “In this example, it’s POSSIBLE to make patties that are smaller than .2 or larger than .3, but a perfect process would have all the patties between the LSL of .2 and the USL of .3… and most of the patties would be exactly centered between those two spec limits: .25, right?”

“Okay…” said John, hesitantly. “Go on.”

“But that’s not the process you’re tracking. You’re tracking time from when the order is called in until when it’s ready to go. There’s two important differences here: First of all, it’s impossible to have data that’s less than zero; you can’t make an order ready before the customer calls, right?”

“Right,” said John.

I continued. “Second, a perfect process here wouldn’t have most of the orders at 22.5 minutes, which is exactly centered between zero minutes and 45 minutes, would it? If this were a perfect process, most of the orders would be as close to the zero as possible, with fewer and fewer orders as you go up in time, right?”

“Okay…” muttered John.

“What we’re describing is a process that’s best tracked by a one-sided histogram. Here, let me show you,” I said, taking control of his mouse remotely. “I’ll generate a one-sided frequency histogram first.”

“Wait… what’s a Frequency Histogram?” John asked, confused.

“It’ll be way easier to show you.” I replied. I selected the tab with his raw data and selected “Histograms and Capability,” then “Frequency Histogram.”

0420 Highlight data Freq Hist 560x425.png

I set the USL to 45, then the prompt asked me for the LSL. “See, there’s a difference between an LSL that’s 0 and a situation where there simply isn’t an LSL, like what you have. The program is telling me that if my situation doesn’t have an LSL, I should click ‘CANCEL.'” I did so.

0420 No LCL 560x309.png

“Now see this? Even though we asked QI Macros to display a frequency histogram, it’s analyzed the data and asks us if we want to see a plain ol’ histogram as well. We’re going to say ‘Yes’ just so we can look at both.” I clicked ‘Yes.”

0420 Also Histogram 560x343.png

“Okay, let’s first look at the regular one-sided histogram. Each of those bars is called a ‘bin,’ and it drops each data point into the bin it belongs in. You can see that you don’t send any orders out in less than five minutes, but most orders go out in just over 5 minutes, the second highest number of orders is just longer than that, the third highest is just longer than that, and so on. It’s not a perfect process, but it’s a darn good one, especially for a process that you came up with on the fly barely 3 weeks ago. Your Cpk is .82… that’s way better than you thought it was, right?”

0420 One Sided Histogram 560x409.png

“Wow, that’s good. Okay, so if this just a regular one-sided one, then what about the Freaky History Gramma that you were talking about?” John asked.

“The Frequency Histogram?” I clarified. “You’re just pulling my chain right now, right?”

John laughed. “Put it to ya this way: I haven’t been able to give you a hard time in a couple of weeks now. You can’t be sore at me for trying to get a reaction out’a you.”

“Fair point. Okay, the Frequency Histogram is like a regular histogram, but instead of separate bins spanning multiple data values, the Frequency Histogram sets up a separate bin for each data point’s value. If there’s multiple data points that have the same value, it stacks them all up so you can see how many of each there are. In this case, it’s set a bin at each minute. You can see that you got one order out in 5 minutes, 4 orders out in 6 minutes, nine orders out in seven minutes, and so on.”

0420 Freq Hist 560x410.png

“Aha! I get it!” John exclaimed. “This all makes sense. Of course, I’ll never remember this well enough to explain this to Willie. That’s why I’m so glad you’re going to be on the call with me tomorrow at 10 a.m. It’ll be 6 p.m. in Switzerland.”

I was dumbfounded. “Wait… what? Willie? As in, the owner of the bar, Vilfredo Federico Damaso Pareto?”

“One and the same,” said John.

“When did I agree to be on the call with Vilfredo Pareto?”

John replied, “I think it was right after I offered to send a six pack of your favorite frosty beverage once a week at half price, as long as you paid for your food order.”

I mulled it over. “You drive a hard bargain, Boston boy. I’m in. Give me a call tomorrow and I’ll help you explain everything.”

“I’ll do exactly that. Thank you so much, sir. Enjoy those beverages… and keep yourself safe and healthy, will ya?”

“You do the same, John. Talk to you tomorrow.”

If you’re interested in learning more about making Histograms and Frequency Histograms with QI Macros, click HERE and HERE to read about it on the QI Macros website.

This entry was posted by Jay Arthur in QI Macros Monthly Newsletters. Bookmark the permalink.