Podcast: Why data collection in machine shops is worth the effort
Key Highlights
- Accurate, real-time data replaces guesswork and quickly boosts utilization and productivity.
- Standardized tech stacks enable faster, smoother digital rollout across multiple plants.
- Data-driven decisions uncover small efficiency gains that compound into major savings.
- Monitoring both high-mix and high-volume work helps maintain margins and improve setups.
In this episode of Great Question: A Manufacturing Podcast, Mike Payne, president and owner of Hill Manufacturing & Fabrication loves data and metrics. If he can pull information from a machine, he's doing it. Mike talked with IndustryWeek's Dennis Scimeca about why he collects all of that data, what he does with it and why, despite all of this technology, succeeding in manufacturing is still all about relationships and performance, not technology.
Payne noted that before data collection, managers would assign tasks to machinists with a time estimate. And, wouldn't you know it, that's how long it turned out they needed to make each part. Was the machine in use that entire time? How much of that time was setup vs. production? Could the company have cut those time allotments by 20% without harming quality? Without data collection to monitor what was really happening, nobody knew.
Below is an excerpt from the podcast:
DS: Hello, everyone. My name is Dennis Scimeca, Senior Editor for Technology at Industry Week, and my editor-in-chief, Robert Schoenberger, has allowed me yet again to hijack the production pulse tonight. We're talking with Mike Payne, who's the president and owner of Hill Manufacturing and Fabrication. We ran into each other a couple of months ago at a conference, and Mike told me this, I thought, fascinating story about how he's wiring his plants up for data capture and what he's doing with the information. So I wanted to bring that story to you. Before we go any further, Mike, would you like to introduce yourself to the audience?
MP: Sure. First of all, thanks for having me. I've enjoyed—I think this is, what, maybe our third time we've talked?—and I've enjoyed it every time. So thanks for having me on.
Yeah, I'm Mike Payne with Hill Manufacturing and Fabrication, which is really kind of an umbrella for five shops that we have all in the Northeast Oklahoma area at this time, looking to expand that maybe a little bit. I'm also the host of a couple of podcasts, Making Chips and Buy the Numbers. So Making Chips is a more generalized manufacturing podcast where we try to equip and inspire manufacturing leaders. And then my By the Numbers podcast is spelled B-U-Y because a lot of my background is in finance and accounting. So it's all things finance, accounting, data-driven decisions—that type of stuff.
DS: As I mentioned, Mike is not just a podcast host; he has networking new plants down to a science, which is why I want to speak to him about the how—and, more importantly, the why—of capturing data. So, Mike, when we first met, you were talking to me about, I believe, the very first plant that you purchased, in 2018, and you were saying how you didn't have any data capture at that plant when you first started. And when you first turned on machine monitoring—when you got the plant wired up—you said you were running 30% spindle utilization. And when they began capturing and analyzing data, that changed to the 50% to 60% range. Now, to what do you attribute—that’s like double—to what do you attribute that change just by looking at data?
MP: Yeah, I think—I mean, everybody's heard their entire careers: what's measured matters, right? And when we weren't tracking it, it didn’t matter. When we talk about not having data capture on the front end, I mean, I guess we did have data capture; it just was garbage in, garbage out, right? It was a paper-based system where we’d give an operator at the machine, like, “Here's this job you're running. It should take eight minutes a part.” And remarkably, how long do you think it took? Eight minutes a part, right? Because they just filled in the blanks: “Well, I was here eight hours and I got 64 parts,” or whatever. It might have taken them eight and a half; it might have taken them seven, but there was no tie to that data. It was just handwritten data.
So our first step into capturing data was implementing a good ERP system, where we were really capturing the time spent on essentially everything, right? We’ll log time even against Kaizen events—how much time are we spending on Kaizen, how much on training, how much on an actual job, whether that's setup time, runtime, first-article time. That gives you an early look. And I would say even with just that, things bumped—but probably not a ton from an efficiency standpoint—because you were capturing better time. You were getting more accurate data. Some were high, some were low on a paper-based system. You know, they level out.
But then the missing piece—you mentioned the machine monitoring. Once we get that, and it's tied to our ERP system, it now knows what we expect to happen based on history, estimates, all that type of stuff. But now you're getting what really happened. And you throw that data up on a screen, and naturally things just get better.
Part of that is the old Hawthorne effect, right? If I see how I'm doing, I'm going to do better. I mean, if you played an entire football game without the scoreboard on, some people might know the score, some might not. But if I see I'm down by three with two minutes left, I’m going to act differently than if I have no idea.
Same thing in a machine shop. Someone’s looking up there, and our machine monitoring gives you a grade—this machine is running at an A, a B, a C, whatever. And just the natural competitiveness in a human being—most people want to perform at an A. We all want to be successful. So I think just that natural effect created that immediate bump from, say, 30% to 50%. Then, when you can start dialing in why you're not at 60, 70, 80—and you have the data to make good decisions—you can start seeing those improvements.
DS: So this is the first plant—excuse me—between 2018 and 2021. That was how long it took to get the systems and processes in place for this first plant. Can you talk about what was happening in those three years? Is there a need to break that down? What was the journey before you got the data and were really able to start rolling with it?
MP: If we back all the way up to my history, 30 years ago when I came out of college, I started a software company, and I actually did shop-floor data collection. So I've been around data collection my entire career. And here I am—I come in, I buy this shop. I can see the data in the financials, but I can't see the data on the floor and what's affecting those numbers. So yeah, I just immediately start looking for opportunities to get more information so that I can weigh options.
Like I mentioned, the first step for us in 2018 was getting into an ERP system that allowed us to start collecting a lot of that data. Then from there, we just started drilling down and going, okay, if we've got this data, here’s where we're still lacking some data, right? The ERP maybe gave us today's performance on a job. Machine monitoring gives us this minute’s performance on a job, right? So where I used to know how we did today, now I know the first time we run a part how we're doing on that.
So when you look at that three-year period, it started with collecting some data, deciding what data was important, and then improving that data. It was an evolution over that three-year period to get there. And of course, we had COVID punch us in the mouth in the middle of that. It changed a lot—although in some ways I would say it changed us for the better, right? Because we had a serious blow to our revenue, but it gave us time to work on ourselves too.
We made it through that without having to do any layoffs or anything like that. So what we did is we were able to use the same labor we had to make ourselves bigger, stronger, and faster. And a lot of that was through better data collection. Even where I talk about an ERP system—being able to track Kaizen events, right? How much time are we spending on improving ourselves? Are we spending enough? Are we spending too much? Things like that.
Or the money and time and effort you spend on having a tool room manager—do we see that in a reduction of setup times? Those types of things. That data gives us what we need to make decisions.
DS: So you said you have an ERP system. What were the other kinds of software or hardware that you installed in your first plant during that three-year period?
MP: Yeah, so we had to start with—I think it's important for anybody looking at their tech stack, if you will—you’ve got to start with your communications platform. One of the first things we had to do was upgrade our network. Whether that was hard lines, soft lines—I mean Wi-Fi—servers, all those types of things. Historically, it had maybe just been patched together. It's like, “Hey, here's this old computer I brought from home; we can use it here.”
So starting with that infrastructure of having good systems in place, standardizing some of the equipment we use—especially on the network platform side—was critical. If you're going to be digital and you don't have a good infrastructure, you're going to be down a lot. You're going to be disappointed in it. So we had to start there.
Certainly, the ERP system was the next step, getting engaged with ProShop ERP and being able to leverage that to start collecting data. And then you start filling in the holes. Sometimes it's an Excel spreadsheet, sometimes it's a Google Form, sometimes it's taking the leap into a powerful machine monitoring system. All the way down to our inspection—having the inspection software we needed, communicating with the ERP system—all of those digital investments we had to make. It took time to find the systems that worked for us, but fast forward now almost eight years, I think we have it dialed in pretty tight.
DS: You've told me that you’ve since bought one plant per year, and your second plant took you 9 to 12 months to get online. And every plant since has been in the three- to six-month range. Does that mean you've really dialed in your tech stack—your suite of tools—and you find they can be applied to other plants? Do you have to make a lot of adjustments per plant? Do you really have it dialed in?
MP: I mean, I feel like we have it dialed in pretty tight, right? So we have—if you listen to me on By the Numbers at all—you’ve heard me say I have my secret sauce, right? I mean, it's not actually secret. I talk about it on the podcast all the time. It’s not that it’s a top-secret sauce; it’s just my sauce, right?
And so, yeah, even down to the brand of Wi-Fi routers we use, all that type of stuff—it’s all standardized. You know, the tablets and the PCs that go on the shop floor are the same. And so you end up, from a hardware and infrastructure standpoint, being very standardized and very easy to implement. And then rolling that out, doing the training—now that we have, we're 70-something employees now, say 75, we'll call it—you know, when everybody knows the systems and the processes that we use, even from a training perspective and a rollout perspective, it just keeps getting quicker, right?
So even the most recent acquisition we did in July—I mean, I can sit here today and tell you it's frustrating; it's taken so long. I mean, the reality is we're 4½ months into it, and it feels like that's forever, and we're essentially done with it. But there are still little hiccups here and there, and a little bit more training we need to do on this side or that side. But yeah, I think the more you dial it in and the more your team—your people—and your systems develop your processes, you can really get down to a pretty quick implementation plan that, in the grand scheme of things, is really not that long. Even if, like I said, I sit here feeling like it is—it’s not.
Yeah, I mean, I would say in general, no. From an “unable” standpoint? Absolutely not. This most recent one, for example, there's been a few little things we haven’t really dealt with in the past that we didn’t have a system for.
DS: I think three to six months seems pretty quick to my ears. Were there any solutions that you weren’t able to port from plant to plant? Something that needed—because I'm imagining the manufacturer listening to this who has multiple plants and wants to get going on a similar type of journey—were there any tools that you weren’t able to easily standardize from plant to plant? If there's going to be kind of a log jam, or something that needs a monkey wrench, what is that?
MP: So, for example, in this new acquisition, we manufacture a machine for the packaging industry as part of the operations, and we sell parts and supplies for that machine online. We accept credit cards for payment and then create a pick list and ship it out. Historically, our companies have not had online sales, and they haven’t had same-day UPS pickups of parts. We haven’t had significant credit card sales, and we haven’t had “order by 2, it ships today” type methodology, right?
Well, there are systems behind each of those things that the company we purchased already had in place. So we've had to integrate those things, which has been a bit of a learning curve, because again, we didn’t really take credit cards on a daily basis. We’d take credit cards maybe three times a year. So we need to integrate that into our entire process of accepting an order and shipping it.
But again, there are tools out there to do it. I mean, it's really as simple as getting with your bank and finding out what merchant account you need, getting it done, and then integrating it with the new systems. But when you look at your core—as far as, again, I'm in manufacturing, so I need to accept an order, get it done, have quality, ship it, and invoice it—the core function of that we have pretty well established. So it's just a matter of a few systems to help with that.
DS: Can you speak to your customer base a little? Meaning, do you have a lot of repeat customers, long runs, or a lot of short runs? Since I understand short and long are relative—just wondering whether the data collection is more helpful for one or the other.
MP: Good question. Yeah, so across our four facilities, I would say we have a pretty big mix. I would say 90% of our business is repeat customers. Now, the long-run/short-run question—that’s where the mix comes in.
So we have Soothill, where I'm sitting today, which is definitely a high-mix, low-volume shop. And I would say all of our other locations are really more high-volume, low-mix. Ninety percent of what they run, they ran a couple months ago.
As far as how the data affects that, I would say it affects them both—but it affects them in different ways. On our longer-run production stuff, that’s where we can really take data and dial in. We can literally look and see: if we change the brand of a drill, did it affect the outcome? Did we save 30 seconds on a cycle time? Three seconds on a cycle time? Whatever. That’s one way we would look at the data on production-style parts.
On the shorter-run stuff—I would almost say the data probably helps us more, right? Because we may have just bid this job. We have targets we need to hit. Being able to track that in real time—through machine monitoring or whatever—and knowing it came off the saw late, right? Maybe it's going to go to three or four machining or finishing operations. If we know it took longer at the saw than expected, we can start to fix that job. We can start looking for improvements so that we hit our budgets, hit our timelines, and so forth.
You know, we're probably not looking at the data as much as “if we use this insert or that insert” on a turning operation or drill. We're looking at a bigger picture. But those jobs can also make or break you. In what we do, most of the cost is in setup times. So if those start getting out of hand, the quicker we can see that, the better.
Also, the systems give us plenty of ability to see—even going back to tooling—picking the right tooling, setting up the right tooling, all those types of things. The more data we have driving those decisions, the better.
DS: I want to take a quick jaunt into something different: infrastructure work. I'm curious: how much infrastructure work do you have to perform at a new plant on average?
MP: It's generally not enough— not much. Like I said earlier, we'll go in— I mean, some of them we've bought have had fine infrastructure to start with, and you can really just hit the ground running. We generally will normalize on a set of, from a hardware perspective, you know, particular brands, and that’s so the IT folks can manage everything under one platform and so forth. The cost on that— a lot of that stuff has come down so dramatically, at least in my opinion, since I've been dealing with it for 30 years now. It just almost seems like it's a commodity, right? So the benefits of having a standardized platform tend to outweigh the cost of that. But it's usually not prohibitive by any means, and it's not that difficult. A lot of people already have a lot of network cabling in their buildings. I mean, that's been going on for a long time. Most of what we do nowadays, honestly, is Wi-Fi driven anyway. So, you know, making sure you have that good coverage out there generally answers most of the questions.
DS: So I assume you're looking at KPIs like OEE and yield and the kind of, you know, basic stuff that we're getting from these systems. So if it's taking you less and less time to install these systems in your plants, are you seeing improvements in your numbers happening faster, or is it still kind of the same amount of time regardless of what you're putting in?
MP: I'm getting real data on day one now, right? So, you know, again, I'll reference the shop we bought most recently just because we're kind of in the throes of it. It's eighty-five percent probably repeat work, and we've had it for four and a half months, five months, something like that. Given those two things, we've now run almost every job we’re running once in the system before, right? So we now have something to benchmark against. Prior to that, we didn't know, right? But now we have something to benchmark again. In another three months, we'll be able to benchmark again and we'll be able to start really seeing improvements.
But to give you a real-life example: we had a part last week that we were running for the second time. They've been running this part for probably 20 or 30 years. Three-minute cycle time on a lathe— I mean, you aren't going to get much quicker than that, right? Probably not a lot of room for improvement. But because we had data to look at— and an inquisitive person, right? I mean, it takes someone caring to look and go, “Can this be improved?”— the operator with my main manufacturing engineer sat down, looked at some of the data, and said maybe if we changed this one part of the… they changed the speeds and feeds. I don't even know what all they did, honestly.
But we took a three-minute, 15-second cycle time down to a minute 58. Well, that doesn't seem like a lot, but that's a 30% improvement, right? I mean, it's a minute. It can seem insignificant, right? But it's over a 30% improvement with no change in tooling costs, no change in anything— just by nature of having data to look at. And over the course of that job, it runs 1,000 parts at a time, right? I mean, that's 1,000 minutes dropping straight to your bottom line. And it's having that data that makes that available.
DS: Could you tell us again what your podcasts are? If somebody wants to go hear you, hear you talk more about manufacturing, where are they going?
MP: If you go to makingchips.com, you'll see all of our shows. We actually have a total of nine shows in the network, all around manufacturing. So MakingChips, and then my show is Buy the Numbers, spelled B-U-Y. You can find it on the website, you can find it on Apple, Spotify— all your podcast platforms. You can find us on LinkedIn. Just about any source that is standard for podcasts, you'll find us.
About the Podcast
Great Question: A Manufacturing Podcast offers news and information for the people who make, store and move things and those who manage and maintain the facilities where that work gets done. Manufacturers from chemical producers to automakers to machine shops can listen for critical insights into the technologies, economic conditions and best practices that can influence how to best run facilities to reach operational excellence.
Listen to another episode and subscribe on your favorite podcast app
About the Author
Dennis Scimeca
Dennis Scimeca is a veteran technology journalist with particular experience in vision system technology, machine learning/artificial intelligence, and augmented/mixed/virtual reality (XR), with bylines in consumer, developer, and B2B outlets. At IndustryWeek, he covers the competitive advantages gained by manufacturers that deploy proven technologies. If you would like to share your story with IndustryWeek, please contact Dennis at [email protected].