Listen to this discussion with Bill Sandblom on using AI to improve business outcomes at his PetroChemical where he is the CIO. He discusses tips for getting more of your team’s time focused on actual AI and Analytics work vs data prep and wrangling. He also shares advice for how to identify the right areas of the business to make a real impact on profits with machine learning & AI.
Transcript
Dave Mariani: Hi everyone. I’m Dave Mariani, Chief Technology Officer at AtScale, and I’d like to welcome you to the data-driven podcast. And today’s special guest is Bill Sandblom. Bill is the Chief Information Officer at the International Group. So I’m really glad to have you Bill. So Bill, welcome to the podcast.
Bill Sandblom: Well, thank you, Dave, looking forward to discussing where we’ve been with our data and the systems that we’ve put in over the last couple of years.
Dave Mariani: Yeah. You know, I got to know Bill through AtScale and, and I just, he, he showed me what he was doing, at the international group and I was really blown away. I was blown away because, you know, we all talk about using AI to improve business and improve business outcomes. but, what he showed me was, it was not, not just was he improving business outcomes, theoretically, it was actually dollars and cents wise. So he’s able to actually just show, you know, legitimately like how much money, his investments were making and have a true ROI. So I just thought, oh man, as people would love to heal here, Bill’s story. And just find out how he’s doing the magic that he’s doing. So bill, before we sort of get into all that fun stuff. just tell us a little bit about the international group. cause it’s a really interesting business. and this tell us about, about the company as well as, your role as the chief information officer.
Bill Sandblom: Yeah, it is quite, quite interesting company. I joined it 15 years ago and it was smaller than it’s been growing ever since then. we’re a wax producer, which is kind of unique. I never thought I would work for a tech, let’s say a company that made wax and, I’ve worked in manufacturing before I worked in mining for 17 years previously, but this company, makes wax and when I first started, we would buy, buy, buy a wax type product from other companies and then process it. But today is it’s evolved. It’s evolved where we have an interest in, oil fields. And then we have our own refinery that, then separates the crude into a diesel and a wax product that then goes to our wax refinery that gets made into the end product that we sell to, big customers.
Bill Sandblom: The big segments, our candles are still, have been and always, probably will be a big part of, of our, of our market. It’s, it’s the biggest, but just behind that is packaging and packaging is a bit of a broader area. it’s everything from having, your wax paper type of products to your boxes that you put a wax coating on to make them waterproof, to wax inside boxes, to when appliances go inside in corner pieces, they don’t scratch, just a wax on your turnips when they come out for Thanksgiving, we kind of say, that’s all kind of the packaging making products better so that we can move them, that type of thing. So that’s big, but after that, then there’s a hundred other uses for wax that you couldn’t even imagine this it’s in cosmetics. It’s, it’s in, for time release fertilizer, the coat, the coat, the little pellets and wax it released at different times. And it just goes on and on and on to the different products. When you talk to that to the salespeople, they say that, yeah, there’s a hundred different markets that we sell into, which is, which is good for the company because that, that gives you some, if one market’s down, hopefully another one’s up, those types of things. so from that sort of go ahead and meet,
Dave Mariani: You know, bill when you were talking about, just about, you know, about yeah, all the uses for wax, but just actually the process for making wax and those different products is actually really complicated and involves a lot of machinery. That’s very sensitive, and, needs a lot of monitoring to make sure that you’re, you’re making it efficiently.
Bill Sandblom: Absolutely. And that, and that’s where, when I started, when we started getting into the details of what, the benefits that we’ve had through BI and AI, that’s, that’s, that’s really that the heart of the matter as much as it’s helping us in a bunch of other areas, in terms of, where I finished the company started back, like I say, 15 years ago heading up the it department. at that time it was a real focus on upgrading networks and business systems and those types of things. we rent JD Edwards. So we upgraded that did a lot of did a lot of efficiencies there. we’re real vision. And I think, I mean, we work with, one of the major consulting companies and they kind of ranked us as their number one or two companies in just efficiencies and it’s, and it’s a lot of the things we’re doing now.
Bill Sandblom: Most companies have caught up, but 15 years ago we were, we were automatically sending out invoices, whether it was fax or email to our customers when most people were still stuffing envelopes. And, when we shipped, confirm a truck, we would have all the documents automatically go to the customer, to the border. If the truck was going across the border, just a lot of those things that we automated that weren’t there. But, but that’s kinda, there’s, there’s always things to improve your business flows, but that’s kind of for us, it’s done now. So then we got focused in, on, back into the production systems is where I’ve, where I’ve gotten back into, which I think is probably my first love because when I started fresh out of school, I was in the process control department at a mining company up in Northern Ontario.
Bill Sandblom: And those are the kinds of projects that I was working on. They had, things where we would launch balloons and track data because of the sulfur coming out of the stacks or writing programs to control the electric furnace, to minimize the amount of energy, those types of projects. So, they always told me I should go back to school and be an engineer, but I didn’t do that. As I said, I T’s pretty, I think it’s going to have a long run for me. So I stuck with that. So it’s kind of neat now that now back at IGI, we’ve taken the systems and continue to work on our business systems, but now have a real focus on, on the, the process systems and the plants is where we’re really spending a lot of time.
Dave Mariani: Yeah. I mean, I love that. I love the fact that you’ve combined technology and manufacturing together to put those two together, to make, to make the product. And so you’ve kind of have a background in love and both of those areas and it really shows in terms of your success and making it work. So, you know, so, so, you know, w you know, you talked a little bit about your path for, for getting to getting to where you are. I mean, did you know that, you know, right off the bat, when you were, in, in school, did you know what you wanted to do or how’d you end up in that being a chief information officer
Bill Sandblom: Well, I know when I went to school and my parents, it was like, you’re going to go to university. I said, okay, that’s good. And, and I went to university and the main thing I had focused on was to play basketball. So I played varsity basketball. I was there and at the site at the same time, trying to get my schooling done and work through that. And then, got working at the mining company and decided to do my MBA in the evenings. So I got that done. And I guess that set me up to work in the management role in it, I guess is that’s what I did. So I guess when I was 30 years old, I kind of had my, my sights set on, on working up the, the it structure within wherever company I was going to be in. And, and, I’ve worked at the mining company. I worked at a university for five years and now I’m here. So I, my list of employers is short lists, but, I’ve had three great places to work. And, this’ll probably be the final one. I would think it’s just a, it’s just a lot of fun. It’s what we’re doing. Works works as much fun as work these days with, with the stuff that we’re doing.
Dave Mariani: So, you know, you talked a little bit about, about, you know, about, you know, you’ve been, you being a leader and all of those organizations. So now what do you look for when you’re building out a team How do you, how do you find people you think are gonna work well and work well together
Bill Sandblom: I mean, they always talk about the, the first 30 seconds or two minutes of an interview. And you’re kinda like, yeah, that’s a purse. I believe like it’s the personalities of a bunch of it to start with. I mean, it’s, you get that feel that they’ve got that drive. And, and I mean, you’re trying these days, you’re almost half the interview. If you’re hiring somebody to sell your company, but at the end of the day, you’ve got to make sure it’s the right candidate, but you get that feel for the person that they’ve got energy. They want to make a difference, those types of things, obviously then you’ve got to get in and make sure they have the technical skills to go with it. But it’s about that energy and want to be a team player. Cause I’ve got a great team. I mean, if we wouldn’t be anywhere without the team and it starts with the servers and the networks and the securities, and then, and then into, the business systems, which you have to have, I mean, you have to be able to invoice and ship and order goods and you track your inventory.
Bill Sandblom: And now it’s like, we’ve got, it’s not like there’s almost a third division. Cause I always thought it was an apartment that was an it department and there was a business system. So we still have those, but now there’s like this analytical people and I’ve got probably a third of the team and we’re not a big team. We’re like, we’re 10, this is what we are. But about a third of the team now is very focused on the AI and the analytics, which wasn’t even a thing through your scope for us. So it’s, it’s, it’s things have changed that quickly and but it’s having people with that energy, it’s the energy to want to make a difference and be part of it and have fun, but work hard.
Dave Mariani: So when did, so when, so how, and when did sort of you decide that you wanted to make that investment in the analytics and AI Like, what was the, what was the impetus to sort of get you launched down that direction
Bill Sandblom: it was 2018 and, we have an annual management meeting where everybody gets together and it seemed the agenda and it kind of, it was coming even before it was there, but it was like, one of them things was we want better access to information. And, before that, before that even got there, I had, Oracle IBM and Microsoft. And just to talk about like, what are the ways we’re doing it And they all talked about a data warehouse. And I mean, I built data warehouses sorta like data warehouses, whatever way back and previously my career. But so what’s what, what is it all about today And they all said that we’ve got the fastest databases. I said, well, that’s great, but I don’t have that much data. I’m not a bank I’m on the manufacturing company, but we went through it and one of them gave me a quote on, and the first thing that the people were interested in was actually getting data out of our ERP to make it easier to get to that data.
Bill Sandblom: And so I talked to them and I got one of them to quote me on. Okay. So what would it take to build the sales cube, for example So give me a quote on that. And, and then w w if we can get that together and then we’ll move on, we’ve got like eight or nine modules and they came back and it was going to be 50 to $70,000 to build that cube and take three months and I need to, and they said, we need to have your systems analyst. So I multiplied that all out and suddenly it looked like that’s gonna be a million dollars in three years, which was mean, I’d just be finishing that part right now. And, and there’d be all kinds of risks and all that kind of things. So what ended up happening is I maybe can put up the first slide though.
Bill Sandblom: This is a bit of the vision that I came back up, and this slide has not changed since 2018. So what we ended up doing was you’ll see the bubble on the right-hand side with JD Edwards and all the different modules. It’s like good luck or whatever, just being up there and aware, found a company called preferred strategies. And they had all the, they had all the data cubes built for JD Edwards. And basically we paid them a fraction of what it was going to cost to build the data warehouse. And in eight weeks we had our data warehouse built. So we saved money and we had it done. And what that did was it allowed the not only did that they were, and that was part of going into it was can we add a rest of our data to the system And the answer was yest and it turned out that they said we never really had a customer like you, that yes, you wanted all your JD Edwards data in there, but at the same time, you were right away working to get all your other data in there.
Bill Sandblom: So you’ll see in the bottom, right There’s a current ORC, that’s all our rail data. We use them, we brought our CRM data and, oh, as sweet as our operational excellence for our plants to get all that data and, and, some fuels data. And then the one that turned out to be the big one is our IOT data, which, the term I always laugh when people say we use an IOT, well, manufacturing plants have been using IOT for well, when I first started back at the mining company. So 40 years ago, 35 years ago, we have PLCs, which basically took readings off instruments, which is IOT. So for us, yes, we have IOT for this company has had it for a long time too. And that was the big ones. We were able to bring our IOT data into the system.
Dave Mariani: So, bill, I noticed that you talked a lot about cubes and using cubes. So, w Y cubes, a lot of people, I mean, I love cubes. That’s, that’s started a company that, that basically allows you to do cubes on cloud data warehouses, but so why did you choose that interface an investor
Bill Sandblom: Well, so the first part is JD Edwards has, I don’t know what they say, 5,000 tables, right And so to, to give the users, if we were to build and we want to use power, BI is what we used on top of it. It’s kind of a market leader that we wanted to go with. And so to give them to put power BI right on top of JD Edwards users, aren’t going to get anywhere. First of all, the data fields aren’t named properly, it’s difficult. There’s just too many tables where this company basically taught to take, let’s say the AR module. And instead of having 200, 300 tables, they brought it back to seven or eight tables, reading the data, organize it into folders and bang. There’s all the data available for the users. And then on top of it gave some sample screens.
Bill Sandblom: So it was actually, AR was our, was our first thing that we actually, if you wanna move to the next slide was the first thing that we did. And this was like a F this was like three months after we, like assigned you to move to the next slide, giving me that. I mean, we started to, we, I mean, we put this together was one of a bunch of screens for AR, but they got to where it used to. They basically had all their summary data like this, that they could see stuff, but they also had customer statements, and customer activity on how will they were with the, a good payer and on a good payer. the HR department used to say, when we’d have a, let’s say a dispute with a customer and they’re, they’re on credit hole, they’d have to go back into JD Edwards and spend three, four hours. And then they meet with them. Now they just pop in the number up, comes the stuff that we use Snagit, they send them a screen print, and the trucks probably they’re sending the money in the trucks moving in an hour versus sitting there for days. So it just became, that was part of what it did for us. It made it very easy where the users can do their own stuff, but we could also build stuff for them at the same time.
Dave Mariani: I love that bill. Basically, you’re saying that, by, by using cubes, using a semantic layer in those caves, you allowed your users to do more self-service without them having to hunt around through thousands of tables and putting that together on their own. So, that’s a, that’s a great tip. That’s a great Testament to the power of, of a semantic layer. So, so a little bit about, I love that, and that’s a great architecture, and it sounds like you’ve, you didn’t try to boil the ocean that you sort of picked each of these areas off one by one. or did you, was it, what was it, what was your approach to sort of delivering what you’ve, what you just showed us
Bill Sandblom: Well, you know what, and I know I watched, you sent me a podcast from another gentleman and he said, don’t boil the ocean. I think we did boil the ocean. And I think we were successful. And the reason we were successful because we had the company that came in for us and gave us the whole ocean in one big package for us. So we didn’t, if obviously, yes, if let’s say we were going back with one of the big consulting companies, and I said, let’s build all nine cubes for all nine modules all at once that would in boiling the ocean. And we would have filled miserable, especially cause they say, well, you need two, two analysts per module. I don’t have that many people. So it wouldn’t have happened. We would’ve just, we would’ve, we would’ve died. So, and for us, really, within three months of signing the contract to get this, where they put that stuff in, we had all our data, we have that, we have the IOT data, we had our rail data, all that data was there.
Bill Sandblom: And then now, obviously it wasn’t that everybody’s using. Cause now there’s like, we have to spend time with each department say, okay, what are your wants And then try to educate them to say, this is what we can do for you now. And then there’ll be interaction back and forth. But within six weeks, we’d have like for the AR we had something built out that they were using. And I remember, I think it’s the only time that the power BI system went down on us for a couple of hours. And both AR people came into my office, what’s wrong with power BI within five minutes. It’s like, I didn’t know what was that And it’s something within weeks, they became reliant on it because it, and I don’t think, I mean, when I talk about the savings later in the IOT area, I don’t even know what that’s worth, but for them it’s worth a bunch. And I think this slide that I’ve got up here right now, it’s really th this is, this is, this is really what it’s about is that we used to spend the red bar preparing and then hopefully some analysis and hopefully some action we’ve reversed that down. The, the, the action is, is, is what we’re spending most of our time on. And then the preparation now is, is, is, is minutes versus analysis and then actually making changes. And that’s, that’s a huge part of what’s there.
Dave Mariani: Yeah. And, and you know what, it’s, it’s the same that those bars look the same for most companies who are trying to do AI, the data scientists is spending most of their time doing data preparation. And then so, so Gardner also supports these numbers. 80% of the time is in prep. and the rest of it, in, in, in, you know, in building models and analysis. So, you know, shifting gears a little bit to, to AI and ML. so, you are a manufacturing and so you have lots of equipment that are sensitive and you’ve got sensors on those equipment and you’ve got, they generate a punch of data. So, so how did you, you know, how is IGI using that data and talk a little bit about what you did, to, to, to squeeze, squeeze real money out of, using analytics and AI, to measure the performance of those, of those, of that equipment.
Bill Sandblom: Yeah, absolutely. So, what we did was we, we have a system called PI, which, and I think it’s kind of a standard for the oil industry, the mining industry, pulp and paper, just about anybody who’s does a lot of manufacturing, especially with liquids. And maybe outside of that, I think they have tens of thousands of customers. And what they do is they basically, you have your PLCs talking to the sensors, on your equipment and the PI system is, it basically gathers, it’s a historian that gathers and saves the data, and it can save data up to like one second, if you want even milliseconds. I think depending if you’re, if you’re doing something where you really need to see fast changes. So we bought some software that allows us to integrate that into the, the data warehouse, the SQL database for hoax, and, and there was some work to do, and we’re still working on because we’re not done all our plants.
Bill Sandblom: So we started our first plant and data point by data point. What we basically did was take one minute averages and, and create, hourly data. So we would 60 samples an hour, and we put all that data for the PO for our plant, south of Buffalo. And we brought that data across and there was probably seven, 800 different variables that we brought across. And, and it was really interesting because when we were doing this project, I was, we were sharing a conference room where there was three of us working. We spent a fair number of weeks down there, and they were having a process improvement meeting. And one of the engineers when the meeting kind of started here over here in a little bit, cause you’re kind of interested on like, how’s this going to help them And, and, and the, and somebody asked like, how long did that take to get the data for this meeting
Bill Sandblom: He said four hours. So, okay. So we spent four hours going through the PI system to pull it together because it’s just, wasn’t that easy to pull the data out. And the meeting went on for about 15 minutes and then they realized, oh, we need this data now. Okay. Meeting adjourned, former hours work, and we’ll be back well now, if he gets that data in, in minutes and stuff like that. So this is what it comes down to. And, and so then what we’ve done then is then we’ve been able to, the first step was, able to look at 5, 6, 10 different points to kind of see how do we, and the big focus was, how do we improve yields Cause every, I think anybody who’s man has liquids. There’s a certain percentage of waste knowing that I don’t think there’s too many process where a hundred percent of the product comes out.
Bill 4: You’re either getting a and B streams are great, but you always have that C stream that has waste. Then you have to figure out what to do with that waste or you’re selling it, or what are you, what are you doing with it So, so they were doing a lot of work that way to try and, improve the waste and that type of thing. And they were, and we’re making some progress, but the big time was, and it was just kind of a, almost say, it’s a thank you. COVID, there’s not too many good things about Coleman, but it was a good thing. There is that I was attending the west coast power BI user. And, and there was a presentation was laid at the end of the day because of the time zone. And it was by a company called Linux. And I have that same slide here.
Bill Sandblom: This is what caught my eye was, was they came up and this was the slide. And it’s like, their product sits right in between power BI and SQL server. And it’s like, wow, like, that’s not going to draw on that side any better. So they were offering a POC at some kind of discount if you’re at the conference. So like got a hold of them. And, it was interesting because we, I started working with another AI company and a big one from one of the big companies. And they were, they were working hard at trying to make we’re playing with our data and not getting anywhere. And we spent about three hours with this company and they came out and said to us, it looks like on your data, that if you raise this temperature as high as you think you can, without causing trouble, you’ll get better results insurance. Sure. As heck it did. And since then, so we’ve built with them. Then we built other models and those types of things. And, and that’s where the big savings came. That’s our first big savings is that on this one unit where the, where the product comes into this plant, we reduce the waste by about 3%. And by doing that yet, somewhere between 10 and $15 million of not revenue, extra profit generated for the company, which is crazy,
Dave Mariani: This just crazy, this shows you the power of, I mean, 3%. And you said, right, bill, you had seven to 800 different variables. And so being able to find which variable actually matters is, is like, it’s like a hunting for a needle in a haystack. Right. So, so, so I think that’s where that’s where Linux came in, right. As, as, as an AI came in as such to identify which variables were actually gonna move the needle for you.
Bill Sandblom: Yeah. And I think that’s that’s, that, that that’s the, where the big part of it came. I mean, we did for this particular unit, we, I call it the knobs is that for this unit, there was, there’s only 13 or 14 different variables or settings that the control the person running the place can control. So it turned out those were the 13 we put into the AI and then it would, it would tell us which way to which, what, what were the things that were most important and what ranges we might want to try and put those in. And I mean, this is a, this is an interesting slide that I got somewhere and it’s it. Basically, you start out with raw data, you clean your data, and these are the steps that companies need to go through. and then companies get to where there’s standard reports and then you’re getting fancy reports and then you try it.
Bill Sandblom: And then you’re trying to use your, your BI and your BI is trying to predict, well, why did it happen And then you cross the chasm into the next level. That’s where your machine learning comes in. And then you’re able to say, why did it happen And then eventually what will happen And we’re somewhere kind of right in these, these green dots here, we’re not here to where the system’s running it and making the changes to the plant. Maybe someday we’ll be there, but they may not let me near the plant, but start to suggesting that. But, but at the end of the day, we’re now kind of saying, we know if we do this, it’s going to do that. And what the AI did for us was, that it’s still doing for us. And we’re using it on two of our towers, is it you’ll create very quickly with the most important variables decision trees.
Bill Sandblom: And we’ve taken those decision trees and we’ve put them into Vizio. And we’ve now the operators and the supervisors are using the decision trees to try to run the plant, because it basically says, you know, what, if, if you do this, your waste is going to be on average, based on your historical data, it’s going to be 3%. But if you do this, it’s going to be 5%. And now if you do that, the good one, and now you do this and that we can get down to 3% and we’re doing that. And we’re starting, there’s a big change process. There’s, there’s no doubt to get the operators because you’re asking them to do things different than what they used to do, but what we’re finding the results. And, we’re now kind of working on our second vessel and, they’re starting to use the flow charts and we’re getting results on there.
Bill Sandblom: And so for us, we’ve only done this in one plant, mainly on two vessels, we’ve got six or seven more vessels. And so we’re working on as quickly as we can to get the data out of these other plants. They have PI, but it just takes time, you know, work with the engineers to make sure you’re getting the data properly. But that’s where we’re trying to go with this because we’re not sure where the, where the end state on this is, but it’s certainly, it was a good start. And, and I, I just believe any, anybody that has a manufacturing that produces liquids, the same opportunities here, which you’ve got to get through these steps. And that’s, that’s a bit of a challenge because I was on a conference call with the people that make pie. And they said, out of all our customers, we don’t see anybody doing what you’re doing. They said, they’re all stuck here. They’re cleaning their data to kind of get to standard reports. They haven’t got here. They’re all dreaming about getting here, but they haven’t gotten here. And it’s like, gosh, you got to understand the value because this is what you probably get more value out of spending whatever dollars this costs than building a new tower. That’s going to cost you 30 million or 20 million or that type of thing. It’s here, it’s in your data to be able to do this stuff.
Dave Mariani: And, you know, in your case, it’s like, you, you, you mentioned right, these, these improvements are hitting the bottom line directly. So it’s not about producing more revenue. It’s actually producing more profit, which is even, which is even a bigger kick. so how, how is it, so what kind of results have you seen from at least from these early investments You’ve got more to go, obviously. So you’ve got more, more money to, to, to, to put back to the bottom line, but what, what kind of results have you seen so far
Bill Sandblom: In like the next stage Is that what you mean Like where we’re heading next,
Dave Mariani: in terms of district just here at your investments in AI, you know,
Bill Sandblom: Yes. Right. So the ROI is endless, right I mean, the software compared to, when you talk to savings like that, it, it, it doesn’t hardly count. I mean, it’s, it’s like, it’s like spend $5 and get a thousand back it’s that it’s, that kind of ratio is, is what it’s been. So it’s, it’s, it’s, it’s kind of a, and that’s not even counting, like I said, the stuff I showed you in AR and I mean, we’ve got sales and pricing modules that are helping everybody. I mean, people are requesting power BI stuff all over the place for everything now. I mean, we, we really are making that digital, becoming a digital company it’s it’s happening and it’s not. And I still feel like we’re only 10% of the way. I mean, that’s the amazing part is that there’s still so much to do when it’s, I mean, here’s a last slide that I have here is a, it’s a really interesting one.
Bill Sandblom: So this was analyzing some of the data and this chart small, you don’t need to see it, but, but it had this bump here. And so I’m, and I’m asking him in a meeting, it’s like, why is that like that Well, he discovered, cause we’re wondering why there was so much of this orange Lionel for the whole year. This was 2018. And so the meeting was over and the part that’s really interesting is that he figured it out. And what he said is we’re only 20 minutes after the call and we have a complete explanation as to why it was more of that product. That’s the huge benefit of having data in front of us. And that was three days of four days ago. He sent that to me. I said, I got to capture that for this presentation and that’s yeah, that’s, that’s, what’s you can stop sharing now.
Bill Sandblom: I don’t have any more screens to look at, but, that, that, that just kind of sums it up. And I mean, he’s what, he’s what, he’s one of my, he’s one of my heroes in the company. Cause he’s one of the guys I’ve worked with a ton and I mean, you’ve got to have some good people out there too, because I can have great AI and great BI systems set up. But if we don’t have plant managers that want to make this happen and engineers that are excited about this and all that, it’s not going to happen. So you’re, I mean, you’re part of a team, but it does show you though what without it, if you go back to where the same guy as the guy that was spending four hours to get data for a meeting, but now he’s solving a problem in 20 minutes and say it isn’t this great. And I’m saying, yes, sure. It’s so, I mean, that’s, that’s where we’ve gotten to and that’s, that’s a change and that’s, that’s, that’s how we’re going to keep increasing the revenue. I don’t, I don’t know with where we’re at, whether that’s half of what we’re going to do, but I think we’re going to at least, I mean, we’ve identified, we’ve identified two more projects that can double that in the next year. And that’s, we’re not even into two or three of the other plants yet that are excited about it. Yeah.
Dave Mariani: Yeah. This was kind of my next question. My next question for you bill was like, you know, you’re, you know, you’re sounds like a small, tight team and lots of demands, of course, from the business. There’s a lot of the areas you can invest. So how do you, how do you determine where you’re going to make that make those investments and, for the team
Bill Sandblom: Well, it’s interesting. I mean, actually after this podcast, I’m meeting with the CEO just to kind of go over the results. I’m kind of detailed out where the results are and what they are. And then tomorrow I have a, we call it an MIS steering team where we kind of review things three, four times a year. I have meetings and, and I’m going to show them the detailed results of what we’ve done last year. And it’s going to be, then we’re going to have to take a hard look at the projects that we want to, the it department. My MAs department to work on this year is that there’s projects there that are, they’re all good projects, but they don’t have ROI. Like it is in the process thing. So I think it’s some of our business improvement projects. We got to take a real hard look and say, yeah, that’s a great project.
Bill Sandblom: Then it’s going to save somebody three hours a day or whatever it is. Should we be doing that Or should that thing go on the back burner for a period of time Or do we bring some more consultants And now that if now that we can truly say, yes, we can, we’ve, we’ve done the audit. That is, those are the real savings or increased profit. What do we do to try and get that faster Because basically every month it goes by millions of dollars, just slipping up the door for us. That’s my feeling. And I think I’m getting, I am getting more and more people on board. You’ve just got to keep talking about it all the time. Right You’ve got to keep saying like, it’s, whenever I’ve got somebody on the phone and I said, and they’ll say, oh, things like, what’s tough day here and we’re struggling the company or the company’s doing great. It’s like, let me show you what we’re doing. And I’ll spend half an hour and show them some of the stuff on what we’re doing and to say, wow, we’re doing that. And I said, yes. And we’re going to keep making things better and better here because those are the types of things we’re doing. And it’s, and the opportunities are there. They’re just all over the place. And it’s, I think every company has to figure it out. But again, if you’re a manufacturing company, you figure out how to reduce your waste. That’s pretty simple.
Dave Mariani: Yeah. You know, it’s like you, you’ve done a good job at to when it comes to balancing, where are you going to make your investments You quantify what the return on those investments might be like, you know, whether it’s saving three hours for this person in finance or whether it’s, you know, reducing your waste, you know, by, by a couple of percent and that what that means to the bottom line. So, so a lot of companies haven’t been able to do that, or haven’t really taken that approach. And it’s sort of, I see a lot that it’s the loudest voice that ends up sort of driving priorities versus, what’s best for the business and what’s going to make the most difference. So it sounds like, it sounds like you’ve, you’ve gotten a good handle on that by measuring and predicting, you know, what the value will be in the ROI will be with the investments you’re making.
Bill Sandblom: Yeah. It’s certainly, yes. It’s some things are harder, right Like it’s like, this was easy because, at the end of the day, you could show between two years that we get this much more wax so that the system by running like this, so that’s pretty easy. That’s hard and fast. Now you give the salespeople really good access to pricing data and sales data and customer data. Hopefully that helps us get better margins and more customers. I guess it’s measurable if you spend a lot of time, but I’m not going to do that. I mean, you just, there’s an assumption and same with the accounts receivable people by collecting money better and keeping trucks not sitting there while they’re on a, on a, on a hold there’s value there, but that those are harder to work out. I mean, it’s there, there’s no question it’s there because everybody wants more of it and more of it and more of it. And we keep sending it out and we’re, we’re getting well, they’re down the business systems, but, that’s a little harder to measure. I mean, the AI company Lytics that we’ve worked with, they’ve spent most of their career, using their tool to help companies with marketing. And it’s like, it says, yes, like we see things, but it’s hard to see where this is. They’re very excited working with us on this stuff and, and hoping to get more customers that are doing this type of thing, because this is hard and measurable
Dave Mariani: And very tangible.
Bill Sandblom: Right. And, and it’s, and it’s, you’ve got that process data and it’s kind of as well, it’s not a hundred percent, but it’s, I think it’s 80, 90%. It’s kind of like when we say that, if you turn this knob to make this temperature, this temperature, you’re going to get an extra percent in yield and that’s true, 90% of the time, like there’s a, there’s a cause and effect relationship. It’s not always there. Cause sometimes there’s something else going on in the plant or the, or the product’s not quite the same, but it’s predictable 90% of the time. And that’s enough to make a difference. I mean, if you do something and you expect to get the results 90% of the time and you can get good results, that’s huge. And that’s really what we’re doing. And we’re learning like, we’re certainly just at the beginning of this. I mean, I don’t know where it’s going to take us with all this, but it’s, we’re still at the beginning of all this that’s for sure.
Dave Mariani: Well, I could say that, you know, you think you’re, you know, I mean, you’re, you’re, you’re far ahead of most of the companies we see in the industry where they’re still trying to figure out just, you know, what they should do with AI. Everybody tells them they should be doing something with AI and ML. And so they buy the tools and then look for a project. and, that’s, that’s, that’s, that’s not the approach you took. so, so you’re waiting you’re way ahead of the game there, bill, for sure. so, so you accomplished a lot, sounds like, so what’s the future look like for you, your team, and your company, you know, w w w what, what are you excited about bill in, in the future
Bill Sandblom: Well, it’s an, it’s the next projects. we’ve identified a couple where, based on the data is different ways. It’s like we’ve identified one where longer batch runs could be worth a lot more product. I mean, it’s, it’s, it’s one of those things you don’t see it until you have the data in front of you. It’s like, it’s like, if you’re baking four kinds of cookies, what’s, what’s the best thing to do. Well, book make a whole bunch of chocolate chip before you make oatmeal. And because there’s a change over time, right. And, and when we actually find is on some of these batches, the longer you run them, the yields improve. So it’s like, okay, let’s change over time and better yields laundry run, but that’s that wasn’t obvious until we got the data in front of us. And then, and then we’ve also, we’ve got some real good communication going between the two plants on our main supply chain that wasn’t there.
Bill Sandblom: And, and, and it’s not just because of the data, but because of the data, we got people talking and we’ve identified some issues around, some different things with quality of the product and things are happening there. And we think there’s some, there’s some big wins if we can change the quality of the product coming across. And, so, so it gets excited. And I was talking to another engineer in the company just on Friday. I mean, it’s so new on this stuff, and he wants to get involved with it. And I was saying, well, there’s one plant. they say things are running well. And he says, yeah, well, things are running well. That’s maybe means that it’s there’s opportunity to make it better and run more difficult. And the data will help show you that, you know, why it might be a little bit harder to run this way, but you might get 2% more yield. And again, that’s worth who knows how many millions of dollars. So the excitement is that we’re just started, I guess that’s what it is. And, once we finish off, we’ve got three more plants. We got this, but what I’ve done in this plant, and we’ve got three more plants and where will it take us That’s the excitement of it all. That’s what gets me excited every day is I come into work and figure out how can we move it faster And, but, but we’re working hard and we’re getting there.
Dave Mariani: There’s so much more value you could do. Cause you just, I mean, like you said, you just started, you have more plants and, and, and, which is going to be more value and more, more revenue and more profits to the bottom line. So you’re truly data-driven, for sure bill, but, so just a last question is, you know, what advice would you give to other folks who are sort of listening to this podcast, for how they can, how they can get to where you are today So what advice would you give them
Bill Sandblom: Well, you got to get up that, that, that graphic showed with the bubbles and get past that chasms. So hopefully your data is reasonable, right I mean, hopefully, yeah, hopefully your data is your raw data. Is there, I mean, you’ve got to, I mean, I’ve always had things in place. We’ve always had master data management in our ERP to try and clean the data and keep it as good as we can. And you’ve just got to get your data as reasonable as you can. And then, and then move along and then get the tools. I mean, I think you need a data warehouse. I mean, maybe there’s other ways to do it, but that certainly worked for us is to get your data in there. And if, and if you’re a manufacturing company, get your IOT data in there. So you can start to look at it and get a tool, whether it’s power BI or athletics or whatever, there’s lots of other tools out there.
Bill Sandblom: Look at those tools on top of it and get your people that are close to the process going. And, and, I know it’s not easy to get up, get up, they’ll get through the blue dots to the green ones, but you’ve got to get on that focus and start to pull that data. And again, maybe you don’t have to do the whole ocean. It worked for us because of the company that we found and that worked with JD Edwards, but pick an area. But let’s say your company is going to want you to focus on that business. And there’s certain areas on the business stuff, but get into your process. I think that’s where the, I think that’s where the gold is for companies that are manufacturing companies.
Bill Sandblom: Yeah. I L I love that. So, AtScale, we have a, data and analytics maturity model, and we, we, we cover the spectrum to bill of, you know, you click on analytics, you can start with, with, you know, historical analysis. when it comes, we call it descriptive to diagnostic, which says, okay, descriptive is like, what happened yesterday Diagnostic is why did it happen yesterday But then when you cross that chasm, you start to then predict the future. So you get into predictive analytics, which is what’s going to happen tomorrow. And then finally prescriptive analytics, which is okay, now that we know what’s going to happen tomorrow, what are we gonna do about it And so that really is that spectrum there. and being able to have that continuum continuum and have a roadmap for getting up into the right is a great strategy for success. So, thanks bill for, for, for, for chatting with me today. And, and this is, that was, a really great, chart, and roadmap for, for other people out there who want to look to, to what you’re doing, and make real difference to the business. So, bill, thanks for joining us today.
Bill Sandblom: Thanks for having me. That was great.
Dave Mariani: And, and audience, thanks for joining me and the data-driven podcast and stay and be data-driven. Thanks.
