In this episode of the Troubleshooting Innovation podcast, Dave Van Laar shares what bakers need to know about identifying, collecting and analyzing trending information (for the product and the equipment). Hosted by Joanie Spencer, Commercial Baking editor-in-chief.
Sponsored by Reading Bakery Systems.
Joanie Spencer: We started this whole podcast talking about how to navigate those pain points that come with innovation and avoid the heartache that can sometimes accompany innovation. Today, I want to talk about how data collection and analysis can help bakers be better innovators.
Dave Van Laar: Sounds like a great subject, Joanie. Let’s see where it goes.
Spencer: I was thinking we could start with a history lesson and tap into your longevity in the industry. In your experience, when would you say you first saw data become available for bakery operations? And what did the adoption look like from there?
Van Laar: As you said, with my years of experience, my first recollection is when we took a rock and chiseled the instructions on the stone. Then we moved up to… seriously, the first thing I remember is a baker taking off his little baker cap and making notes on his cap. That was data that he needed, and he needed it available to him, so they would take off that hat and write on it. Maybe it was a recording in a manual, something they tracked like an oven log. Information was available for them at a later date. But we’ve always recorded information. We’ve always had data points. But it’s what we do with them that I think is more critical.
Spencer: Absolutely.
Van Laar: Going back to the early years, in 1987-88, I was involved in a project at Pepperidge Farm. The Greenfield plant was a new operation, and there was an automated ingredients handling system that measured precise ingredients into each batch. There were major, minor and micro systems for that. Almost every ingredient was weighed automatically. There should have been no human interface in the whole process. There was a data highway installed in the whole plant; it managed the entire process. Well, operators sat and watched that. But all the information was sent to a central control room. The equipment controls were fed into a central computer system, and everything was tied to the control room. The technicians manned that 24/7 and they monitored everything they could. It was a thing of beauty; however — and there’s always a “however” with new technology — it became overload in a lot of areas. We could not handle all the information. We had “analysis paralysis” at times. We had the data, now what would we do with it? Not all the equipment was feedback-friendly. We could monitor the equipment, but could not always use that information to go back to the machine itself. So if they saw something going out of control, they’d have to pick up the phone and call that operator, tell them to make some adjustments. In addition to that, there were power failures. Just a blip in the system caused resets. Valves malfunctioned. Motors burned out. All those things can happen, so nothing is foolproof. The information was only as useful as our ability to do something meaningful with it. But all-in-all, it was the right direction to go and many meaningful lessons were learned.
As we venture into new territory, there’s often pain associated with it. I remember one of the sessions early on in that project, I sat down with an equipment manufacturer and they were in control of the process, from handling the dough to out of the oven, and we got quite carried away in one session. I thought what we could do on the depositor was to check-weigh each lane across the oven band and each dough ball that came out, and then we could take that information and feed it back to a servo at the depositor and automatically change the weight of each piece as adjustments needed to be made. Of course, that was way out. It was the right thing to do, but we weren’t able to tie everything together.
Spencer: I think it’s still a challenge today, that question of what do you do with the information? I’ve talked to bakers, they’re like, “We have all this data, then we print it out and it’s just numbers on a page and it goes in a drawer.”
Van Laar: Yep, it goes in a drawer in case there’s an audit, in case somebody asks, in case they need it for something else. And I’ve been a longtime proponent of inbound ingredient tracking. We do a lot of work with inbound ingredients. We take measurements, we do statistical analysis, we get information from the [Certificate of Analysis]. We get all this information and we typically put it on a piece of paper or maybe into the computer, and there it is until we get an audit. Then at the audit, we say, “Hey, we’ve got all this!” Well that’s great, and we do need to do that for the audit, but why don’t we do something more with it?
We’ve been tracking lot control for years, inbound materials, outbound shipments. We’re able to track a lot of flour from the mill to the store shelf today. That doesn’t help our process. That’s all just recordkeeping. We do have some proactive things that we use, like the master sanitation schedule, preventive maintenance programs (which I think have come a long way and are a big help in the industry), oven logs are something we’ve always kept, line QA sheets we’ve always kept… But how do we do something meaningful with that data? And that’s exactly what you’re hearing, Joanie, is now what do we do with it.
Spencer: To your point with the audits, data collection is a wonderful tool for FISMA compliance, and like you said, it’s there when it’s needed. But what would you say is the learning curve these days? I’d guess that the curve looks a bit different if you’re an old school baker vs. a younger generation baker, between collecting the data and knowing what to do with it.
Van Laar: The old-timers had that information in their head. They would look at things when they needed it, for instance, if they were having a problem that they thought was a flour problem, they’d go back and look at incoming sheets and check the protein, other statistics on the inbound shipment. They knew where they needed to go to get that information, but they had the experience and the knowledge to know that they needed to go look. Today, we don’t have that with people that are coming into the industry. It’s a learning curve of two things, I believe. It’s a learning curve of the industry to make the information available and usable, but it’s also a learning curve with every new employee. How do they walk into the system and be able to use the information? That’s what we need to understand for this upcoming generation. How do we make the information usable for them?
Spencer: That sparked a thought for me. There’s the veteran bakers who understand the process; they can probably analyze that data really well. Then there are the younger bakers who understand the Internet of things and understand the tools that are available. Perhaps the new ways of data collection and analysis could be the bridge to bring these two groups together.
Van Laar: There’s no question about that, Joanie. We have to somehow build that bakery knowledge into these new systems. I go on back to the Pepperidge Farm days in the early ’80s, I was in a group that was trying to do artificial intelligence by capturing all the information that these master bakers knew. I’ll never forget sitting in a room with these three people while they argued about the best way to fix a problem. These people could all go into a plant and fix the problem, but they’d do it in different ways. They would develop products in different ways. The results would be nearly the same, but their procedures were different.
I mentioned in one of our recent podcast episodes that the operators had their process running in the yellow all the time, perfectly in control. They were measuring product parameters in real time, as real time as they could. Color, moisture, the weight of the product, both at the depositor — the wet weight and the final packaging weight — and the finished product size. This was real. This was not just something for show, because a lot of times, we do things for show. We say, “Look how we’ve got our process in control!” Well, with this lady, I wanted to ask a bit more. I wasn’t certain they were that good at it. She showed me again, real time, the screen of the checkweigher. She said, “I know because here’s the average of the checkweigher, here’s how many lightweights, how many overweights I’m running.” I was really impressed with that. It was not smoke and mirrors; it was meaningful data being used in real time. In talking to her further, when a different crew came on — and I witnessed this — they got more conservative in their adjustments. They were afraid of getting into the yellow so they were well into the green all the time. It was well in spec, but they were giving away a bit more product than they should have, maybe running on a higher side of the color. So it depends on the person taking that technology and doing something with it.
Spencer: That makes sense. I was going to ask about how data can identify how the equipment is performing and how that helps with the product, but that story is such a good anecdotal explanation for how data can directly impact the product.
But let’s talk about the equipment for a minute, because I know you have really good relationships, not only with other bakers but also the suppliers out there. In your experience, and your conversations with the equipment suppliers, how have you seen data help bakers identify how the equipment is performing and how has that led to efficiencies and better product?
Van Laar: I think we need to make that information more visual. We need to make it look like a video game, if we can, and that may sound silly but that’s what the people coming up understand. And not the video game that I learned on, like Pong, we’re talking about video games that are amazingly complex. They require critical thinking skills to compete successfully.
One of my grandsons is regularly playing games with people around the world. There are lots of things going on that I don’t understand in that game, and I probably never will. But can we tie the characteristics of inbound ingredients, for instance, with the final product? How do we make a video game of how to bake properly instead of blowing up the world?
Once the experience of the bakers in the past… How can that be put into those games? The other people learned by experience. We talked about this earlier, of how can we visualize what’s happening in that big, hot box that the product goes into wet and comes out dry? Can we put the perfect baking curve on this screen and have them match it somehow? But also give them some tools to understand what there is, and again, that’s the human side of the learning curve. Maybe they won’t be a master baker but they’ll understand, when they make an adjustment here, that’s how it affects the final product there. I think there’s something to that. In other words, the equipment manufacturers can help us take the mystery of baking — both the art and the science — out of the process and give our operators tools to better manage the process.
Spencer: I think you’re really onto something. In a previous episode, we talked about how veteran bakers had that ability to use their senses, whether it was feeling the dough and sensing something is out of spec or knowing there’s something going on in the machine just because they can hear something is a bit off. Maybe the data that’s available to bakers now, and the propensity that younger bakers have for that gaming mindset, can help them bridge that gap with their ability to sense when something is out of spec or something is going on with the machine, and take advantage of that skillset so they are going further than just turning on the machine and watching it go.
Van Laar: Exactly, Joanie. I think that was our intent back in the ’80s, to somehow capture that information, and the technology was just not there. There was no way to put that down on a screen and have a simple decision “tree” come up. When telemarketers call us — you know, I’ve renewed my car warranty about 16 times last year — they have a tree they go through. When you answer a question one way, they give you one reply. When you answer another way, they give you another reply. That just didn’t work for what we were trying to do, but I believe the major players have certainly made great strides in both collecting and using meaningful data.
One thing I want to caution us all about is not to make it too complicated. We’re always going to need human interface, even that plant that early on in the ’80s, we built so people didn’t have to interface with it. They had to. We need to give our people meaningful data, but just as importantly, we need to educate them on how to use it. As we were talking about, I think we can take some of that ability to analyze the data away from them and put it into a system. But if you go back to the science vs. art discussion, which will always happen, we live in an industry that has inputs with variable characteristics. I go back to Dr. Irfan Hashmi of GrainCorp, “Two loads of flour are never exactly the same. A perfectly standardized flour is impossible from mill to mill, less even from crop year to crop year. The baking characteristics of flour are not definitely indicated by chemical tests.” So we’re not going to be able to get discrete information from testing our ingredients coming in, and they’ll always be variable. However, we can use the data we have to educate our employees and teach them what to do with that information. I think we can do some analysis with it for them, then they can turn around and adjust the system accordingly.
Spencer: When you see the information of how a product is running on the line, have you ever seen any times where it’s like, “We see where we could make those product better.” Like a better crumb, a better taste…
Van Laar: Absolutely. That’s where we have done that out of need more than out of technology in the past. The better bakers will always go back and analyze their inputs to see what may have changed. I talked earlier about that great wall between packaging and processing. Far too many times, I’ve seen a product that’s out of spec go to the packaging area and they’ll adjust the equipment to try to get it in the package, but they continue to run more and more out of spec. That feedback is critical. We need the information highway to do that. That’s what bakers are getting better at: the feedback from one piece of the process to another. That’s the example with the “running in the yellow” people; there were four or five people that were all in touch with what the others were doing.
We talked earlier about the mixer that went on break and kept cheating on his lay time. Everybody else in the process was affected. They all knew that, but management thought they had taken away all the excuses, but they had not. Our people are very intuitive in their processes; they know what they’re doing. I think more and more, we need to bring our people into the process of figuring out what to do with the data. They’re the ones that do it everyday and deal with the information the best they can. They’d like to do it right every time. They would prefer not to have to go back and redo anything. So I think we need to take the input from those people even more seriously than we have in the past.
Spencer: I agree. When we take our view of the process and look at specific equipment that’s lent itself really well to providing data… What types of equipment innovation are bakers overlooking? What’s the low-hanging fruit? Have you seen any tools out there for data collection and analysis that bakers maybe don’t know about yet, or that they’re not taking advantage of?
Van Laar: I think we’ve always been looking for ways to do that, Joanie. Moisture measurement, for example, is something that we’ve tried to do real time at the end of the oven. Some people have been successful with it, others have not. Color measurement at the end of the oven is something we’ve done successfully. But when we do those types of things, they’re not as discrete as what we’d like them to be. There’s a range we kind of calibrate to. If a reading is 2.6, it has to mean something else. It’s a matter of interpreting what we’re getting.
I feel the people are such an important link to that. We tend to overcomplicate. We tend to give information and we think this will really do well, but it turns out to not do much more for us than be reactive to the process. A lot of companies do a great job of helping employees understand what they’re doing afterwards. I’ve toured plants where each department proudly talks about their process. They show graphs and charts to track the process, the metrics are measured, success is measured by the results of those charts. Those are all reactive statements about their process. What I’d like to see is being proactive with that information. How do they realize the process is in control real time? A time when graphs and charts show what we already know? I think a lot of the tools are there, to your question, Joanie. We’re using them already. But how do we apply that information back into the system?
Go back to that big, hot box that the product goes into wet and comes out like a pretty piece of product. How do we understand more of what’s going on in there? We know we can send a device through the oven to understand air movement, moisture, temperature across the band and through the oven. Can we do that real time somehow? Can we take that information and show the interaction between the product and the piece of equipment moving it down the line? I think there are attempts to do that. There are lots of measuring devices out there and they all have their good attributes, but they have to be used correctly. I think that’s the biggest issue that we find.
Spencer: That’s a really good point, to look at it proactively and in real time, not so reactively. It makes me think about the team you talked about running in the yellow, and when they bring a new person onto the team then they have to play it safe and go back into the green. What are ways that teams can get those new bakers coming into the operation up-to-speed quicker so they can get back into the yellow and run those process controls as tightly as possible?
Van Laar: It comes back to education. That’s what we’re always going to have to do. If you look at that process, it’s the exact same information available to both the new person and an experienced person. It’s the exact same information available on the screen when they change operators. The difference in that? The operators and their experience. We know information can be obtained — I mean, all the information they were getting was real time and it was usable information — but we haven’t yet found a way to take that information back and adjust the system. We have not found a way to take weight of every dough ball coming off the rounder and go back to make weight adjustments on the equipment doing that. And is it necessary? Can we go too far with things, make it too complicated? I think good, solid knowledge of the very basic parameters is where we need to start, and it needs to be real time. We never get away from the fact that we have to educate our employees what to do with that. That’s where the technology can come in. Perhaps the video game concept can do that for us. It can list and say, “If this happens, that happens.”
In our education courses, we’ll show a product that is dark in color, burned, and we asked the employees to guess what made that happen. Guesses are all over the place: not enough water, too much flour, too much sugar, too much heat in the last zone. All those things could have caused that, so we ask them to taste the burned product, where you can see the crystallized sugar in it. You know something happened there, probably causing the sugar to crystallize more and it gives you a clue as to the steps to repairing it. A lot of things can look the same, but the cause is definitely variable.
Spencer: Right. The data can help bakers make more educated guesses, right?
Van Laar: Absolutely, and there’s no question about that. I marvel, at times, when bakers have flour crop changes and they don’t tell production. They just let them live through it. It’s an extreme example, but it happens. I don’t understand why, but it still happens. That’s one thing we need to do: Take the data we have and make it usable to our employees.
Spencer: And understand why it’s important.
Van Laar: They need to know that! I’ve never seen people’s eyes light up so much as when they understand what’s behind what they’re doing. It’s amazing. We talk about labor shortages… We don’t give these people enough credit sometimes, in our bakeries, to give them enough information to trust them with it. I think we need to do more of that. We need to give them the information and watch them get the job satisfaction that comes from that.
Spencer: I totally agree. As you were talking about bakers making a guess and thinking about all the potential possibilities as to why, I thought to myself: That’s got to be really satisfying, to have the opportunity to unpack why. I’m sure it leads to job satisfaction on the plant floor, but I think that job satisfaction is something that’s been a challenge lately.
Van Laar: I go back to the training session of explaining elasticity, and the lady going and making a pizza for her family that night and explaining that’s what she does at work. They all kind of stretched the dough, played with it together at home. We could find a way to measure that elasticity. I don’t know that we do anywhere, but we could find a way to measure it. But then if you did, what would you do with it? We’ve got to get past the point of, “This is bad, don’t run it.” We need to get to the point of, “This is not right, what caused it?” With all the variables and the inputs, that’s going to be fairly complex, but we need to have that ability to get back to the gold standard of where we started. That’s too often where we go. We make an adjustment, that becomes the new standard. We make another adjustment, that becomes the new standard. A lot of companies do a great job of making sure we get back to the base formula, but there’s always going to be a tendency to make those adjustments with the dough. And we have to. If you look anywhere in contract manufacturing, we had real tight specs on equipment, but they still understood that bakers would have to make adjustments at times. We had a +/- 2% on water, for instance. We had a +/- 1% on flour. So those are things they knew we’d have to work on to keep the product in spec.
Spencer: What do you think the future looks like when it comes to this ability to collect and analyze data? We think back to the ’80s when the baker took off his hat and jotted notes down on it, and now we look at what we can do now. Where do you think we’re going? Like, when we can fly our cars to work?
Van Laar: Haha it’s so dependent on our workforce, isn’t it? It’s so dependent on the people who want to do something. In the ’80s, we thought we could replace a lot of thinking with technology and that just doesn’t happen. I don’t think it’s ever going to happen. The art of baking is always going to be there, but we need to understand the science more. I think that’s where technology can come into play, to help us understand the science of what’s occurring at the time.
Wouldn’t it be nice if we could look at that burnt cookie, put it into a machine, and it would tell you that you put in 2% too much sugar? That would take the guesswork out for the employee. We go in and if they leave out the soda, we can taste it, check the pH, there are some things we can do in that way. Can we do that with other things? Can we do that with the flour, shortening? I think the way to go is to take more discrete measurements on the inbound variables, then turn that into meaningful information for the employee.
Spencer: You’ve used a word several times in this episode, and that word is “meaningful.” I think that’s the key. The information has to be meaningful, the analysis has to be meaningful, and we have to make the process meaningful for the operator. Right?
Van Laar: Absolutely. That’s where the touch and feel of the dough that the old-timers had, that was intuitive to them, how can we measure that in a way that interprets it to the new employee? That’s going to be critical. Can we put that dough on a little machine that says, “All the parameters are good here,” or “The flour is a little high in protein; you’d better do this to the oven.”
There are two steps. The first is knowing the information, and the second is to know what to do with that information. If we knew, for instance, something changed on the inbound ingredients, the other side of that is, we need to know what to do about it. That’s where I believe the equipment manufacturers can come closer to the bakers in determining that. I’ve been in labs with equipment manufacturers and they are really looking to the bakers to help incorporate that information into their new technology. Everyone is trying to do that. We’re looking for great breakthroughs in that whole thing. If you go to a trade show, you’ll find many people with instrumentations that will tell you they can help your process. A lot of that is cutting-edge, a lot of it is trial, but it’s coming together. There will be a way someday to put that piece of dough in a device and it’ll tell us more about it than we know today.
Spencer: When you think about how far we’ve come with technology in such a short amount of time, it’s really exciting to think what’s on the horizon.
Well, Dave, those are all my questions for this episode. And thinking about what’s on the horizon, I’m really excited for next week because that’s when you have the opportunity to answer listener questions. We’re going to put next week into the hands of our listeners and see what is important to them and meaningful to them and tap into your expertise. I’m really looking forward to that.
Van Laar: I’d love to hear from experienced people also. Lessons they’ve learned in helping the new generation understand the process. If there’s anything they have to add, that would be extremely beneficial so that our listeners could take and put it to use on the line. Submit not only your questions but also solutions. If people have found some of the ways to do these things, I’d love to hear it and we can share it with the listeners.
Spencer: Absolutely. If anyone has a question for Dave, or a comment or suggestion, just email info@avantfoodmedia.com. Dave, I will talk to you next week.