Data Innovation in Manufacturing and Supply Chain Panel

Manufacturers and Supply Chain companies are data rich organizations. In this roundtable discussion we will discuss the techniques and experiences of our panel, pitfalls to avoid, and how they approach data innovation and AI in these industries.

Contents

Our panelists will discuss:

  • How to get started and target use cases.
  • What to expect while innovating
  • Turning Data Innovation into Data Products

Video

Data Innovation in Manufacturing and Supply Chain

Want to be a more effective analytics organization? Need help with being data-driven? Stumped on an advanced analytics project?  Contact us for a free consultation:

Schedule Your Consultation

Transcription

Tony Olson 0:02
Alright, let's go ahead and get going here. Thanks, everybody for joining. For those who don't know me, my name is Tony Olson from Excelion Partners. Thanks for attending this webinar on data innovation and manufacturing and supply chain. From an agenda perspective, pretty simple. Today, we're going to go ahead and kick off with introductions from our webinar sponsors, Excelion, and Dataiku. After that, we'll switch over to a panelist roundtable, which you can see here on the screen right now we have our panelists ready to go. We're really excited about this today. Each comes with a different perspective and the data innovation innovation process and everybody hears from large manufacturers and they'll do some shelves shortly. But why don't you just welcome them and say thank you. As we get this going. You know, topics that we're going to cover in the in the panel discussion is how to get started. And with target use cases inside of innovate Data Innovation, what to expect while innovating and then turning those innovations, that data product. The breath from our panelists is really interesting. Each one comes with a different expertise inside of that lifecycle. So we're excited to dive in. Lastly, this is a panel discussion. So as the discussion proceeds, we'd like to have a q&a at the end, at the end, please put your questions in the zoom, we'll get them answered at the end. And if there's any question for any of the panelists specifically, just please call it out. And we'll make sure that that that question gets addressed that person. So that'll be at the end, we're looking forward to a lively q&a, we reserved about 10 to 15 minutes for that also. So we'll go ahead and get going here, here shortly. So first off Excelion Partners a little bit about us, what we do is we're a professional services and consulting organization been around for about 10 years, we do everything from BI to AI with our clients. A lot, we do a lot of projects inside of the applied data science analytics space, where we are building advanced data solutions into business processes of our digital products. And with our experience in the best practices, people process tools. A common thread that you'll see with us is that we're we're very focused on accelerating your analytics teams, we help accelerate the maturity of data analytics and data science teams through training through tools, working hand in hand with you on those initiatives, we really focused on efficiency there. And then lastly, if you don't want to build an analytics practice, we do have managed data solutions. From a services perspective, we help build data solutions. We also do some data resources slash data staffing efforts from data analysts, data engineers, data scientists, data architects, we also provide data mentoring services and Dataiku training services, we are gold partner of theirs. Next Dataiku. So we were actually introduced Dataiku, back in 2018. It's a unified platform to systemize analytics and AI. And really one of the our big messages when we speak to our clients, about this product and about their analytics organizations is that we look to scale them, and, and then also mature them. And this workbench or this platform allows you to do both. It's really a centralized workbench for everybody, from everybody from IT ops, to data scientists and data engineers, more coders, from financial analysts that are more maybe they're more visual data preparation, people like business systems analysts. And then lastly, the consumers of that data are consumers of the solutions, the enterprise leaders, having that in one cohesive package really prevents those or eliminates those data silos from coming up for work silos and coming up. Another great thing about Dataiku is it has a bunch of different features, all related to the orchestration of your analytics and data science solutions from cataloging and data exploration all the way to ml ops and AI governance, and it can handle all those different items. Finally, it is available in the cloud, or on premise. So it's it's a, it really speaks to any enterprises infrastructure, it's flexible and elastic. And then, obviously, has security and compliance and all those things that have become an expectation for data platforms today. So enough about that. Let's get to the main event here. I'm going to hand it over to Pierre from Dataiku For our discussion today. Pierre, go ahead and take it away.

Pierre Goutorbe 4:41
Thank you very much, Tony. It's really a pleasure to be here today. And we have a really good panel discussion today with Sarah from Schreiber foods. Kyle from Purdue farms and and Mike for most kosher we discussed today during lunch. 45 minutes, 45 minutes together, and we have some time for open question again. So I think I mean, Tony has done a good prediction of the day Chris is as good, I won't have to do this my company and but now I think it's time for for you, the three of you to introduce yourself and maybe potential companies start business business, you're in the type of people who are working at your company and the revenue of those. So let's start with with us our

Sara Stabelfeldt 5:31
answer. Thanks here. Hi, everyone. My name is Sarah stable thoughts. I'm with Schreiber foods. We are a global food company that has been around for 75 years. And we focus in a lot on dairy beverages and plant based food alternatives that we manufacture and supply to customers in retail foodservice and CO manufacturing. For us. Our journey is all about doing good through food, and using innovation to improve our customers lives and consumer lives as well as the lives of our employees. So you'll hear us talk about innovation from product through data through efficiency realms in our manufacturing, and ops. And we'll talk more about that as we go today. I come with 20 years of experience in innovation both at Treiber foods and other CPG namely Kimberly Clark prior to this, so thank you for having me today. Looking forward to it.

Pierre Goutorbe 6:36
Thank you very much, Sara. It's really nice, really nice introduction of Schreiber food. So now let's let's move on to CHI.

Kyle Benning 6:45
Hi, I'm Kyle Benning and I represent Purdue farms. We are in the in the same arena, just focusing on protein manufacturing, as well as agribusiness or agribusiness, which is grains and feed production for the protein that as it's growing. That is international. And then we are focused on United States for our food, which is chicken, beef and pork. We have farms in every time zone. We're approximately about an $8 billion dollar company. And we've got about 22,000 employees, 19,000 of which are in our factory floors. So, you know, pretty small when it comes to our corporate offices. And that's kind of our mentality is you know, keep it lean and fresh is how we move things forward. We're just beginning our data journey here. So I've got 25 years of experience in data in this is my fourth industry. I've worked in public health, manufacturing and finance prior.

Pierre Goutorbe 7:58
Thanks, guys. That's the sort of introduction and now let's move on to tonight is working at Oshkosh.

Mike Schuh 8:08
Hi, everybody. Mike Schuh. I'm here representing Oshkosh Corporation. We're a fortune 500 company, worldwide presence about 15,000 employees last time I checked. And we're headquartered, right in Oshkosh, Wisconsin, probably have heard of us or seen us before, we're not the clothing company just in case that comes up all the time. So what we do is build heavy duty specialty vehicles. And really our motto is, is building serving protecting communities around the world in various ways. So whether that's military vehicles, commercial vehicles, like concrete, or refuse trucks, fire and emergency vehicles, or even JLG boom lifts and scissor lifts and things, usually orange vehicles, you'll see all over the world anywhere you go. We've got a presence and a place there to help people do their jobs better. We were founded over 100 years ago now back in 1917. And we patented four wheel drive so that was kind of our our stake in the game back in the day. And really now it's a lot about digital innovation, adding product intelligence to our portfolio. So I've been here at Oshkosh for about five years, leading kind of our Data Innovation journey. And before that, I came from academic research labs so spent about 1015 years in data science machine learning space. So pleasure to be here.

Pierre Goutorbe 9:39
Thanks a lot, like really great introduction for us. So maybe now, what could be interesting to know like, digit to this discussion is what do you see as the main challenges today for your company into manufacturing and supply chain and and maybe relate to that, and we started to talk a bit about that, what do you see? If you had a magic wand? What would AI data analytics and innovation without you, and what you would like to tackle first. So let's start again, same order. We saw,

Sara Stabelfeldt 10:18
I'll jump in and start. I think for us, when we talk about challenges, it is just very much a time of high uncertainty in our world. And we're shifting from areas where we can use very tried and true approaches and get very predictable outcomes to a space where the future is pretty unclear. And I mean, we're all living in this world where we have massive uncertainty in supply chain, and incoming ingredients, orders that are being placed by our customers, even the labor and staffing within our plants are all things of consideration. And I'm sure resonate with many of you out there. And so when we think about each of those spaces, and what can we do to help decrease the risk that we're seeing and improve our predictability and stability, that's where we're looking at and turn into some of our data internally. And how we can better have tools that augment our decision making capabilities for, for our employees to help diagnose problems as they come up. So those are some of the things we're looking at when we dive like one layer deeper into the challenges and how we solve those through data. Some of the next things that we face is disparate systems and disparate areas of information within the company. How do we bring those together and cleanse the data and format it such that we can solve some of these problems? So that's kind of the macroscopic level? And I'm sure we'll dive in more, but maybe Mike or Kyle, you'd like to chime in to?

Kyle Benning 12:12
I think you hit actually everything that I would have said, Sir, it, it is all those and we're specifically from a manufacturing standpoint, that, as you said, you know, labor being a shortage, we're kind of focusing on looking at labor, and how can we start to use analytics to tell us exactly who's clocked in what skills they have, where we can go, then that helps us to actually plan out where, where we're going to win what product we're going to make, because right now the way our lines are set up, it is still very person centric, we've got to do a lot of things to really trim down the meat and do that, that harvesting process. So if we've got certain people in that can do certain jobs, we do more of that. And so we'll move those things around to help us make that more efficient and get the best product out. We're also then using that to help us with some of the automation so that it does cut down on that dependency on labor. So I think those are really from a manufacturing standpoint, we're focusing, we've obviously got our back office stuff that is focusing on all those other pieces that you talked about, to make sure they're working efficiently. And I'm just exposing data to that whole new level of users.

Mike Schuh 13:31
I can add a couple little things that I think you guys both touched on on the two big sides of it. So we don't do a lot my team specifically doesn't do a lot with the manufacturing line itself, although we do have some work there on efficiencies, as well as machine maintenance and operational logistics. But the big thing, especially in the last couple of years with the global pandemic has really just been supply chain, and proactive risk management. So, like Sarah mentioned, dealing with disparate data sources, having our supply chain folks basically logging into dozen different systems trying to coordinate orders and things. And really what we've done is tried to put all that together in a one stop shop for them. So they can get a handle on things ahead of time rather than always behind and reacting. I think right now we've got somewhere around 1000 suppliers that we interact with on a pretty regular basis. So even kind of going towards the data management and the ground truth to say, you know, these are the numbers. These are the actual orders in progress right now, and help help the team kind of have their own data to do their job better.

Pierre Goutorbe 14:50
Super interesting comments from from all of you like I really liked it. I think that's something that we see among many of our customers or developments in the manufacturing space and like you bought this subject. So maybe if we if we would dive a bit more like, but what could be interesting to know from from the three of you is have you have you started your data transformation or your adoption of phonetics and AI? And, and maybe tell us, when did you start when? How did you start? If you are facing difficulties? Did you try different solution? Or did you choose the best solution? And when you had to face difficulties, or did you overcome them such a failure, anyone want to start? Sara?

Kyle Benning 15:39
vectors, I can start, that's fine. So from a Purdue perspective, my role is really actually the Center of Excellence. So what we've done is we've set up a federated model, we have the center who so our job is to help kind of grow the data literacy practice, and just data management practice here, as well as doing the analytics. So prior to the launch of this program, really the only warehouse solution that we've had is our finance Essbase solution. So just getting people to understand even what a data mart, what a data warehouse, all of those things are. So we're much more on the beginning of our journey to get to AI, we have a couple of use cases that are jumping straight to AI, again, starting to look at our package validations and things on the line. So we're kind of exploring it all. But from a center standpoint, we just sort of starting with the basics of the you know, descriptive and diagnostic analytics, really looking at how we move that forward. And then, again, with this federated model, because the business people are the ones with that great data knowledge, we're going out in each of our functions and finding those people with business knowledge and excel, that are really good at the Excel macros and are really doing a lot of that data work today, bringing them into the fold with this tool set. So that they can help us create the recipes and the transformations. And we're training and upskilling them and more of the data practices to make sure it's done in a performant way. And then we automate it after that. So that's a lot. But basically starting with the very simple stuff of just starting and trying and getting people's awareness of what are the art of the possible is really where we started. And we'll get a few users who are very advanced, and want to try the more advanced things. So we'll partner with them as well. Hopefully, that was helpful, sir, I know that you from an innovation standpoint, probably have a different approach a little bit different approach.

Sara Stabelfeldt 17:37
Sure, I think I mean, some of what you said how resonates definitely, you know, from a data standpoint, it's really like, in my eyes, a continuum or a journey that we're on that is about getting the most out of the data we have and the information we have and then just gradually pushing out into some of the more predictive spaces augmenting decision making in different ways. And we're on a journey in that area, too. So there are small steps that have been taken within our plants to help understand our assets and keep them running efficiently. But then there's bigger areas that we can look into as well in terms of how we're predicting some of our ordering systems and data coming in, if you will. And I think where we've really be gone in that more exploratory space for machine learning and AI applications, we've begun with proof of concepts in a variety of spaces. And what we've learned as we've been on that journey, is that we started with, Hey, this is a, you know, this is what machine learning can do, let's bite off a project that can use that technology. And we just kind of slide it in opportunistically. But what we had to be mindful of is that there is, um, there's a behavior change for people as they adopt these tools. And so how likely they are to use the tool and maximize how actionable it is, really depends on what problem that we're trying to solve with it. And so we took a step back and said, for for each of these applications, let's think about how we use our innovation process throughout. And so everything in our innovation process starts with understanding the problems or the pain points that people face. And so we started grounding our proof of concepts now in what is the business problem, what is the employee problem that we're trying to solve? And then what's the right technology to to couple with that. And so there's a little bit of a back and forth to define in our problems defining the tech to solve it. And then those are going into our proof of concepts now, so that we're able to right size, things for business impact, and then likelihood of actionability against it.

Kyle Benning 20:22
Yeah, I agree. I mean, I think one of the interesting parts that we've had is incubating those new ideas, this toolset kind of helps us do that small incubation of ideas before we get into the bigger tools. Because most of my users, if I start coming in saying, we want to do a data initiative and roll it, they kind of gloss over and go, What is that, but when we actually have, what is their business problem, and oh, we're here to help you solve that problem. They're like, great, more feet. Like I said, we're a lean company. So more feet on the ground to help me solve my problem, everybody's in and glad for that help. And then you just kind of start talking about data as you're solving the problem and creating that awareness, and then they they start to see those pieces. So your point saying focused on what's in it for them? And what what is their problem that we're trying to solve the data, just the data solution comes along for the ride, and then they kind of upgrade their literacy through that.

Mike Schuh 21:24
That's spot on Kyle, I took the words right out of my mouth off that to add to that, I think, for us cashier, we would probably come, our journeys been a little different than than both of you guys kind of in the middle of it, maybe. Really, you know, if we step way back, we started with that classic scenario of, we need to do data science, you know, let's hire a data scientist. Let's do magic. And all our problems are solved. And you hear about that in industry quite a bit. And I think the unique success or advantage that we had was finding some short term ROI and solutions, as well as having a longer term scope of what we're trying to achieve here. And so really, in the short term, it's about identifying those pain points. And getting that business by ends that you guys are talking about, sell coming to them and saying, you know, what's maybe not working as well as it could or what can we help you improve on, so we're not trying to solve a problem that the business is not interested interested in in the first place. By doing that, that also lets us then gain trust and confidence in what we're doing, and start building those relationships. So when they see us solve a, you know, smaller, low hanging fruit type of problem. Now, they say, well, now I've got this next problem or this next problem, and you start building that collaborative relationship to get into the more difficult challenging, and, and transformative applications that you can apply at the same time. So we did a lot of kind of short term ROI, prove, you know why you're here. Because you see that a big risk when people start a data science journey is, a lot of times, it's a lot of spinning your wheels, really trying to find the data in the first place to do the work. So kind of buying that time, then let us also scale out an actual team and data strategy and starting to shift that culture of the organization to understand it's not a sprint, you know, it is a long journey that you're on and it takes a lot of time and a lot of people all working together to really achieve something. So for us, it's a lot about applying data science to our products. And so knowing down the road, wanting to connect our products with telematics adding AI back onto the products for customers, those are, you know, 235 10 year visions down the road, but the way we can start making those right decisions and getting that buy in now with the the business entities required to be able to achieve those in the future. So I think we kind of hold both visions at the same time, and it's been pretty successful here.

Sara Stabelfeldt 24:15
I really liked a lot of what you're saying there, Mike about building the trust, through the short term things to get you into the long term, I think that's really important, is everybody has a way that they're tackling these spaces right now within the company. And so getting them to adopt these new areas, if you can tap if you can find these low hanging fruit or these bite sized wins, to build that momentum and then buy your way into the future is really important.

Kyle Benning 24:50
Yeah, I think the trust pieces team, right, um, you know, we one of our taglines and one of the last iterations was trusted data trusted in Science because if you didn't trust the data and trust the insights, and a lot of times as a new organization, we would stand things up. And we would vet the definitions and vet the formulas with everybody. And we'd come up with this answer and people go, Well, that's totally different than what I ever seen before. So the data must be wrong, right? The new way must be wrong, not the old way. And having, that's one of the reasons that we use more than a federated model. So you've already got the subject matter experts that most of the business goes to that they say those are the knowledge people, now giving them this new way that then brings their analytics up, or their analysis up a level into more of that, you know, augmented AI and things like that. You want to answer changes, or the result changes a little bit, you still have the credibility of the existing analysts, the existing people helping you do it, which then helps the organization adopted more. Otherwise, it's the new group and like, oh, the new group just doesn't know what the heck they're doing, when really what we did was we found a new thing in the data that was important. So it that credibility and trust has kind of been the key to making it successful.

Mike Schuh 26:15
Yeah, I would agree. I think we've done that a lot. Finding those, you can say, key stakeholders or data stewards that exist already. Right. And tying into that. So they're almost vouching for, you know, what is this analytics effort we're doing now? And know that we're using the right data to do that. I think you see, as well, a lot of business as usual, that you're up against. So you're really, it's not an easy job trying to tell upper management or decision makers to change the way they're making decisions. So you can't come at that lightly, or just say, Well, you know, here's my answer. Here's what the what I think the data is telling me, you really need this, to have that buy in and that confidence and trust from the beginning to say, you know, is this the data, we agree on the data in the first place the ground truth, you know, the fundamentals of what we're doing, so that when you do show that solution, they already have all those questions answered, right. They already know how we got here, which is easier said than done. But I think essential to start building upon that as you keep doing more projects.

Pierre Goutorbe 27:31
Thanks, thanks for the for the comments, like it's really, really interesting, we'd like to depart into trust, I think it's really, really important I fully agree with with all of you and bidding on continuing on the same kind of topic. So now, you have started the POC. You have been successful on short term project. So what in casing has been answering a business question earlier? Like, how do you organize your team? Is it something Central? Like a central team that is working on on all those innovative data projects? Or is it more decentralized? And all those project now, are you able to, to see the benefit that for only specific people such as the expert? Or do you see that goes up business people? And you mentioned you were already starting to answer this kind of question, like, people were not necessarily like that data people that are experts or that they know, but you know, software? And are you able to track this result Have you been able to, to, to capture the value of also see initiative.

Mike Schuh 28:50
I can go first on this one, I suppose. Go down the line, I think, a lot similar, our journey to what Kyle described as well. So we kind of sit at the center of excellence, quote unquote, or may or may be an umbrella team in Corporate Engineering, serving the other segments of the organization. So when we first started here, really what we wanted to bring to the organization as a whole was kind of three things. One was big data to do kind of that that new framework of how to do analytics in the modern environment. Another was then all of your like web based self serve tools. So people can go and get their data themselves and interact with it and start answering their own questions. And then really, the third thing we want them to bring was our own expertise to help guide people on that journey so we can upscale the organization as a whole. And I think what we've seen over the time is by kind of sitting as this Center of Excellence outside of it and outside of specific lines of business. We've been able to kind of scale our limited resources, more broadly speaking to the larger org, and then identified those, whether they're kind of citizen, data scientists or analysts, or subject matter experts that are data centric data focus, then empowering them in their own lines of business to start upscaling and adding value to kind of the larger group that we're forming. So really, over the last couple of years, we've went from our own pocket of about a dozen people on our team, to now dedicated teams in each segments, and then dedicated kind of contributors in specific lines of business within all those segments around the org. And it all still kind of feeds back up to us still being able to talk about our tools and techniques and processes and, and having some sort of best practice mindset on the direction we're all trying to go together. So I think that's been really successful. And I'd be remiss to not say Dataiku has helped in the last year or so, by giving us that collaboration in that space to say, you know, here is how we're doing this, here's where everybody can come together, especially with collaboration, but also the compute and storage resources we have around the org. So a lot of people that maybe they don't know how to interact with Hadoop or a big data system. Now, they don't necessarily have to know all that. So it lowers the barrier of entry for them to get involved and play a part in the organization.

Pierre Goutorbe 31:42
Kyle and Sarah?

Kyle Benning 31:44
So I mean, obviously, like I said, I've been kind of foreshadowed this question a little bit, but you know, as the center, one of the things that, you know, as we were talking with the CFO, who's our, who is the program sponsor, and we talked originally about, hey, look before it systems wherever around so you know, we're a company that's been around since 1918. And before we had IT systems in the 70s, and 80s, right? 90s, the business person, when you said, Hey, what's on your general ledger, what's on your monthly chart of accounts, what's all that they knew that data, they knew where it lived, because it lives physically on their desk as a physical chart of accounts as a physical ledger, that they would do that, you know, and she goes on to the story of, yeah, I used to have to take my monthly pad, and then roll it up into this. And this is why this works. And she's talking about the business process and how the data flowed across her paper. When we shifted to IT systems, everybody, kind of the business tended to lose that sense of ownership. And it shifted to it being the owner of the data, and it had to keep it and it knew where it was. And as we've started, one of our goals is bringing that data back into the hands of the business and saying business, you truly own your data. That's why we go with that federated model. And bringing to your you know, two. Point before about tools. Now, previously, it was all IT tools, very tech savvy, you had to be very tech savvy to interact with the data, now you get a set of tools where you can interact directly with the data as a business person. So because that's one of our goals is getting business ownership of data, we've created that more federated model and teaching people around the business, how to interact with their data is a key goal of our program, not just the analytic output. So it's, I think it depends on what your win is, right? All that said, we still at the center had to come up with one solution that kind of we picked a set of data that the whole company uses. Some of our competitive analysis data, showed how it worked through the system showed how we documented give them kind of a visual of what it looks like. And then they were able to interact with that more. And it's data they already knew. So it helps to show that what the future looks like to really get that buy in. Because we're trying to do the whole organization, it's really risky, because you're asking a lot of people to do change. So if that OCM journey, organizational change management is always the hardest thing. And either people get it either the rock, you're able to push the rock all the way to the top of the element starts to roll down or it rolls back over here, right? So you kind of got to watch how you do that. But I'd say that change management and getting to that federated model so we can get business ownership of data being what is our driver so I think picking a key driver to say what you want the organization to do is kind of one of the keys to what organization fits best in what you're doing. Sorry, was long winded there, sir, I'm sure with the from an innovation standpoint, you're a little bit different on that focus.

Sara Stabelfeldt 35:00
Yeah, yeah, no, that's great. We are, we are a Trevor in in evolving space on our organization and structure. So we have begun our innovation journey in a more formal way in the past 18 months. And as part of that, we've been evolving how we look at the structure and these opportunities for advancement. So right now we're in a space where our data team is within our is department. And then we have a cross functional group that is looking at each of the opportunities, where to hunt for problems, how to match these proof of concepts up. And we bring those ideas together on a regular basis to size them, prioritize them, scope them in a way that is actionable. So it's a group of people from the business, from technology, from innovation, and from the various functional groups like manufacturing, that come together to sort through those, I would say, this is still though in a space where we are proof of concept in this and I do expect that as we as we learn will continue to evolve in that area were young in that this area of the journey.

Pierre Goutorbe 36:36
So that's our offer for this for your comments. And I think it's interesting feedback from four of you like making code as well. So maybe maybe as a, as a question to look further towards the future. What's What's your what do you see as your potential next steps by adopting even more like, such tools as innovation technologies? And maybe a question that I mean, we didn't really answer so far as, as taste going to be even bigger, like, you're gonna adopt even more like, I can, you clearly like, make sure everything is governed properly, and everything, you keep track of everything. And you really shows the long term value of all those projects.

Sara Stabelfeldt 37:33
Yeah, so as I think about our future, I think about really embedding the mindsets and the tools within the company and focusing on how we can kind of get a flywheel going of showing some successes, and then using those to launch us into new spaces. As we've looked at some of our most recent proof of concepts, what's exciting about it is, we've scoped out a specific opportunity, and then, you know, delivered on that, but at the end, it's just starting to naturally trigger more questions and thoughts and applications that the team wants to take into a next generation space, which I think is one measure of success for us, because it shows that you're having an impact, and it shows that you're building the overall capabilities of the ark to see where we can take it in the future and build those next, next set of opportunities for us, and then also starting to build the internal champions for these tools, this data and way of working, which I think is a really important success metric. So 100% Yes, each each I'm proof of concept needs to have a success criteria and an ROI sort of metric but I also think much beyond that in terms of setting the right culture and, and flywheel of opportunities for the future.

Mike Schuh 39:17
If I can jump in right off of that, Sarah, I love the flywheel mindset and mentality. We talk about that all the time, actually. So it's, it's great to hear the same thing and anybody else listening? hasn't thought of that. I mean, there's lots of motivation online. I think a few key players in this space talk about it a lot, but it's really that momentum building you know, spinning of that flywheel, all that energy that you're building up, to keep it going without friction, you know, removing as much friction as possible, from your team from the org from the tools, so that you can see that that gain and improvement faster and faster. You know, basically It'll start smoothing out all the little ups and downs that your team might have, whether it's fires, or ad hoc projects, or deadlines and whatnot, and give you that just kind of upward trajectory, momentum of growth and innovation. And so we've really been probably the last year, we've been talking about that a lot, internally for our team, as we've been scaling out our people, processes and technologies, trying to try to not take that step back. How but kind of harness everything we've done and keep going forward. So. So a little tangent on that. But I think back to the question, really, the two key things I'll keep it brief for us is, is operationalization, which has kind of always been the thing in recent years for AI initiatives. For us, it's now talking about different deployment areas, whether it's cloud base, as you base or production level, AI and telematics for vehicles coming off the assembly line. And then the other thing is really just scaling broadly speaking, like we're talking about, so still building out our teams, we we have a data team of about a dozen that were, you know, all the different hats of the end and line and scaling that out, scaling out the products that we touch, as well as the projects then that we're working on for that. So just everything up and growth with that flywheel mentality right now.

Kyle Benning 41:31
I think that that's, I mean, I haven't used the flywheel before, but I'm it's perfect sense, right. Um, and, again, with our federated model, what we've really done is started to segregate to say which teams are responsible for which key deliverables, right, so our tagline is, skills, plus technology, leveraging data equals insights every day in whatever piece of the business you have, right? So stride and striding the business forward. So the Center of Excellence is really focused on the skills, the technology upskill, you know, upskilling our people and making sure the technologies performance. And then the insights in which questions we're asking, we use those federated teams to say, what insights are you after, what are you monitoring, so our manufacturing has a set, our finance team each have a set, and they're really meeting their business where they're at. So we're able to take those next steps. And then when we look at the platform as a whole, we think about it more from a data lifecycle, as opposed to just projects. So harvesting raw data, no matter which team asks for data, we're bringing that data in and keeping it at a raw level. So any other team that might need that data in a future project would have it, then we have our workbench faces where each of those teams are able to incubate all of those ideas with the center coaching them on upgrading their skills, right, and making sure the right technology, so they're incubating their ideas. And then we have our consumer, which is what when they have an idea that's been incubated, and it's like working, it comes back to the center, and then we can tune it and automate it, and make sure that it's performance and running over and over again, into our consumer layer, which is those data assets that facilitate whatever the question or the insight that people were after. So really focusing on those three levels, and how the teams interact to take the data through those levels is kind of how we keep this thing churning forward and grow everybody forward. So specifically with manufacturing, they're kind of looking at, you know, again, we talked about processing and making sure that production is good. Then we have our animal health, which is the raising of the flocks, and they're starting to look at how do they interact with the birds more and actually do you know, give the bird their little RFID chip, and we can track them all the way through the farm and know how much they're eating know exactly what's happening with each one of those birds. So we can do better prediction on when we process them how, how that's looking, and then give feedback so that our farmers are raising happier, healthier birds along the way. And it's not just about processing. So but I think that yes, projects in proof of concepts are important. We've I found that when we get focused on a proof of concept too much sometimes we lose focus on the bigger picture of do we have the data ready for the next proof of concept for the next idea? And if it's only if the data is only form in a way that answers the one question, you kind of lose some of that momentum. So keeping it focused on each one of the steps gives us some of that ability to be ready to scale faster.

Pierre Goutorbe 45:03
Thanks. Thanks a lot.

Mike Schuh 45:06
I was gonna say real quick. I like that last point, Kyle that I don't think we've touched on. I know I had it on my list to mention at one point, but the iterating on proof of concepts, right? So I think a really important thing is sure you can do this proof of concept idea. But now, while you're doing that, be thinking about the data pipeline? And can you do it again? Can you answer the same question next month with new and up to date and fresh data? Rather than Alright, here's your answer, one and done. You know, never again, it's hard to start all over. Rather than thinking about that through the process, maybe it's certain pieces of data, certain features, or certain ways you're building models, that if you do it, you know, right the first time, it's going to help you out in the long run.

Kyle Benning 45:56
Yeah, and we're even a step, we're even a step back from that Mike of just, you know, a lot of people spend a lot of time just finding the data and going where was it and, you know, the first part of every was, you know, the good 40% of the project lifecycle was just getting data so that we could in a, in a place where we can use it. And what we're trying to do is take that first proof of concept and the data, they get structured and ready to use, make sure it's stored in a way that is accessible for all people that should be able to see it at the company that have that visibility, so that they don't have to on their next project, spend time with that set of data. And the hope is that the more proof of concepts, get more data in that lake. And by time we get to, you know, proof of concept seven or eight, they've only got to do 10% of their data, because the other 90% was already ready for them. And then going on to do the modeling and all that other fun stuff that they can spend more time, which is where their intelligence really is right. Or, you know, where we're getting the value add for the company.

Pierre Goutorbe 47:06
Yeah, definitely.

Yeah, fully agree with, with those last comments, it's really, really something important to keep in mind. So I think I think we, we have finished, like the conversation and we will put we can open to q&a. Tony, one thing that I think it was really important in the conversation, and Nic was really, I think, discuss all along the petition is the fact that people is is really important. And I think it's the right culture and putting the people first, including into that. That innovation journey is really, really important. And I think it was actually the discussion, something that was reflected like a couple of times by all of you.

Tony Olson 47:59
Absolutely. Thanks, Pierre. Yeah, one question that came in. So Sarah mentioned momentum being a piece of success criteria. What are some? What are some other techniques or measures for success criteria that helps justify that next data innovation effort? Sarah, you know, I think they called you out in that one. Why don't you go ahead, and maybe if you if you have some other techniques, or criteria, that'd be great to share. And I can move through the rest?

Sara Stabelfeldt 48:23
Sure, sure. So we do have a formal scoring mechanism, if you will, when we bring new ideas forward. And it's really related to Financials around efficiencies that we can gain or revenues that we can gain. There's a feasibility metric around being able to acquire the data we need or that we have it and can use it. There's definitely success metrics around alignment to our overall corporate goals and strategic initiatives so that we're focusing on things that hopefully we'll have more than this, you know, n equals one small impacts, but we'll be on a trajectory towards our overall growth and strategy futures. Those are some of the top of mind. To start with.

Kyle Benning 49:31
Yeah, I think I would just add to that, you know, a lot of times we think about, we struggle that the first going, how do we calculate ROI, what's the return and all of that, but when we but Sarah, you mentioned it earlier, right? What is their question that they're trying to answer? What is the reason that they're asking something? Most of the time executives will tell you I care about this because it'll make the business better this way? And they've already got the story of why they're asking the question. So then you go, Well, if you're making the Better Business better this way, how do you quantify that to even justify, that's where you want to take the business. And they tend to bring those things out. And then it's a matter of tying in what technology and how much the technology might cost to get. Is there an ROI? But focusing on what is their question they're trying to answer and why they're trying to answer it. A lot of times, we found, I found anyway, that it could lead me better to what am I? What value am I going to give to the business? But as soon as I start to ask them, Well, what is data doing? What is it? They're like? Oh, no, it just is there, right? Because it's, it's an enabler to making that decision or that improvement, it's not really the win. So getting our executives to realize, we're just giving a new enablement, a little bit more cost, maybe to that business improvement that they would do anyway. And that's what they were driving. That's kind of how we ended up getting to our, what was the right ROI? Or what was the right measures? I don't know if that works for you certain innovation or not. But

Sara Stabelfeldt 51:09
Kyle, you actually triggered something else. In my mind, as you talked about the why behind? I think it's really important to acknowledge that there are various elements of quantifying these things. And so some of them, it's hard, it's a hard exercise to quantify. And because there are intangible or hard to quantify elements of it, like if you take labor as part of that, you know, there's elements that might not be at the surface about training people, retention of people, and things that can be more than just that headcount itself. So when you work on talking through the why, with executives, and and stakeholders, that can help flesh out some of these other components. And it's important to not just look at that one number at the surface. But what is the impact over all throughout the chain of those involved?

Mike Schuh 52:24
I can, I can throw in my two cents, I think you guys covered it really well. But from a total different perspectives that some people may want to hear on the phone from engineering of the platform and services you're providing. I heard a really good statement years back when we were building our architecture was that, you know, you're being successful when you start getting a lot of complaints from people trying to do more than they are able to do in your system. And I thought that was just kind of reassuring and validating for us at the time. Because you know, if you build it, they count they will come they say right, but are people using it or not? And you don't really know, and you're trying to provide this self service experience. And we felt the inflection point where, you know, a few people were dabbling, but then it became part of their work process, they became the tools they're preferring, well, now, can I do this? And can it why can't it do that, then, you know, and you start getting those those questions and feedback. And really to see that as a success for yourself and your team that you're building. And of course, the do better now, but at least shows you that that feedback and involvement. So I always said, that always sticks with me.

Tony Olson 53:41
That's great. I'd like both of those. But you know, I really like to talk through the why I think that's an important. It's a good way to succinctly answer that question. Right? It's just to make sure that you're talking through all the different aspects of the Y getting to the root of the ask, so that's great. I don't see any Oh, sorry. Go here.

Pierre Goutorbe 54:03
Yeah, do do we got any other question?

Tony Olson 54:06
No, I don't see any other in the chat. So up, here we go. Yeah, so this one is, what is the main user interface with data that users at your company use? For instance, like, like, for instance, a tableau, or, or others.

Kyle Benning 54:30
So I can take a first shot? Again, we've broken down into those three sections, right. So it all depends which users you're talking about. We have, we've developed personas of report writers, and users report writers and data analysts are kind of our three personas that we're trying to enable. And we would say the data analysts are using Dataiku is their main tool set to do Data Prep data wrangling data analysis, modeling, those kinds of things. The data scientists obviously being that next level statistical data analyst, also using Dataiku. Our report writers and our end users are consuming through Power BI, where we're connecting directly to our data sets in the back end that we've, you know, we've created the datasets through Dataiku. They're governed, signed off on fully all the metadata capture into that consumer layer that we're calling it, which is really think about collection of data Mart's each function having their own data mart that's fully governed, we then can connect Power BI to those data assets. And people are using that report writers to create an end users tickets.

Mike Schuh 55:43
Were pretty similar here. I'll say In short, I think we're we're a Python and a sequel shop. So a lot of it was code based originally. But as we're scaling, really pivoting to use data, I Kumar for everybody at any any skill set or tool experience, they can all jump in. And Power BI we use as well here.

Sara Stabelfeldt 56:11
And nothing to add on my end, we use a variety.

Tony Olson 56:15
Oh, it's great. All right. Well, I know that we're coming up here at at at time. So I just want to take some take a moment say thank you to the panelists, thank you for to the attendees for joining. Really appreciate everybody taking an hour out of their day, to come here and share your knowledge and just kind of breadth of experience too. I thought, again, having such a good variety here has been excellent for all different levels of maturity, all different levels of roles. So again, appreciate your time and willingness to share with a large community Sara, Kyle and Mike compare. So thank you. One more logistics note is that for any anybody that was on the call, we will be sending this out as a as a recording. So if there was a tidbit, you want to go back to or listen to again, you're going to have that opportunity also. So again, Sarah, Kyle, Mike here for everybody on the call. Thank you very much. And that's a wrap. Have a great day. Thanks. Thank you very much, everybody.