How Big Data and Analytics Can Transform Manufacturing

sobel art

mic Listen to the podcast:

Sight Machine CEO Jon Sobel explains how a new generation of data analytics is transforming manufacturing.

Manufacturing companies are fast realizing that data and analytics can help tremendously in improving operational efficiencies and business processes, and in transforming business models — and they are investing heavily in it, says Jon Sobel, co-founder and CEO of Sight Machine, an analytics company focused on the manufacturing industry.

In a conversation with Knowledge@Wharton, Sobel discusses the changing dynamics in the industry and explains why the focus of solution providers has shifted from offering local solutions to enterprise-level insights. Following is an edited transcript of the conversation. (Listen to the full podcast using the player above.)

Knowledge@Wharton: Let’s start by talking about how manufacturing is being transformed by digital technology and especially with the coming of age of big data. What’s going on there?

Jon Sobel: Manufacturers are looking very systematically at everything they do, from the methods they use to make things, so technologies like 3-D printing and additive printing, all the way to their business models. These large giant factories that have produced things are being broken up and distributed around the world to become much more flexible. And they’re realizing that as manufacturing becomes more networked and takes on the characteristics of a system, just like in the virtual world, the key to making the system effective and being strategic about it is the data that is generated in the system.

And so, in the same way that a bunch of technology companies spent 15 or 20 years hooking everything up and realizing, “now we have to use big data to make sense of it,” they’re starting to look at all the data that’s generated in production as an opportunity. First, [they want] to improve the efficiency of manufacturing operations — so time honored problems — how do we improve quality? How do we keep our factories running? Next, [they want] to improving business processes and then all the way to business model transformation. And so, they are investing heavily in the capability to use data that’s already there. There’s a huge amount of data and manufacturing that just sits on the floor. And they are starting to think very purposefully about using it.

“What’s needed is a very strong technological capability and an ability to help these manufacturing companies evolve.”

Knowledge@Wharton: How big is the market? And what are some of the dynamics going on between manufacturing companies trying to become more digital and software companies entering the manufacturing space?

Sobel: It’s a fascinating question. By most accounts, if we look at the Internet of Things market, McKinsey believes manufacturing is the largest opportunity of any industrial category. Other firms echo that. The numbers that are thrown around as far as the expected value creation from using data and Internet of Things technologies are in the trillions over the next 10 years. If we get very precise and say, “What is the market for big data software to help manufacturing?” The most recent estimates are that it will go from approximately zero now to over a billion dollars within the next five years. So by whatever scale you use, it’s a big opportunity.

You asked about all the people approaching this market and it’s fascinating. Many companies are trying to build do-it-yourself solutions, which is understandable. It’s an extremely challenging problem, so a lot of them try and then, at some point, look for help. The major industrial companies such as GE and Siemens have invested, literally, billions of dollars in this opportunity. And then, the largest software companies, the cloud companies, the business process software companies, they’re all funding and developing initiatives to provide insight.

One way to look at the market is to see all the people who are touching manufacturing. They all want to be a provider of analysis, of insight. And everybody from the software supplier for the front office to the logistics or supply chain provider wants to be able to help the customer understand better what’s going on. And it is a fascinating collision of industry and business cultures because everybody has a toehold on the opportunity, but what’s needed is a very strong technological capability and an ability to help these manufacturing companies evolve.

Knowledge@Wharton: Since you mentioned the collision, is there anyone who is winning this battle? Or is it still being played out?

Sobel: It’s very early. No clear winners yet. In our market, what we do — analyzing production data — the focus has shifted to scale. There have been dozens of venture-backed startups that offer a sort of magic bullet and say, “I can help you understand this kind of machine or this kind of problem.” The focus [now] has really shifted from local problems to scale solutions. If you think about the modern manufacturing enterprises, companies like GE have 500 factories in their company alone, to say nothing of the thousands of factories in the supply chain. The winners here are going to offer a scale solution because that’s what the enterprise needs. It’s enterprise-level insight.

Knowledge@Wharton: So talking about your company — where do you see the opportunity to play in this space?

Sobel: Our company very early on focused on the scale analysis opportunity. The ingestion and analysis of huge amounts of varied data at once from multiple sources and multiple plants. And what we did was to spend several years of building a sort of AI-enabled data engine for making sense of all of this data. And if what comes in from one side is a bunch of raw data, what should come out from the other side is useful information about production. That’s the opportunity we focused on.

We didn’t start with a complete understanding of where the opportunity was. What happened was exactly what was supposed to happen for a startup. We went and talked to a bunch of factories about what they needed and we were guided by their pain to develop solutions. Early on we were hired by a very large global manufacturer that is investing heavily in new automation. And we realized that in order to serve them we had to provide an enterprise class solution that could take data from many different types of sources in many locations. Once we built that, we realized this is what everybody’s talking about and we were quickly pulled into operations by other enterprise manufacturers. And so, our conception of what’s needed here is developed as the market is starting to develop.

Knowledge@Wharton: Can you give me an example of the kinds of insights that are now available through the analysis that you are able to do that were not possible previously?

“The capital markets’ interest in digitizing traditional industry has exploded in the last year or two.”

Sobel: Previously, if you wanted to understand how the same machine is performing in two different plants — and if we think for a moment about manufacturing — small percentage gains in efficiency can be huge dollars at scale. So, if you had 30 plants with the same process going on and you wanted to know exactly which ones are doing better and why, you’d have to have people with clipboards, go measure things, it would take weeks and weeks and weeks. And understanding why one is better than another is all but impossible.

Today, we can put up a screen that shows the actual performance and the reasons for variation. In some cases, operators might be doing things differently. In other cases, aspects of the process itself are different. Another example is quality. We often find that in manufacturing there is a large percentage of scrapper rework. If you’re making drugs, for example, sometimes a batch is bad. This can cost a million dollars. What happened in the process that explains why that went wrong? Many times we don’t know.

Amazingly, manufacturers have huge amounts of data on production. And then, sitting right next door, there is a big pile of data about quality. But putting it all together so that you can know immediately — this batch was bad because of parameter A — is a very challenging problem for them to solve in any sort of systematic way. So those are some very basic examples. You can get much more sophisticated quickly. Bottleneck analysis — looking at a number of lines and seeing where in each line the process is being held back. It varies from line to line.

The data to answer all of these questions is there. And then, if we go to an even deeper level, a supply chain. If an automaker wants to look into the production of its suppliers, it can now know, “You’re on time. You’re late. I better not depend on you alone. Your quality is good. Your quality is bad.” It’s actually in the interest of the supplier to transparently share the information. It helps the supply chain. So these are all areas where the data to answer these questions exist, but we’ve never been able to use the data to answer the questions before.

Knowledge@Wharton: You talked about how huge the opportunity is. What is Sight Machine’s strategy to take advantage of this opportunity?

Sobel: Our strategy is to partner with the leading manufacturers in a variety of industries, work closely and deeply with them and build — in a very measured and sustained way, proof of success in their operations — and then, move beyond those early flagship customers. There have been several counterintuitive insights along the way.

Standard startup theory says, “Go to one vertical, nail it, take a half step sideways and go to another vertical.” That was our plan. We began working in the automotive industry. To our surprise, a large cohort of early customers approached us from other industries. And what we learned is manufacturing is manufacturing. So if you make shoes, cars, drugs — everyone who makes those different things wants to know the same things. So one counterintuitive lesson has been that working with a variety of industries is helping us develop a very robust approach. We use the exact same piece of technology to support them all. There’s no difference in our software, whether you make cars or drugs.

And social proof is very important in manufacturing. It is an industry that is very practical, appropriately skeptical, isn’t really into your PowerPoint. If you get results for somebody that’s respected in the industry, then others will be more than happy to work with you. So our strategy has been to engage with the leaders in the industry and make them successful. There is an aspect to this which has been fascinating and that has to do with the change dynamics in an industry. Think for a moment about what it means truly to have transparency in a large organization. Some cultures are set up well for that, some — it’s not so safe.

And so, useful data brings a level of transparency to organizations which represents a change. So one of the things we’re trying to do is develop very close relationship with progressive manufacturers and make them successful. And then, move out and beyond them. We’re working in almost 15 countries right now around the world and with leaders in several industries who are investing heavily in these capabilities.

Knowledge@Wharton: Could you talk a little bit about the start of your entrepreneurial journey, how that came about and where you are in your journey right now?

Sobel: I did not self-identify as an entrepreneur for a long time. I began my career as a corporate lawyer and worked at some entrepreneurial companies early on in my career. I went to Yahoo! in the late 1990s when it was a couple of hundred people. I was chief counsel there and was exposed to a tremendous amount of early innovation on the web. I see now that I was drawn to a lot of entrepreneurial activity, I just didn’t think of myself that way. I came to Wharton as a mid-career student, feeling that I wanted to learn more about business, but not expecting to come out the other side as an entrepreneur.

Wharton exposed me to a lot of thinking about entrepreneurship and provided an opportunity for reflection. And when I graduated 10 years ago, I realized that I very much wanted to be part of building a company instead of just fixing companies. I was a good fixer, but I really was hoping for the chance to be part of building one. Because I had worked with and been around a lot of startups, I understood that good ideas and good teams are hard to find, especially a combination of both. So I chose my opportunity carefully. And when a technologist that I admire very much, Nate [Nathan] Oostendorp, who I had worked with, approached me about starting this company, I was very flattered and I jumped at the opportunity.

Knowledge@Wharton: What was the original inspiration for starting this company?

Sobel: The original inspiration is Nate and Nate’s a really interesting person to start a company like this because of the combination of his experiences. And this is something about our company that I think has helped us. Nate grew up in Western Michigan. He started a well-known technology site in the late 1990s called Slashdot.org, which was at the center of a lot of the technology community’s activities on the web. He was part of a group of students at Hope College, all from Western Michigan, who built and ran the site from their dorm. Nate had worked in a tier-one automotive plant in college and he thought, at the time, his career was going to be a controls engineer in manufacturing. [But] he was so successful at the web that he went on to do many things.

Nate and I met in 2009 at a company where he was site architect and we were together working on some very interesting big data problems. A few years later, Nate approached me and told me that he’d been thinking about a next application of big data, and he had identified manufacturing as an industry that might be ready. This was heresy at the time. I spent about six months studying the industry myself, because I had seen the pattern of industry disruption ripple through a number of industries. And it seemed like it might be time, but it might be too early.

“Wonderful visualization tools don’t know what to visualize unless you tell them what’s going on, and that’s what we do.”

But of course, you can never know as an entrepreneur. And sometimes, being a little early means you’ll actually be right on time if you hang in. So, Nate had the inspiration and there was a group of others in our founding team with very eclectic backgrounds. One was a robotics and machine vision integrator for factories, who is a world-class hacker, another is a data scientist and there is a long-time businessperson who grew up in Indiana, his dad ran two factories. So we all came together. And I think the fact that from the beginning, we had people who respected and were curious about manufacturing, but also, deeply value technology, allowed us to put those two in the same house and tried to understand our customer.

Knowledge@Wharton: So it sounds like the leadership team came together as a matter of shared interest.

Sobel: Yes.

Knowledge@Wharton: Another very important challenge for any startup is raising capital. How did that happen for you?

Sobel: Raising capital at the beginning was very difficult. At that time, we’re talking about 2012, 2013, most large venture funds were focused on things like social media and mobile. I had come from the world of consumer Internet, so I felt like, “Well, everyone will see this as the next industry, it’s obvious.” And it was an incredibly lonely time because manufacturing is one of those industries that is on the short list of industries that people in Silicon Valley say, “Don’t touch.” That’s changing now. Our early investors were very foresightful and brave. The capital markets’ interest in digitizing traditional industry has exploded in the last year or two. I remember struggling and wondering, “Are we wrong? If no one’s excited about this, what are we missing?”

It was the true startup journey from a sheet of paper to something beyond that. We spent about two years just going and talking to factories — the founding team wasn’t getting paid — and what we kept hearing from factories, from customers, was, “I need something exactly like this.” And then, when we showed them our early product — our first customer was literally a one-factory company in Detroit, Michigan, who fell in love with what we were doing. And we realized the customer knows something that the capital markets haven’t figured out yet. So if we can hang in long enough, we’re going to get there.

But there is this moment that as an entrepreneur where if you’re doing anything new, by definition you’re going to be lonely. I had always been with these big brand names, so I hadn’t experienced that. And it was very difficult. But now, when I meet entrepreneurs, I try to encourage them to hang in. You could be lonely and wrong, but if you’re right, you’re probably going to be lonely, too.

Knowledge@Wharton: Who’s your competition?

Sobel: Everyone and no one. And here’s what I mean by that — we track at least 200 companies that say they will help manufacturers analyze their data. Without getting too deep into technology, as far as we know, we are the only platform technology company that has systematically taken on the data challenge and manufacturing. The data challenge, in big data terms, is variety. Big data means volume, velocity, variety. As far as we know, we’re the only company that has developed a systematic, unified way to make widely varied raw production useful.

We get awards for AI [artificial intelligence] and things like that, but fundamentally, what we do is we make the data useful. That particular problem, that particular benefit is something that as far as we know, no one else is doing yet. And I think that’s why we’re getting hired by our clients because they’re very smart about the gaps in the market.

They build data lakes, they collect data, they visualize data, but wonderful visualization tools don’t know what to visualize unless you tell them what’s going on, and that’s what we do.

Knowledge@Wharton: What are the biggest risks that you face? What keeps you up at night?

Sobel: Most of the risks, thankfully, are now things that we control. That’s one thing that’s changed. I used to worry about market timing but there is a lot of interest from customers now.

Here’s where the risks are: The first is you have to have compelling scalable technology. There’s no reason that a Fortune 50 manufacturer will hire a startup unless you’re bringing something distinctive and new to the party. I think we’ve covered that risk, but you’ve got to keep your eye on that ball.

Second, these are demanding customers who are entrusting you with very important business functions. You’ve got to execute flawlessly. You cannot drop the ball. And most startups, by nature, are not equipped to support global manufacturers.

The third thing, which is very unusual for a startup, is you have to be able to partner effectively with an enterprise that’s going through change. That sounds like a big abstract idea, here’s what it means — people worry that their jobs are going to go away, different functions have turf battles over who does what, you can get caught in the crossfire if somebody wants to innovate in a company and maybe they don’t want your innovation.

You have to find a way to make as many people successful as possible and handle that journey. We have a seasoned team of leaders in our company who have been through these journeys before and, like it or not, if you’re going to lead in an industry, you’ve got to be able to handle that piece. So the risks are that we don’t keep innovating in technology, we don’t execute flawlessly or we don’t sustain and strengthen our capability to be a good partner to large companies that are going through change. Thankfully, we control those things. Every single one of them is significant. 

Knowledge@Wharton: What do you think have been your biggest successes so far?

Sobel: I have to offer a tip of the cap to our technology leadership. The only reason that we get in the door is because they built something truly distinctive. I think the company’s capability to understand and effectively work with the very different points of view around the table is a strength that has helped us. There are these words that are thrown around in business writing all the time. Empathy. The notion of really trying to understand your customer or cultural skills. EQ. We are very fortunate to have an eclectic group of people who can relate to the factory foremen, a hardcore open source developer, a data scientist, a venture capitalist and a CIO, all at once.

“There’s an aspect to our work that involves the equivalent of hitting a machine with a hammer to get the data out of it and it takes about a day or two.”

That cultural sensitivity — and what we’ve learned is people want to hear what time it is, really, just what’s going on. Be straight up. Tell them what you really believe and try to focus on what’s going to help everybody. It sounds easy. But like most things in business, it’s easier to say than to do. I think our company’s strength in addition to technology has been genuine curiosity and skill in navigating some of these dynamics.

Knowledge@Wharton: What do you think have been some of the biggest mistakes that have been made and what have you learned from them?

Sobel: We’ve made many mistakes. Early on, we were so humbled and honored to work for some of the companies that we worked for that we weren’t assertive enough about bringing problems to their attention. We were always trying to make everybody happy because we just didn’t want to lose the customer. What we learned was they want and need us to assert what the challenges and problems are and help them be successful by being a little tough on them.

Another learning was the operational aspects of what we do. So the kinds of thinkers who come to a company with a technology challenge like ours are very deep system-level thinkers. There’s an aspect to our work that involves the equivalent of hitting a machine with a hammer to get the data out of it and it takes about a day or two.

That’s a very different mindset than building a big, beautiful piece of technology. So we recently created a function called “factory connect,” and it’s a different kind of person. It’s somebody who literally gets on the phone with a customer, doesn’t leave them alone, hassles them and goes after whatever the bottleneck is in getting all this stuff happening. And we had to think very intensely about what kinds of dimensions we need to scale. Those were two mistakes.

I think a third was maybe a little too much modesty at the beginning. We often got lumped in with “this is a SaaS product.” It took us a while to appreciate this was actually a very ambitious piece of technology and, chalk it up to everybody being from the Midwest, we didn’t want to oversell to our clients. It was the clients who helped us appreciate how foundational the technology was. And so, we got better at articulating what’s new here.

Knowledge@Wharton: Looking forward over the next four or five years, where would you like to be?

Sobel: I think we can build a company of significance.

This is a real opportunity to lead an industry and where we would like to be is pushing the frontiers of what’s possible with this technology and making a number of leading manufacturers materially more successful in what they do. All of us have been at other companies, we’ve seen what it’s like to participate in something that’s growing and that’s great. And we came together to build a real company. So four or five years from now, we’d like nothing better than to be many times the size of what we are today and having outsize impact in the industry.

Citing Knowledge@Wharton

Close


For Personal use:

Please use the following citations to quote for personal use:

MLA

"How Big Data and Analytics Can Transform Manufacturing." Knowledge@Wharton. The Wharton School, University of Pennsylvania, 21 September, 2017. Web. 18 October, 2017 <http://knowledge.wharton.upenn.edu/article/big-data-analytics-can-transform-manufacturing/>

APA

How Big Data and Analytics Can Transform Manufacturing. Knowledge@Wharton (2017, September 21). Retrieved from http://knowledge.wharton.upenn.edu/article/big-data-analytics-can-transform-manufacturing/

Chicago

"How Big Data and Analytics Can Transform Manufacturing" Knowledge@Wharton, September 21, 2017,
accessed October 18, 2017. http://knowledge.wharton.upenn.edu/article/big-data-analytics-can-transform-manufacturing/


For Educational/Business use:

Please contact us for repurposing articles, podcasts, or videos using our content licensing contact form.