"IT can't be what holds this project back."
That's what Charlene Begley, CEO of GE Home & Business Solutions told CIO Alan Kocsi.
The project she was talking about has become the poster child of the "Revitalization of U.S. manufacturing." That's how it's hailed by politicians, business leaders, and workers. They view it as proof that America is still a great place to make things. At its core, the project is about manufacturing refrigerators, water heaters, and washers at GE's historic Appliance Park in Louisville, Kentucky -- a plant that was on the verge of shutting down just four years old.
In 2009, General Electric Chairman and CEO Jeff Immelt said that GE was committing to a $1 billion re-investment in U.S. factories and the creation of 1,300 new American jobs by 2014 -- mostly manufacturing jobs returning from China and Mexico. As a result, GE's Appliance Park has sprung back to life, sprucing up its warehouses, retraining workers, and purchasing the latest manufacturing equipment so that it can launch new product lines of appliances from the Louisville plant.
However, the company had an IT problem that threatened to derail its ambitious plans for 21st century manufacturing. The GE Appliances division was running outdated software and systems that were cobbled together over decades. It was a tangled mess of custom apps that didn't talk to each other and the result was that it took way too long to get reliable information about the state of business processes and the overall health of the business.
The systems were simply not nimble enough to support the kind of lean manufacturing operation that GE needed in order to make it economically viable to build appliances in America. IT was a game-stopping obstacle.
How GE Appliances overcame that obstacle -- or, more accurately, is still in the process of overcoming it -- says a lot about where GE is going as a company and how today's enterprise IT departments are having to make radical changes in order to adapt to the relentless pace of a faster, leaner, and more global business environment. Inside GE, this initiative has been dubbed "ERP+" and TechRepublic interviewed executives, IT leaders, and key employees at GE Appliance Park to get the story on how they are pulling it off.
The near-death experience
GE broke ground on Appliance Park in 1951. Just two years later, the site employed nearly 10,000 workers and became a paragon of American industry when GE purchased a UNIVAC computer to handle payroll for all those employees. It was the first business in the world to own a computer -- only governments had owned them up until that point -- and Appliance Park was the first non-governmental site to host one. Appliance Park was a beacon of progress in America's post-World War II manufacturing boom.
By the 1970s, Appliance Park was churning out millions of appliances -- from washers and dryers to refrigerators to air conditioners -- and employment at the manufacturing facility peaked at almost 20,000. With that many employees, the site became a hotbed for labor disputes and was the site of several highly-publicized worker strikes.
But, by the 1980s, GE began shipping some of its manufacturing work overseas to Mexico and China to reduce costs, and Appliance Park started to shrink and wither. From 1994 to 2007, GE avoided hiring new hourly workers at Appliance Park. Employment dropped to about 5,000 workers.
Then, the 2008 global recession nearly wiped out the facility. The U.S. housing market collapse brought new home construction to a halt and since over 95% of the sales of GE Appliances are to the American market, it had an immediate and devastating effect.
In 2007, the appliances business had been the business of the year inside GE, riding the hot U.S. economy and the red-hot housing market to strong sales. But when the bubble burst on U.S. housing, the weakness of being so dependent on a single market led GE executives to consider spinning off or selling the appliance business to one of its appliance competitors that was more diversified globally. However, when GE shopped around its former golden child, it didn't find any buyers. It was too expensive and the long-term prospects of the U.S. market were too uncertain.
Those were dark days in Appliance Park. No one knew which company might end up owning the business and there were serious doubts about whether any buyer would want to keep the aging facility open. Employment at the plant dropped to under 3,000.
But, then came a ray of light. When GE failed to find a buyer, Immault announced in 2009 that not only was the company keeping its appliance business, but it was going to make a dramatic reinvestment in Appliance Park and bring a lot of the product development and manufacturing for appliances back to Louisville.
There were three factors behind the startling about-face:
1.) The federal government stepped in and provided assistance from the Recovery and Reinvestment Act of 2009 and ultimately gave a slew of tax breaks and incentives for GE to build a new breed of energy-efficient appliances using U.S. labor (a.k.a. "green jobs"). The Kentucky state government and the Louisville city government also put together aggressive tax incentive plans.
2.) The local union in Louisville agreed to a wage reduction in which new manufacturing employees would start at $13/hour (roughly $26,000/year), or the same wage that workers at the plant received in 1980. That was down from $22/hour, as the union and the community made the trade-off of more work versus higher paying jobs.
3.) GE decided that it could change its business model for appliances. It had previously outsourced much of its product development and manufacturing to overseas suppliers (many of whom were now entering the market with their own products). GE had relied on the strength of the GE brand and the company's distribution to make money. However, GE made a bet that it could "invest, in-source, and innovate." In other words, by bringing product development and manufacturing back to the U.S. it could decrease the time it would take to bring new products to market and out-innovate its competitors in the appliance market.
The IT problem
Even with the victory of keeping Appliance Park open plus all of the excitement around bringing jobs back to the U.S. and launching new product lines from Louisville, there was a dark cloud that threatened to rain on the parade. The existing IT systems in Appliance Park were not prepared to support the kind of hyper-innovative environment that GE Appliances was staking its future on.
"We didn't really invest in IT for 20 years," said Kevin Uhls, IT Director of GE Appliances.
GE Appliances had 2,673 disconnected workflows, 530 applications, 444 IT platforms, and thousands of different databases. The integration between these systems was almost non-existent. Uhls characterized the IT environment as "very good silos and very little trust."
It was mess. And it couldn't support the high-tech manufacturing juggernaut that GE wanted to create at Appliance Park.
So, GE went hunting for a CIO who could lead an IT transformation.
They called Alan Kocsi.
Kocsi (right) had worked at Appliance Park 15 years earlier and since leaving Louisville he had worked at several different divisions within GE, including Healthcare and Capital, and he spent a good deal of time in Asia with GE Corporate. Since an important segment of the IT department at GE Appliances was located in India, that experience was highly valuable.
When GE Appliances approached Kocsi in late 2010, he was living in Florence, Italy and working as the CIO of GE Oil & Gas. Kocsi earned recognition at Oil & Gas for leading the integration of the company's ERP, PLM, and CRM systems, which saved $38 million and reduced the time it took to create proposals from 15 days down to a single day.
"I got the phone call," Kocsi recalled. "And they said, 'Would you be willing to come back to Appliances to lead this project?'"
Kocsi was intrigued.
He remembered the GE executives telling him, "We're reinvesting in the business. We're going to change the game. You know the [appliance] business."
It was a new challenge, a big challenge. But, Kocsi was quietly confident that he had the experience to help Appliance Park pull off the kind of big transformation that it needed.
"I had a track record at GE for doing some different things," he said. "I was the one who had set up all of our software development centers in India. And I had a history of doing ERP projects. It was good fit."
When Kocsi returned to Louisville, it didn't take long to figure out that he had a huge task in front of him and a lot of urgency to drive change quickly.
"When I got here we had no budget, no plan, no team," said Kocsi. "But, we knew he had these M1 product go-lives staring us in the face and we didn't want to put the new business model on the old systems."
M1 was GE's internal name for the reinvestment program that was bringing product development and manufacturing back to Appliance Park. GE was on the verge of committing a lot of time and resources to the legacy platform that tied together the most important parts of the computing environment. Kocsi knew that it wasn't the answer for what the team needed to build.
"I sat down with the CEO and the CFO and they told me that the thing that they loved about the IT department here is that it did everything that they asked. The thing that they hated was the IT department did everything they were asked. They were looking to us to provide guidance as opposed to just following orders; to take a swing and do something different."
That was January 2011 when Kocsi arrived. He was ready to do something different -- really different.
Kocsi quickly evaluated the tangled mess of different systems that Uhls described. In fact, he recognized a lot of them.
"I knew most of the systems because they really hadn't changed in 15 years," he said.
He decided to pull the plug on the vast majority of them. The problem with the existing systems wasn't just that they were outdated and didn't integrate well with each other. It was also that they were so customized and specialized that only the employees at GE Appliance Park knew how operate them. In some cases, that meant only a handful of people. There were times when the company had to practically beg employees not to leave or retire because they were the only ones that knew how to operate the systems. A standard solution was needed.
Kocsi said, "We cancelled about $6 million of stuff that we were going to do on the legacy environment. We sat down and in 90 days we crunched a plan that said, 'This is the tech stack we're going to go to.'"
The tech stack involved taking hundreds of different apps and platforms and a bunch of different custom ERP solutions and consolidating into one standard ERP from Oracle and a handful of supporting apps. Here's what the primary stack looked like:
- Supply Chain Management (SCM): Oracle and Proficy (a product from GE Intelligent Platforms)
- Order Ship Bill: Oracle
- Financial Resource Management (FRM): Oracle
- Product Lifecycle Management (PLM): Windchill
- Customer Relationship Management (CRM): Salesforce and Siebel (an Oracle product)
The first three items on the list are traditional components of ERP, which is all about driving efficiency by closely integrating these systems. However, Kocsi and team wanted to tightly integrate all five components, in a move that they dubbed "ERP+."
"We rolled up an original estimate of what this was going to cost. It was in the neighborhood of $150 million," said Kocsi. "We sat down with the business and we said, 'Look. We're going to stop working on a whole bunch of other things that are not strategic to the business and we're going to replace about 75 percent of the environment."
The reason Kocsi was able to do that without freaking out the entire business unit was that he went to Mark Shirkness, General Manager of Appliance Distribution Services, and made him a partner in the project. Kocsi said he had heard a lot of great things about Mark as a leader, so he went directly to Mark's office and sold him on the plan. Then he got the CEO to appoint Mark as the business leader for the ERP+ project.
By the end of the first quarter of 2011, Kocsi and Shirkness had their plan.
"We pushed about 70 percent of our resources to [ERP+]," said Kocsi. "We looked at licensing costs and we said 'If we stop doing the other work, it's about $40-60 [million]' at the time over a five year spread and we said, 'That's not bad. The business can digest that.'"
But, as part of the overall technology plan, GE Appliances would also need a modern data center to support the kind of cutting edge systems that it would need to run this type of environment, and that was going to add additional cost to plan.
So, at a time when more and more American companies were outsourcing their data centers to the cloud, GE Appliances decided to build one of the world's first LEED Platinum Certified data centers right in the middle of Appliance Park -- not far from where it had implemented the world's first commercial UNIVAC computer five decades earlier. (It should be noted that a variety of tax breaks and government incentives helped make it economically feasible to put the data center in GE's Louisville facility.)
Even with its aggressive plan to move to a more standard technology stack, getting full buy-in from the business stakeholders, and a new data center on the way to support the whole operation, Kocsi and his team still faced one seemingly-insurmountable obstacle: Time.
After taking 90 days to put the plan together, the leadership team entered the second quarter of 2011 with a lot of enthusiasm about the plan, but no idea how it was going to pull it off in time to help the business start shipping its next-generation appliances by the first quarter of 2012.
Using the traditional "waterfall" project management methods, a project of this scale would have taken at least 18-24 months to implement. It would have meant gathering tons of requirements, creating sequential schedules, and delivering all of the value at the very end when the project completed.
"We knew that there was no possible way of using traditional waterfall [if] we were going to hit any of our dates," said Kocsi.
"We would have overdeveloped," added Shirkness. "We would have been trying to create too much solution, instead of trying to phase it. Our appetite is too big."
So the team started searching for alternatives.
"We also looked around the industry," Kocsi said. "For example, we talked to some folks that had been involved in the Whirlpool [ERP] implementation that broke the business. We did some benchmarking around people that weren't as successful and that was our number one [goal]: 'Don't break the business.'"
What the group ended up doing is the part where this story gets really interesting.
Knowing that it needed to create business value much faster than the usual enterprise IT project, the team decided to try a mashup of bleeding-edge techniques from several different playbooks -- business management, manufacturing, and software development. The result was creative, risky, and highly unorthodox. But, it has been so successful that it may turn out to be a new model for how to run business technology projects and how to organize IT. A decade from now, I expect you'll find a full description of it in college courses and textbooks on business management, and it will be filled with fancy jargon about Agile, Gemba, Lean Manufacturing, Moonshining, and Andon. For now, we'll mostly try to sum it up in simpler terms.
What GE Appliances decided to do was break up the project into smaller chunks. Things that essentially would have been software components or features in a bigger rollout were now treated as individual products, and as they were completed, they were rolled out into production so that users and the business could start benefitting from them -- and improving them -- right away. This approach is based on Agile Development from the software programming world, where the focus is on a continuous set of incremental releases rather than one big software release.
"We started the process in the April/May  time period," said Uhls. "The team went through the Agile process and getting into their sprints and builds and releases. We had our first release in late October."
In other words, within six months of starting the project, the group was already pushing out new pieces of the platform for employees to begin using. But, part of the formula was that they pushed out the release to a department or subset of users, who used it and gave feedback. Then the release got refined and improved and the rollout went out more broadly to the rest of the company.
Uhls said, "Instead of pushing it out to everybody, we push it out to one product at a time. That gave us the ability to learn quick and fail fast. We learned that there were issues with some of the things we needed to do. We were able to work quickly to get that fixed within two weeks and then continue the rollout process."
One of the challenges of the incremental approach is that some processes were moved to the new system while other processes are still in the old systems.
Uhls said, "Because we're doing it in stages, we had to build bridges -- temporary bridges."
Shirkness admitted, "It's a challenge to the user because we're used to delivering the whole pie or cake. It's part of our culture. We're used to getting it all. Now we're just giving them one piece of the cake at a time. And there's a challenge there in saying, 'I'm going to give you something but you can't do your whole job with it.' You kind of have to grow into it. So you get pieces of the functionality, and that's a little unsettling because they're worried about when's the next piece coming, what does it look like? … The user needs to get used to a different way of dieting, of getting their nutrition."
The Big Room
The way that the team overcame the user challenge was to make the users part of the process from the very beginning. They broke down the walls between the business and the IT department and between the technologists and the users. What was the magic formula? They put them all in the same room.
They created a mission control room for the ERP+ project and co-located the IT people, the business stakeholders, and the employees who would eventually be operating the software all in the same room.
Shirkness said, "That's the whole idea behind The Big Room: It's to create collaboration from the get-go and it's really one seamless team. You really can't tell the difference in many respects between one of Mark's business people and one of Kevin's IT people. In many aspects they walk and talk and do many of the same activities. The idea of The Big Room is to get the people that have to make decisions as close together as possible without barriers in between."
"There are no handoffs, by design," said Uhls. "If you're in a circle instead of a sequential, there are no handoffs."
Kocsi said, "We drive decision-making faster. And you just break down all the communication barriers by doing it this way."
This is based on the Lean Manufacturing concept of Gemba, which means "The real place" in Japanese. The idea is that you have to go to the place where the real work is being done in order to create value and make a change.
At Appliance Park, one of the key components of The Big Room where most of the work for ERP+ was being done was visual decision-making. All of the active teams and working groups had charts, diagrams, timetables, giant post-it notes, and other visuals stuck to the walls in their area of the room. In a traditional project, this information would be locked away in spreadsheets, documents, computer files, or employee notebooks. Putting it all up on the walls made it easy for anyone on the team to walk around the room and see the status of all the active parts of the ERP+ project and established a consistent level of accountability for everyone working on it.
It's an analog process in a digital world -- and in the future maybe all of the paper will be replaced by LED screens -- but the visual element is a centerpiece of The Big Room and one of the factors that drives a powerful sense of unity among the larger team. It's also critical for executive engagement in the project, since it gives them an incentive to do Gembas -- or walk-arounds -- on a regular basis so that they can check on the state of the project.
"They've used things like this in manufacturing," said Kocsi. "We've just taken that approach, stole it, and applied it to an IT project. Never done that before."
Shirkness added, "We stole the principles of ‘Moonshine' [rapid prototyping], ‘Andon' [bottom-up quality control], and ‘Gemba' from the manufacturing world here and have been able to apply it… It's mission control but a vast majority of time is actually spent with the user. That's Gemba. Andon is the ability to raise your issues and get a resolution and speed decision-making. So those are the principles that the room is built on that allow us to work with such speed. You can make good decisions and be informed because you're at the place where the work is. You're able to raise your issues and get decisions made [and] evaluate your risk faster. And you're able to experiment and get experience and grow and learn from 'Moonshining.'"
The new normal
One word that you rarely hear from the CIO and the tech leaders at GE Appliances is "IT."
Rob Freshman, IT Governance Leader, said, "It's about business process transformation. It's not an IT project. And that's the difference-maker."
ERP+ still has over a year left before it's completed, but it's on schedule and the team continues to methodically roll out the pieces of the platform. The new data center launched in August 2011, and the whole IT revitalization initiative remains on budget at $200 million.
Most importantly, GE Appliances manufactured and launched its first new product, the GeoSpring Hybrid Water Heater, in February 2012. Shortly thereafter in March 2012, it launched an even more ambitious new product line from Appliance Park with a set of French-door bottom-freezer refrigerators.
The motto for ERP+ that I heard over and over again from Kocsi and the other members of the team was, "Fail fast to succeed sooner." Everyone I spoke with at GE Appliances has embraced the incremental approach.
"Never let best get in the way of better," Kocsi said. "It's a continual evolution and improvement as opposed to designing something perfect up front... I love the concept. I love the visual management. I love the Gembas. I think this will be the way we work on projects for the foreseeable future. "
Ultimately, it's given GE Appliances the speed, flexibility, and efficiency that it needs to produce better appliances. That makes this brand of IT a force multiplier and a competitive advantage.
"It's about turning the company around to play offense against our competitors," Freshman reminded.
IT is playing its role in that. If necessity is the mother of invention, then the IT department at Appliance Park should be strangely thankful that it had to re-think its operations in order to meet an impossible deadline. In the process, it came away with powerful new ways of running the company. And, there are plenty of other organizations that can benefit from the example.
"Seriously, I'm sold," said Kocsi. "I'll never go back to managing an IT department any other way."
- GE taps lean and green to revitalize US manufacturing with slick refrigerator launch (TechRepublic)
- GE thumbs its nose at outsourcing, builds world-class data center (TechRepublic)
- GE unwraps 'Industrial Internet': M2M for planes, trains, manufacturing (ZDNet)
- The Insourcing Boom (The Atlantic)
- The problem with the return of manufacturing (Reuters)
- Factory Jobs Gain, but Wages Retreat (The New York Times)
- New era launches at GE's Appliance Park (Louisville Courier-Journal)
- What's behind enterprise insourcing of IT? (TechRepublic)
Jason Hiner is Editor in Chief of TechRepublic and Long Form Editor of ZDNet. He writes about the people, products, and ideas changing how we live and work in the 21st century. He's co-author of the book, Follow the Geeks.