A Day in the Life of Celanese’s Big ERP Rollup

In the go-go ’90s, large companies everywhere decentralized and customized their ERP systems. Now that the party’s over, it’s time to clean up and work on the difficult task of integrating many systems into one. This kind of systems rollup will be the predominant type of ERP project CIOs at large companies will face during the next half-decade.

At Celanese, a $4.5 billion global chemicals maker, rollup means integrating many SAP systems?plans called for 13 when the project started, but that’s been scaled back to seven?into one. Celanese’s project, called OneSAP, began in summer 2001 and won’t end until mid-2004?1,000 days. This story is about one of those days, Oct. 3, 2002, with Celanese’s IT group in Dallas, when CIO Karl Wachs and his project leaders are finalizing the system’s blueprints. The day starts early.

[6:45am] Project Leadership Morning Meeting

Where Is That Software Module?

The purple night sky is turning morning blue, but the sun hasn’t yet risen over greater Dallas. Two dozen Celanese project leaders are settled into their regular morning meeting, taking turns giving progress reports on OneSAP.

Wachs is here, though he usually avoids these meetings for fear his team will speak less freely with the boss around. Standing in front of a business process flowchart on the wall, Wachs sips a Pepsi One. (This could be a product placement; Celanese makes Pepsi One’s sweetener.) One of the project managers introduces this reporter, perhaps as a caution to the team to behave. Across the room, a staffer leans back in his chair. “So, have you chosen a genre for your article yet?” he asks. “Suspense? Thriller?”

The room goes silent as everyone waits for a response. “Horror story?” I suggest, smiling.

As nervous laughter cycles through the room, Wachs points his soda can at the team, smiles and says in a thick German accent, “There. That’ll get the juices flowing. Let’s go.”

Wachs and his team know that OneSAP will sound like a horror story to those unfamiliar with the rollup concept. Celanese manages 13 SAP systems (versions 3.1, 4.5 and 4.6) scattered across five data centers that serve five business units. (Since our visit, Celanese sold off one business unit, leaving only 10 SAP systems.) OneSAP, conceived by Wachs after he received marching orders to cut costs (and for other reasons that will become clear), will roll up the systems into one instance of the enterprise software. At the same time, OneSAP introduces several new and critical features to Celanese, such as supply chain forecasting. After an 11-month approval process that ended in June, the board gave Wachs 21 months to pull it off.

Celanese officially declines to say how much OneSAP will cost. Wachs told CIO in a previous interview that it will cost between $100,000 and $200,000 per day. A project manager let slip a figure of $60 million. ERP experts say that’s a reasonable number. Celanese believes OneSAP will pay for itself two years after it’s fully deployed.

The meeting moves efficiently through a litany of status reports, using a red, yellow, green system for marking progress on different elements of the project. While certain aspects such as the supply chain planning group are in yellow (between two and five days behind), OneSAP as a whole is in green. It is about 1 percent, or less than two days, behind schedule.

The meeting is marked by German, Indian, British and Texan accents speaking the languages of SAP, Six Sigma, project management and Celanese.

One acronym, SCEM, doesn’t whiz by. When it’s brought up, the meeting takes on a concerned tone. SCEM, Supply Chain Event Management, is a new ERP feature from SAP?still in beta development?that is meant to help a company like Celanese better manage its supply chain. The ICE (interfaces, conversions, enhancements) group at Celanese needs the vendor’s SCEM beta code more or less today.

“What can easily drive us into red,” the project manager who needs SCEM says, “is not having that software.”

ERP, a Two-Track Mind-Set

One way to look at major ERP projects is as if they are railroad tracks. There are two rails to follow, a technical one and a cultural one, which are tied together. There’s a sense at Celanese that the technical rail?the code, interfaces, documentation?is easy. Even building OneSAP doesn’t much worry Wachs, or anyone else in the room. Culturally, though, OneSAP is the hardest project in the world.

That’s because Celanese has traditionally operated as a holding company, letting its five business units run themselves. Over time, they developed a fierce independence, which made it hard for Celanese corporate to make decisions for the whole company. Cam Carlson, the project director, will say twice today: “Historically, Celanese’s culture has not been so good at decision making.” Over time, the staff says, Celanese earned its reputation for building consensus and avoiding decisions. (Under current project management wisdom forwarded by gurus like Patrick Lencioni, consensus is bad. It means you’re trying to placate everyone; it leads to mediocrity.)

OneSAP is meant to change all of that?to get the five businesses sharing processes and systems while accepting that they are accountable to Celanese as a whole. “The real project is, we’re trying to become one company, and this is how we’re doing it,” says OneSAP Integration Manager Russ Bockstedt.

Culturally, that means Celanese must turn itself inside out. And to do that, Carlson says, “we’re going to have to make people feel pain to an almost near-death experience. That’s the only way to get to the ultimate goal.”

If Wachs pulls it off, Celanese wins big. OneSAP means decommissioning 14 major systems?both ERP and non-ERP?and the costs that go with them. Wachs has talked to peer CIOs who have realized operational savings in the 30 percent to 50 percent range by reducing the waste that comes from doing the same thing five different ways in five different business units. Analysts suggest returns can be had in less than two years. OneSAP has the power to transform the company.

Or, if it fails, debilitate it. At a time in the chemicals industry when the cost of raw materials is up and revenues are down, Celanese has set a goal to double sales in five years, leaving behind peers like BASF and Eastman Chemical to compete with the Dows and DuPonts. Part of that revenue doubling will have to come from acquisitions. To pay for them, Celanese must boost its stock price (which, for three years, has bobbed around $20) and keep costs in check. All of that depends on the rollup to OneSAP. If that rollup fails, so does Celanese’s aggressive business strategy.

“If we don’t do it,” Wachs says, “we’ll be second class. We have to do it.”

8:00 AM Workstream Design Meeting

Inside the Sausage Factory

On this day, OneSAP is mired in the muck of the blueprinting phase, when integration happens at an arduously technical level. Developers here will argue over the implications of the ship to schema versus the sold to schema in SAP’s Advanced Planner & Optimizer module. Before that, though, the subject of the tardy supply chain event management module comes up. If the technical team doesn’t get the package from SAP right away, the team will have to postpone its work on SCEM for four weeks for scheduling reasons?and put the project behind.

“Rollups are not so much about replacing ERP as they are about redeploying what you have,” says AMR Research analyst Jim Shepherd. “Decentralized businesses are inefficient, and everyone needs to be more efficient now. They want centralized financing, for one, given they have to sign off on this stuff now. Customer requirements, like single invoicing, are driving them back to one system. And they’re also realizing that a single view might give them opportunities to cross-sell and upsell customers between business units.”

Wachs boils it down. “Top management calls this inefficiency. The lower guys call it stupidity. When you have five different pricing models, it creates inconsistencies,” he says.

The integration meeting plods along, led by Bockstedt, the former CIO of Celanese’s Ticona plastics business (where he oversaw that SAP deployment). Today, the team is checking progress on tasks such as batch determination and demand forecasting. The accounts payable scanning feature is late. The team says it will call a consultant who already wrote 3,000 lines of code that was supposed to make that feature work. Bockstedt presses, “Should we descope it?” Representatives from the finance group convince Bockstedt to keep it in scope for now.

True Integration Requires Faith

There’s something religious about embracing ERP. It requires faith that the destination will justify a long period of sacrifice and change. And it requires discipline.

The late ’90s were secular, and its distinctions?independence, entrepreneurship, decentralization?can be seen as euphemisms for what was really going on: an utter lack of discipline. Instead of adapting to ERP, large companies bought ERP, hired consultants, dismantled the software and rebuilt it to fit existing business processes. They were designing religious tenets around how they lived rather than living by religious tenets.

The Celanese ERP rollup is about rediscovering religion. Instead of tailoring software to the business, the business conforms to the software. Of course, that deliberately brings pain, as Carlson noted, and demands that people change. The obvious risk here is that too much pain can maim or kill the organization, that is, if the company can even find the discipline it lacked in the first place.

Wachs says he’ll know when discipline flags. “If we tolerate business units explaining why their outputs are different so that they don’t have to change their inputs, then we have lost,” he says. “The mistake of the past was trying to adapt software to the business.”

Eventually, the integration meeting gets stuck in the mud. Sensing this, Bockstedt gives it a push. “This is what you’ve agreed to do over the next week,” he says, calling up another spreadsheet. “It’s quite a bit.” For a while no one speaks. SCEM is still hanging out there, unresolved. Finally, one staffer speaks up: “We’ve communicated with SAP by phone. By e-mail. We got no response. Should we send Karl in to do something?”

Leaning back in his chair, Bockstedt says, “I think it’s time to pull the trigger on that.”

8:57 AM CIO Karl Wachs?s Office

Call Him Confident. or Crazy.

Wachs speaks rather softly. His demeanor echoes his business-casual dress. His most demonstrable emotion is puzzlement?his face goes blank in an instant, and he volleys your questions back as terse rhetoricals. It is not rudeness so much as it is incomprehension. For example, Celanese recently snapped up a $150 million emulsions company, Clariant, that now must be slipped into the OneSAP strategy. Wachs is asked if that adds risk to OneSAP. Baffled and agitated he shoots back, “What are you going to do? What? That’s normality.”

When Celanese AG demerged from Hoechst AG in late 1999, the company launched a slew of integration initiatives known internally as One Celanese. Hiring a global CIO was part of the effort. Celanese found Wachs at the Mount Olive, N.J., wing of chemicals giant BASF, where he was director of systems integration. Prior to BASF, Wachs worked with the North Jersey Media Group, managing IT for the company that publishes The Record of Bergen County, N.J.

Wachs started at Celanese in 2001 with no direct reports and one directive: Cut costs.

First, he hired Paul Peters, someone who embraced this cause. “We saw $30 million on the table if we could consolidate IT,” says Peters, the project manager for OneSAP. “We took information from 30 other chemical companies. Everyone was moving toward consolidation.”

SAP was a fat target. Wachs paid consultants to look at his options. They came up with three: Do nothing; consolidate by division, with some governance; or roll up into a single instance of SAP. Each was riskier and more expensive than the last, but each also carried exponentially greater benefits. Wachs spent the better part of the next year building a business case for OneSAP.

Wachs’s soft-spoken exterior belies what his team says about him. “He is a change agent,” Peters says. “A spark.”

“He doesn’t mind being the only one on the other side of arguments,” says Tony Perroni, the lead BearingPoint (formerly KPMG) consultant on the OneSAP project.

“Karl’s fearless,” Peters adds. “He likes?OK, he loves?change and friction.”

This explains how Wachs convinced Celanese’s rabidly independent business units, despite strong resistance early, that OneSAP was tenable. He simply didn’t care that he would, as Peters put it, “get shot down every time he told them we were doing it.”

“They saw this would cost money, and they saw very few benefits,” Wachs recalls of the early meetings with business unit IT leaders. “They were trained to be independent. When we got them to understand they weren’t perfect and they told their bosses that this could be a good thing, that was a major milestone.”

Bockstedt, then part of the Ticona plastics division, was skeptical. “I figured we’d end up with three regional systems. But Karl convinced us. Without him, we wouldn’t be doing the project,” he says.

The Approval Campaign

“There was no one event that made us go in this direction,” Wachs says. “It was a process of research. There were 70 people building the business case. Never, ever start this if you don’t understand the risk. This is not a four-week decision cycle. It should take a year for approval.”

In fact, approval came after about 11 months of stumping, in mid-2001. Then Wachs recruited his troika of project leaders, Carlson, Peters and Bockstedt, with whom he felt comfortable leaving the day-to-day operations. That freed up Wachs to deal with OneSAP at a higher level, which is best for the project, Carlson says. “Karl is not so much a detail person. In fact, he is not that at all.” (Wachs himself says: “I’m not here to manage the project. If I am doing that, I am not doing my job.”)

Instead, he’s thinking six months out, about progress reports to the board. He’s consulting peer companies. On this day, he has a major budget meeting with COO David Weidman.

Wachs also thinks about big-picture issues like burnout. “We’ll send the team away for two weeks at the end of the year. We’ll say, Go home, because we need 150 percent next year,” he says.

If Wachs’s soft-spoken exterior hides a tenaciousness, it also reveals a great deal of confidence. He says ERP “is not rocket science anymore. It’s not easy, but it’s mechanical.” He says nothing about this project keeps him up at night. He’s irked each of the several times he’s asked about the possibility of failure, as if the question is a non sequitur.

“I am not naive,” he says, finally. “Proposing this to the board was the most difficult thing of my life.” Wachs leans back in his chair. His feet are up, his hands clasped behind his head. “If this project goes bust, I will have serious issues.” He laughs. “If I were you, I’d be asking, Is this guy confident or stupid?” He laughs again, harder.

11:20 AM Lunch with Cam Carlson and Russ Bockstedt

The Three Types of Pain

Over lunch, Carlson, the project director, and Bockstedt, the integration manager, lay out OneSAP’s structure.

They run the show with project manager Peters and BearingPoint’s Perroni. Bearing Point has 30 consultants on the job averaging 10 years of SAP experience. (The last time Celanese built an SAP system, it used 86 consultants and 30 employees. This time, it’s 70 employees to 30 consultants.)

The project has seven tracks. Four are functional: finance, supply chain management, manufacturing and order-to-cash. The other tracks are business intelligence reporting, technology and change management. Each track has several functional stakeholders?men and women responsible for overall progress. Each track also has one business process owner (BPO). “While the functional stakeholders work on the day-to-day, the BPOs work at a higher level,” Wachs says. “For example, the finance BPO is talking with the CFO and controllers to design the business processes for that organization.”

Celanese uses training principles from the Project Management Institute and the Six Sigma philosophy of “total quality management.” OneSAP gets money in predetermined chunks, when the team hits its milestones. Risks to the project are scored every month (see “The Top Eight Risks to the OneSAP Project,” this page). And project leaders vouch for what they’ve accomplished in regular validation sessions.

Over the course of December, Celanese held the blueprint sign-offs?when more than 100 BPOs, functional stakeholders and other key players signed (“in blood,” Wachs says) a document that finalized the OneSAP blueprint. Planning ended there. System building began.

Like Wachs, both Carlson and Bockstedt say they are plenty confident that blueprinting and building this system isn’t hard, just laborious. And like Wachs, they fret more over cultural issues. They see two.

First, Celanese must fight its own consensus-building heritage. That is why the morning meetings moved at a no-nonsense pace. It was part of a conscious effort to change how Celanese operates.

“We want adequate analysis but not paralysis,” Bockstedt says at lunch.

“We’re seeking consensus only when it’s appropriate,” Carlson adds. “We are saying, What are the facts? Let’s move on. It hasn’t always been like this.”

And Celanese hasn’t mastered it yet. Startlingly, it’s learned during the day that certain business units have fought for, and successfully kept, their SAP systems out of OneSAP’s scope. Seven of the 10 existing enterprise systems will be decommissioned, composing 90 percent of the business. But the U.S. corporate HR system, an outsourced HR system in Germany and a few other small systems staved off inclusion in OneSAP, for “legitimate business reasons,” says one team member.

That could be prudent. But it also could be Celanese’s consensus culture creeping in.

The second cultural challenge is managing the massive pain Celanese will bring to its 12,000 employees. It’s Carlson’s job to manage that change. And as he sees it, there are three kinds of pain he’ll bring: “Jobs with new responsibilities. Jobs that change completely. And jobs that, yes, go away.”

1:35 PM Project Management Office Meeting

Who Gets to Go First?

With the lights dimmed, the senior leaders of OneSAP look at a screen where they see projected two potential rollout strategies for OneSAP.

If the project stays on schedule, it will take about nine months to build the OneSAP system, according to the blueprint. Then Celanese will start deployment in the fourth quarter of 2003, and the rollout will run through the second quarter of 2004. The subject of this meeting is basically, who draws the short straw? The leaders are discussing two options.

Acetate, one of Celanese’s five divisions, is planned for later in the rollout because it has one of the newest versions of SAP already running within Celanese. Someone suggests moving Acetate to the front, because its version and the new OneSAP are so close, it will be less difficult for those employees to adjust. But Perroni interjects, “We still have major change management issues with Acetate. As late as last week we were dealing with the question from them, Why are we in this?”

Someone else suggests, why not move Celanese’s new acquisition, Clariant, to the front of the line? The acquisition was so recent that no one is sure what version of SAP Clariant uses. “Version 4.5b-ish?” Perroni theorizes. But the sooner Celanese gets Clariant on OneSAP, the sooner Celanese stops paying Clariant’s monthly six-figure support fee (a figure that Celanese says was later reduced). Still, someone else offers, how will it play if we acquire this company and foist this pain on these people right away? Not well, seems to be the consensus.

Perroni offers a new option: Combine all of the small rollouts, leaving plastics from Ticona and chemicals (roughly 80 percent of the total system) for later. That eliminates lots of risk early while still giving the team rollout experience it can apply to the big dogs later. “We could do Mexico, Singapore…,” Perroni says.

Peters is interested: “If you think we could do it that way, that becomes very attractive.”

After more brainstorming, this new third option?small sites first?is called “doable” and Peters is jazzed, calling it “clever” and asks, “How reasonable is it to gain a month next year so we can start these small sites in August?”

For the most part, a three-year project moves like one of those revolving rooftop restaurants. You don’t feel the progress, but now and again when you look up, you’re in a different place. But here, for a second, you could feel the motion. You didn’t need to look up to know OneSAP was in a different place now.

Postscript

While Peters, Carlson and Perroni are figuring out how to bring OneSAP to life, upstairs Wachs will e-mail SAP to press the vendor on the supply chain event management issue. SAP, Wachs will learn, doesn’t think the beta code is ready to be shipped. Not good for the OneSAP schedule.

The e-mail exchange will turn into conference calls. And Wachs will have his CD-ROMs within four weeks’ time. The supply chain software module won’t put the project behind schedule.

Wachs will also attend a key budget meeting with the COO Weidman. He’s probably showing Weidman what he once described as the hockey stick curve?”that shows the tremendous expense now will lead to a future that’s bright, I promise.”

In his office, looking out on the LBJ Freeway, a shiny ribbon in the morning sun, he says, “You see why we have to do this? If we don’t go to one system, our five units will drift apart. Looking at other companies doing this, we see substantial cost reductions. You know nearly a half in some cases. All that cost and complexity can go away.

“But if all that goes away,” he says, smiling, “what is my job?” He laughs.