Nestle USA's costly and protracted struggle with its SAP project is a cautionary tale for any company intent on an enterprisewide implementation.

In June 2000, Nestle SA signed a much publicised $US200 million contract with SAP - and threw in an additional $US80 million for consulting and maintenance - to install an ERP system for its global enterprise. The Switzerland-based consumer goods giant intends to use the SAP system to help centralise a conglomerate that owns 200 operating companies and subsidiaries in 80 countries including Australia.

Not surprisingly, a move of this magnitude sparked scepticism. Anne Alexandre, an analyst who covers Nestle for HSBC Securities in London (the company is traded only in Europe), downgraded her recommendation on Nestle stock a year after the project was announced. While she says that the ERP system will likely have long-term benefits, she is wary of what the project will do to the company along the way. "It touches the corporate culture, which is decentralised, and tries to centralise it,"she says. "That's risky. It's always a risk when you touch the corporate culture."

It is a risk that Jeri Dunn, vice president and CIO of Nestle USA, the $US8.1 billion US subsidiary, knows all too well. In 1997, the company embarked on an SAP project code named Best (business excellence through systems technology). By the time it reaches the finish line, Best will have gobbled up six years and more than $US200 million (the same amount its global parent intends to spend). Dunn now says she sees the light at the end of the tunnel. The last rollouts will take place in the first quarter of 2003. But the implementation has been fraught with dead ends and costly mistakes. It is a cautionary tale, full of lessons not only for its Swiss parent but for any Fortune 1000 company intent on an enterprisewide software implementation.

"I took eight or nine autonomous divisions and said we are going to use common processes, systems and organisation structures,"says Dunn. "[Nestle SA is] looking at 80 autonomous countries and saying the same thing. They're just taking it up a notch. If they go in with an attitude that there's not going to be resistance and pain, they're going to be disappointed."

Nestle's global SAP project, which is tied in to a larger $US500 million hardware and software data centre rehaul, will be integrated with its American subsidiary's soon-to-be-completed ERP. And Dunn is even lending 70 of her own staffers for the global initiative, as well as some of her hard-won expertise. But while the verdict is still out on the global project, the pain - angry employees, costly re-engineering and long periods when it seemed the project would never end - was worth it for Nestle USA, Dunn says. To date, she claims, the Best project has saved the company $US325 million.

Regardless of the project's exact ROI, the lessons learned are real. The primary lesson Dunn says she has taken away from the project is this: no major software implementation is really about the software. It's about change management. "If you weren't concerned with how the business ran, you could probably [install the ERP software] in 18 to 24 months,"she says. Then "you would probably be in the unemployment line in 19 to 25 months".

Nestle learned the hard way that an enterprisewide rollout involves much more than simply installing software. "When you move to SAP, you are changing the way people work,"Dunn says. "You are challenging their principles, their beliefs and the way they have done things for many, many years."

The Problem - 29 Brands of Vanilla

Vanilla may be the world's least exciting ingredient - the word is even a synonym for bland. But that wasn't the case at Nestle USA, where vanilla represented a piquant plethora of inefficiencies and missed opportunities.

Before 1991, Nestle was simply a collection of independently operating brands, such as Stouffer's and Carnation, owned by the Swiss-based parent. In 1991, the brands were unified and reorganised into Nestle USA. Even so, the new company continued to function more like a holding corporation than a single entity. Divisions still had geographically dispersed headquarters and were free to make their own business decisions, although they now reported to corporate Nestle USA executives in California rather than in Switzerland. The new company was trying to introduce economies of scale and common practices, but years of autonomous operation proved an almost insurmountable hurdle.

In 1997, a team examining the various systems across the company found, among many other troubling redundancies, that Nestle USA's brands were paying 29 different prices for vanilla - to the same vendor. "Every plant would buy vanilla from the vendor, and the vendor would just get whatever it thought it could get,"Dunn says. "And the reason we couldn't even check is because every division and every factory got to name vanilla whatever they wanted to. So you could call it 1234, and it might have a whole specification behind it, and I might call it 7778. We had no way of comparing."

While the American brands were willing to go about their business as autonomous companies - headaches be damned - the Swiss parent knew that similar problems would continue. In 1991, the same year that Nestle USA was created, Dunn, then associate director for application systems of Stouffer's Hotels, one of the many Nestle brands, went to Switzerland to help implement a common methodology for Nestle projects worldwide. In 1995, she was promoted to assistant vice president of technology and standards for Nestle SA, where she came up with technology standards for every Nestle company to follow. Dunn figured that common systems across the Nestle empire would create savings through group buying power and facilitate data sharing between subsidiaries.

Yet when Dunn returned to the US to take the more hands-on CIO job at Nestle USA in 1997, she found that few of her recommendations had been acted on. "My team could name the standards, but the implementation rollout was at the whim of the businesses,"she says during a recent interview in her sparsely decorated fourth floor office. Dunn takes cigarette breaks at every possible opportunity and isn't afraid to dress in a leopard-print skirt and blouse. At 47, she is a survivor who is refreshingly open about her mistakes and is respected throughout the company. Her staff speak of her in almost reverential tones.

The Proposal - One Nestle, Under SAP

Dunn's arrival in the northern spring of 1997 came a few months after Nestle USA chairman and CEO Joe Weller coined the term One Nestle to reflect his goal of transforming the separate brands into one highly integrated company. In June, Dunn joined with executives in charge of finance, supply chain, distribution and purchasing to form a key stakeholders team and study what was right and wrong with the company. When the time came, the key stakeholders were initially allotted a little over two hours to present their findings to Weller and other top Nestle officials.

The team baulked at the time limit. "I told them that they would either throw me out in the first 15 minutes or they would cancel the rest of the day, and we would really have a great discussion,"says Dick Ramage, Nestle USA's vice president of supply chain and a member of the team. "It took them an hour, but they cancelled the rest of the day."

"I don't think they knew how ugly it was,"says Dunn, referring to the company's condition. "We had nine different general ledgers and 28 points of customer entry. We had multiple purchasing systems. We had no clue how much volume we were doing with a particular vendor because every factory set up their own vendor masters and purchased on their own."

Soon the stakeholders team presented Weller with a blueprint for major changes they thought could be made in three to five years. While the cornerstone of the recommendation was an SAP package, Dunn says, "We made it very clear that this would be a business process reorganisation and that you couldn't do it without changing the way you did business. There was going to be pain involved, it was going to be a slow process, and this was not a software project."

Despite that warning, it would later become apparent that neither Weller nor the key stakeholders really understood the degree to which the Best project would change the business processes at Nestle or the amount of pain it would cause. "They still thought that it was just about software,"Dunn says.

By October 1997, a team of 50 top business executives and 10 senior IT professionals had been assembled to implement the SAP project. The team's goal was to come up with a set of best practices that would become common work procedures for every Nestle division. All the divisional functions - manufacturing, purchasing, accounting and sales - would have to give up their old approaches and accept the new pan-Nestle way.

On the technical side, a smaller team spent 18 months examining every bit of item data in each division in order to implement a common structure across the company. From now on, vanilla would be code 1234 in every division. The SAP system would be customised around the uniform business processes. In the case of the supply chain, the team decided not to use SAP because the ERP company's supply chain module, Advanced Planner and Optimiser or APO, was brand-new and therefore risky. Instead, Nestle turned to Manugistics - at that time an SAP partner. Manugistics' supply chain module followed all the SAP standards and could easily be integrated.

By March 1998 the key stakeholders had a plan in place. Nestle would implement five SAP modules - purchasing, financials, sales and distribution, accounts payable and accounts receivable - and the Manugistics' supply chain module. Each would be deployed across every Nestle division. For instance, the purchasing group for confections would follow the same best practices and data as the purchasing group for beverages.

Development work began in July 1998. The deadline for four of the modules was Y2K. The new systems would have to double as code fixes and be in place for the millennial change. Nestle USA made the deadline. But its haste created almost as many problems as it solved.

The Process - Nestles Team Players

Even before three of the SAP and the Manugistics modules were rolled out in late 1999, there was rebellion in the ranks. Much of the employee resistance could be traced to a mistake that dated back to the project's inception: none of the groups that were going to be directly affected by the new processes and systems were represented on the key stakeholders team. Consequently, Dunn says, "We were always surprising [the heads of sales and the divisions] because we would bring something up to the executive steering committee that they weren't privy to."Dunn calls that her near fatal mistake.

By the beginning of 2000, the rollout had collapsed into chaos. Not only did workers not understand how to use the new system, they didn't even understand the new processes. And the divisional executives, who were just as confused as their employees - and even angrier - didn't go out of their way to help. Dunn says her help desk calls reached 300 a day. "We were really naive in the respect that these changes had to be managed,"she admits now.

Nobody wanted to learn the new way of doing things. Morale tumbled. Turnover among the employees who forecast demand for Nestle products reached 77 per cent; the planners simply were loath or unable to abandon their familiar spreadsheets for the complex models of Manugistics.

A technical problem soon emerged as well. In the rush to beat the Y2K deadline, the Best project team had overlooked the integration points between the modules. All the purchasing departments now used common names and systems, and followed a common process, but their system was not integrated with the financial, planning or sales groups. A salesperson, for example, may have given a valuable customer a discount rate and entered it into the new system, but the accounts receivable department wouldn't know about it. So when the customer paid the discounted rate, it would appear to the accounts receivable operative as though the invoice were only partially paid. In its haste to unify the company's separate brands, the project team had essentially replaced divisional silos with process silos.

In June 2000 the project was halted. The company removed Marc Richenderfer as project co-leader and gave Dunn full responsibility. (Richenderfer was reassigned to work in Switzerland.) It was time for self-examination. In October 2000, Dunn gathered 19 Nestle USA key stakeholders and business executives for a three-day offsite at a hotel not far from Nestle headquarters.

Jose Iglesias, director of information systems, says the retreat started off as a gripe session. The time constraints necessitated by Y2K had put too much pressure on the people in charge of executing the changes. The project team had lost the big picture of how the various components would work together. And there was still work to be done. The existing modules had to be integrated and the team still needed to roll out two more SAP modules - sales and distribution on the domestic side, and accounts receivable - as well as a new module for the supply chain. Since Dunn had rejected the SAP supply chain module two years before, it had improved and been named a Nestle global standard by Dunn's old standards group in Switzerland. So she decided to replace all but a couple of parts of the Manugistics system with APO. Dunn estimates that last-minute switcheroo accounted for 5 per cent of Best's $US210 million cost.

The offsite group members eventually decided that to finish the project they would need to begin at the beginning, starting with the business requirements then reaching an end date, rather than trying to fit the project into a mould shaped by a predetermined end date. They also concluded they had to do a better job of making sure that they had support from key divisional heads and that all the employees knew exactly what changes were taking place, when, why and how.

The End Game - Sadder But Wiser

By April 2001, the end-state design was complete, giving the project team a highly detailed road map to follow. A month later, Tom James came on board as director of process change for the Best project, having the sole responsibility of acting as a liaison between the divisions and the project team. James says that he was shocked by the still poor relationship between the divisions and the project team. He and Dunn began meeting with more of the division heads. They also started conducting regular surveys of how the employees affected by the new systems were dealing with the changes.

They were not afraid to react to what they found. Dunn says that Nestle recently delayed the rollout of a new co-manufacturing package for six months based on feedback indicating that the would-be users were not prepared to make the process changes that were involved.

ERP projects are notorious for taking a long time and a lot of money. Jennifer Chew, an analyst at Forrester Research (US), found that 54 per cent of respondents to a recent survey said that their project lasted more than two years (the other 46 per cent brought theirs to fruition in less than two years). Nestle USA's project "sounds on the high side"for both time and money, says Chew. Still, success is ultimately measured by what the project accomplishes. Chew points out that Kmart had to write off $US130 million for an ERP project that was never completed.

Dunn herself is not ashamed of the length of the project or the numerous dead ends. She insists that slow and steady wins the race. Nestle USA has already achieved significant ROI, she says, with the largest chunk of savings from better demand forecasting. "The old process involved a sales guy giving a number to the demand planner, who says: 'Those guys don't know what the hell they are talking about; I'm going to give them this number','' Dunn says. "The demand planner turns [that number] over to factory, and the factory says the demand planner doesn't know what the hell he's talking about."Then the factory changes the number again.

With SAP in place, common databases and business processes lead to more trustworthy demand forecasts for the various Nestle products. Furthermore, because all of Nestle USA is using the same data, Ramage says, Nestle can forecast down to the distribution centre level. That allows the company to reduce inventory and the redistribution expenses that occur when too much of a product is sent to one place and not enough to another. Ramage says that supply chain improvements accounted for a major chunk of the $US325 million Nestle says it has saved from SAP.

If Dunn were to do it over again, she'd focus first on changing business processes and achieving universal buy-in, and then and only then on installing the software. "If you try to do it with a system first, you will have an installation, not an implementation,"she says. "And there is a big difference between installing software and implementing a solution."

Nestle USA at a Glance Business Divisions:

Seven (beverage, confections and snacks, food services, foreign trade, nutrition, prepared foods and sales.) Popular Brands: Alpo, Baby Ruth, Carnation Instant Breakfast, Coffee-Mate, Nescafe, Nestle Carnation Baby Formulas, Nestle Toll House, PowerBar, Stouffer's Lean Cuisine, SweeTarts, Taster's Choice.

Annual Revenue: $US8.1 billion.

Number of Employees: 16,000.

Projected Cost of SAP: $US210 million.

IT Staff (including outside consultants): 250.

Nestle Australia at a Glance Established: 1908.

Business Divisions: Seven (beverage, confections and snacks, food services, foreign trade, nutrition, prepared foods and sales).

Popular Brands: Nescafe, International Roast, Nesquick, Milo, Kit Kat, Minties, Perrier, Lean Cuisine, Go-Cat, Lucky Dog.

Five Take-Home Lessons from Nestle's ERP.

1. Don't start a project with a deadline in mind. Figure out the project requirements, then determine how long it will take you to accomplish them.
2. Update your budget projection at regular intervals. So many things happen during a long project that you will be lucky to stay on target during a particular year, let alone the life of the project. Frequently revisiting your numbers will help minimise troublesome surprises.
3. ERP isn't about the software. It's easy to put a new system in place. The hard part is changing the business processes of the people who will use the system.
4. Nobody likes process change, particularly when they don't know it's coming. Include in the planning the people whose processes you are changing. Keep the communication lines open while the project is in the works, and measure the level of acceptance before, during and after the rollout.
5. Remember the integration points. It isn't enough to simply install new systems; you need to make sure that they can talk to each other.
   More about HIS Limited, SAP, Forrester Research, Unify, HSBC, On Target, Billion, Finish Line, Cornerstone, Manugistics