Softwar

by Matthew Symonds

  Buckley doesn’t look any happier. The expression on his face suggests that Ellison is the worst kind of snake-oil salesman. It turns out, however, that one of his firm’s divisions is in the process of doing a 100 percent 11i implementation. Ellison suggests that they audit how it goes and the time it takes to get a return from the software. Sandy Sanderson will follow up and report back. If that works, maybe Buckley will be more inclined to take up Ellison’s suggestion of diverting funds from other projects.12

  The message that Ellison wants to emphasize is that his guests should spend less on their IT rather than more. By stalling some spending and taking his recommended path of putting in 11i without modification, he genuinely believes that they will save money. The key, he says, is speed. Most SAP failures happened because the projects never got finished. The money spent went to IBM and Andersen. Don’t take a three-year flyer like that. The longer a project runs, the less likely it will be to finish. Freeze your budget, and we’ll get you a return in very short order. It’s the 80/20 rule: a project aiming for 100 percent of the benefits will never finish. But if you go for 80 percent, you can get it done in six to nine months. David Lawrence of Kaiser, who has the manner of a kindly M.D., asks whether there is any evidence that going for the last 20 percent of functionality ever paid back. Never, Ellison says. That last 20 percent can cost you tenfold.13

  It’s just the opening Ellison needs to describe the approach he used at Oracle—what he calls spending less to know more. The pitch has now been carefully honed to the new reality of shrinking IT budgets, disenchantment with the Internet, and the absolute requirement for rapid payback. “When we first started planning the replacement of all our local country systems—there were hundreds of them—with a single global e-business system, the planning team told me I needed to increase our IT budget by $250 million and not to expect to see any payback for at least three years. Sound familiar? Of course, our IT budget was already way too high; it needed to go down, not up. We had to find a way to fund our global e-business transformation out of our existing IT budget. So I reviewed every one of our internal IT projects; if a project wasn’t going to pay for itself in twelve months, we killed it. That didn’t mean we canceled every big project. It just meant we had to break up our big projects into multiple steps and make sure we were getting deliverables with benefits and cost savings at every stage along the way. If we couldn’t break a big project into steps, then we killed it.14 Every time we killed something of questionable value, we freed up people and budget to work on the global e-business project and a couple of other cool new things I had in mind.

  “Now we targeted two operational systems that were burning money like it was going out of style. We had so many e-mail systems and Windows file servers that we couldn’t figure out how much we were actually spending on them—but we knew it was a lot. Replacing our ninety-seven local country e-mail systems with one global e-mail system saved a fortune. Replacing our two thousand Windows file servers with one global file-sharing system saved even more than global e-mail.15 These two projects were the closest you get in my business to ‘cheap thrills.’ Both projects were quick and easy to do and had huge near-term payoffs. Everyone should do them. It only takes about ninety days to lower your e-mail and file-sharing costs by 50 to 75 percent. We actually ended up saving more than the $250 million we needed to build our global e-business system. By the time we finished putting in our global e-mail and file-sharing systems and consolidating our data centers, our IT budget had dropped by half.

  “As predicted, our global e-business system did take about three years to put in, but we saved money every step of the way. Every time we folded another local country’s system into our global system, our IT costs went down and the quality of our business processes and information went up. The first local systems we turned off were in Latin America. Separate off a small part of your business—a country or a division—and move that small organization to the E-Business Suite first. The ‘start small’ approach minimizes both risk and cost; you’ll find out very quickly if the suite delivers value to your business. If it does, then you can start adding other countries or divisions into your working E-Business Suite system.

  “We started in Latin America because they were the least automated part of our business and they were the most willing to move to our new standardized Internet business processes. After we got Latin America running smoothly, we started moving other countries into the global system. In our big countries—like the U.S. and Japan—there was a lot of resistance to the change, but it had very little to do with our e-business software. People didn’t want to adopt our new global business processes because it meant they’d have to change the way they worked. The resistance was fierce, but we finally broke through by making the people write down exactly why they couldn’t or wouldn’t change to our new business processes. Making them write it down forced people to carefully study and think through the new process. Most of the time that resulted in them realizing that the new process was okay—98 percent of the cases—but sometimes they found real deficiencies in the software and they wrote down exactly what the engineers needed to do to fix it, which they did. Once our users and systems people were communicating in writing, we started making real progress. We added country after country to the global system in rapid succession. Every time we added another country to the global system, our information got better and our IT budget went down. The less we spent, the more we knew. It was amazing.”

  After Buckley, the two other voices of skepticism are from Wells Fargo’s Kovacevich and from Justin Jaschke, the CEO and founder of Verio, a big Colorado-based Web-hosting specialist that was bought by Japan’s NTT Communications a few months before the dot-com crash for an eye-watering $5 billion. The point that Kovacevich wants to make is that surely there’s a danger of companies’ losing competitive advantage and whatever is unique to them if they slavishly adopt business practices that suit the software they’re putting in. It’s a common misperception that has to be dealt with politely. Oracle’s Steve Perkins, who handles much of Oracle’s federal government business says that compared with SAP, Oracle’s apps are very flexible and you have plenty of choices about the way you want to work.

  Jaschke, not surprisingly, is more of a geek than the other guests. He wants to know what’s wrong with the currently fashionable approach to systems integration in which applications from different vendors are made to work together using specially designed software, known as middleware, that acts as glue, and central data schema created to form a common language. It’s the kind of question Ellison loves. “It just can’t be done very well,” he replies. “If you glue, say, E.piphany and Clarify [two popular CRM applications] together, the customer databases don’t map to each other, so the systems can’t pass data back and forth very easily. It’s like one guy speaks French and the other guy speaks Japanese. You can get them on the phone together and they can hear each other talking, but they can’t communicate. The only thing that really works is a common database schema like we have in the suite. Engineers have tried the data-passing and messaging approach for twenty years without much luck.” IDT’s Jim Courter helpfully adds that he’s trying to do something like that now and “it’s a complete nightmare.” “Absolutely,” says Ellison. “It might work a little bit some of the time, just enough to tempt you to keep trying. You can put wings on a car, drive very fast, and it will take off . . . landing safely is another matter.”

  The next question turns out to be trickier. The air force’s Ron Orr wants to know, reasonably enough, where Oracle has an example of 11i up and running that has stood the test of time. Ellison points out that the suite has been out for only a year but that a few customers had adopted the “no-modifications” model early on out of poverty. One such is Techtronics in Oregon—five years ago its business “fell off a cliff,” so it put in global systems because that was the cheapest way to go. It’s an odd example to give. Sensing it, Oracle’s Mark Jarvis adds that Cisco is a wel
l-known example using Oracle financials to achieve a daily “virtual close.” Somebody points out that it didn’t help Cisco from making a hash of its forecasting.16

  Afterward, when the guests have gone, the Oracle execs strike a resolutely upbeat note, confident they will get solid orders out of the re-evangelized customers. But with the prospects for the economy as foggy as San Francisco Bay, the question is: For all Ellison’s eloquence, how much software will they actually buy?

  * * *

  1. LE writes: Izanami and Izanagi are the names of the two Shinto deities that gave birth to the Japanese islands, or so legend has it. When the local newspapers started pointing out that Izanami was “I’m a Nazi” spelled backward, I had the choice of explaining Shintoism to the reporters at the San Francisco Chronicle or changing the name of the boat.

  2. LE writes: Our new sales force automation software was an Internet system that ran on top of a single global database. We were able to get it up and running very quickly and inexpensively. Guaranteeing a firm fixed price for the software and the labor to put it in seemed the best way to differentiate us from Siebel.

  3. LE writes: We can’t do all the business flows in ninety days. Campaign-to-cash takes longer. But we can do opportunity-to-forecast—global forecast—in ninety days, even in a fairly large company.

  4. LE writes: At the height of this madness, we lost a very big sale to Commerce One because I refused to give the prospective customer any Oracle stock as a part of the deal.

  5. LE writes: It made no sense to release the compensation flow until it was working successfully at Oracle. I had made this mistake too many times in the past. Never again.

  6. LE writes: Gee, I guess I should have named the company “Ellison Systems” and put a huge ELLISON sign on the side of our buildings so everyone could see it as they drive by. And can someone please tell me what “transitory fulfillments” means?

  7. LE writes: And crash he did.

  8. LE writes: I met Mark out on my deck just as I arrived back home from the gym. I was wearing a black cotton tank top with a Sayonara logo on it. I have never owned, nor would I ever wear, a pink tank top. This is very important.

  9. LE writes: SAP wasn’t on the stage with us, but they did officially certify and start reselling Oracle 9i Real Application Clusters under their applications. Interestingly, SAP did not certify either IBM’s or Microsoft’s database-clustering technology. So our biggest applications competitor is now our best database reference. SAP loves RAC.

  10. LE writes: CEOs as a group are incredibly pissed off about how much they spent for their computer systems and how little they got in return. They have a right to be. As for our own consulting organization recommending best-of-breed, well, that happened regularly in those days. All consulting organizations, including our own, made a lot of money pushing very large, very expensive best-of-breed integration projects.

  11. LE writes: I wasn’t feeling smug. I was feeling frustrated. Over and over again I hear about consultants’ recommending overly complex systems that end up costing far more than the customer planned to spend. If customers would demand a price guarantee for the total cost of the system before they signed a contract, it would go a long way toward solving this problem. (Total system cost includes: hardware, software, installation, and annual operating costs.)

  12. LE writes: We did a complete E-Business Suite implementation at one of Brunswick’s divisions. The project was delivered on time and under budget. They’re a very happy customer. The integrated suite approach works.

  13. LE writes: Even if the software could be made to do 100 percent of the things customers say they need, there are still a couple of serious problems that remain. First, the customer may not fully understand each and every one of its own needs, and second, even if it does, those needs are likely to change over time. A far better approach is to buy a system that delivers 80 percent of the benefits of a “perfect” system at 20 percent of the cost. After the 80 percent solution is installed and delivering substantial benefits, you then have the option of incrementally enhancing your system’s automation and information capabilities a little at a time. Going for a 100 percent solution requires a perfect understanding of your business before the project starts and a lot of customization to make the software fit. That is a very expensive and risky approach; you might end up with nothing.

  14. LE writes: The longer a project runs without delivering value, the greater the likelihood it will never deliver value. Go to zero-balance budgeting for IT projects, and beware of throwing good money after bad on long-overdue projects.

  15. LE writes: Both our global e-mail system and our global file-sharing systems are based on what has become the Oracle Collaboration Suite.

  16. LE writes: As I’ve said before, our forecasting system is not clairvoyant. Our forecasting does statistical extrapolations based on historic trends. If something that’s outside our mathematical model of the business changes, like a war in the Middle East, our forecasting becomes inaccurate.

  12

  HUNGARIAN LESSONS

  Hungary

  May 2001

  The customer Larry Ellison refers to more than any other when selling the E-Business Suite is GE. It’s a good choice: there are few more admired companies.1 But the bit of GE he’s improbably declared a model 11i customer is in a far-flung outpost of Jack Welch’s sprawling imperium.

  The road east from central Budapest toward the tiny market town of Veresegyház some forty-five minutes away is not a bad metaphor for modern Hungary. The capital itself is a job half done. There are plenty of new buildings—mainly banks and office blocks—and a concerted effort is under way to refurbish the grand government buildings, such as the absurdly Ruritanian Parliament that sits on the Buda side of the Danube and is covered in scaffolding. The road to Veresegyház, narrow, rutted, and bordered with ads for the latest in mobile phone technology, carries a mixture of modern European cars, a few wheezing Trabants, and, as the tatty suburbs give way to open countryside, increasing numbers of horse-drawn carts. On a warm afternoon in early May, the rolling landscape of neat fields is unspoiled and beautiful, until the view is interrupted by a shining new steel-and-glass factory that stretches across more than 500,000 square feet and has changed forever the view of the valley. There’s nothing to indicate what goes on inside—a reasonable guess might be some kind of agribusiness plant producing fertilizer or turning out farm machinery. Except that every few days a noise that blasts out of one of the huge metal containers lined up outside one end of the plant sounds oddly like a Boeing 747 taking off.

  This building is a $100 million investment by the Power Systems division of America’s mighty GE, put up in record time to make replacement parts for industrial gas turbines—highly sophisticated nozzles, shrouds, and combustors—and to manufacture what are known as “aeroderivative” power packages. These are $15-million-a-pop transportable gas turbines that harness the thrust of GE jet engines to produce up to 50 megawatts of electricity—enough to run a large factory or keep a small town humming. Before each of the turbines is disassembled and packed into twenty or so containers for shipping, it is given full load testing to verify its power output and heat rate. Within a year of the groundbreaking for the factory construction—even before the offices for site managers and engineers were completed—the first turbine package rolled out of the gates, on its way to a customer in Paris.

  The man who makes it happen is Craig Kipp, a tall, athletic-looking forty-five-year-old who was yanked out of his job at GE’s nuclear fuel business in Wilmington, North Carolina. When he arrived in April 2000, Kipp knew that he was facing the challenge of his life to get the plant up and running in the time demanded by his unrelenting bosses in Atlanta. What he didn’t know was that Oracle’s Larry Ellison would turn out to be staking even more on his success or failure than either GE’s legendary Jack Welch or even John Rice, the hard-charging new head of the Power Systems division.

  GE has been in Hungary for severa
l years and has several major businesses in the country. The first, Tungsram Lighting, was acquired in 1990 in the first major phase of post-Communist Western investment. Since then, GE Capital has bought Budapest Bank, and GE’s Medical Systems and Engine Services divisions have established outposts. Welch early on decided that Hungary would be one of the leaders in the Eastern European pack and the decision to build the turbine plant wasn’t a difficult one, with the country expected to accede to European Union membership by 2005. As Kipp puts it, he can plug into an existing GE infrastructure for support services, and he has access to low-cost, well-educated local engineering talent. What’s more, for American expats, Budapest is a “livable city.”

  However, within GE, the significance of Veresegyház would be more than just a new facility to help meet the booming demand for the firm’s power generation products. Under Welch’s project to digitize every aspect of GE’s extraordinary spread of businesses, Veresegyház was to be given the status of a laboratory to test how far, using Internet technologies, the power systems operation could be transformed into a full-fledged e-business. There was also another consideration. The speed with which Welch and Rice’s predecessor, Bob Nardelli, now boss of Home Depot, were insisting that the plant come on stream meant that a traditional approach to systems integration wouldn’t work, particularly as the other outfits in GE that Veresegyház would be dealing with, such as oil and gas, based in Italy, and energy products, based in France, had a mishmash of ERP systems from different vendors.