Softwar

by Matthew Symonds

  A few days later in Paris, Ellison seemed unusually nervous and flustered. The main theme of his speech was on the folly of customizing Oracle’s software and why it was better to buy a complete suite from Oracle—in Ellison’s analogy, a finished car that you knew would work—rather than a bundle of kit glued together by mechanics from IBM. He said, “We’ve worked hard to be the first car company. We don’t want to sell you parts and labor. If you customize our software, when we bring out a new version you can’t use it. You trap yourself in an old version.” Ultimately, customizing the E-Business Suite was futile because Oracle itself would soon add the functionality customers were looking for. “We will bring out a new version of the software before you can finish customizing our old version.”6

  There was nothing wrong with the familiar argument. But without a prepared text, Ellison sometimes puts things in starker terms than he—and certainly Oracle’s marketing department—intends. A significant part of the speech was a rant against system integrators—evil folk who persuaded customers to part with huge sums of money to let them mess up perfectly good software. As usual, IBM was singled out as the worst villain: “IBM has thousands of consultants who are eager to help. IBM has alliances with five hundred software partners. IBM has thousands of guys with glue guns who come and stick it all together. At least that’s what I think happens. What is systems integration, anyway? What do those thousands of people do all day?” Unfortunately, most of the AppsWorld sponsors just happened to be systems integrators. Mark Jarvis says, “I was just thinking how proud we were of ourselves that we had done a great first AppsWorld, when we suddenly picked a battle with every systems integrator in town.” Oracle had lined up a series of anti–systems integration ads. One said, “Just Say No to Systems Integration,” while another asked, “Do You Have the Big Systems Integration Blues?” Naturally, “Blues” was written in the style of IBM’s logo. After Paris, they never saw the light of day.

  The other big “take-out” from the Paris speech was Ellison’s frank recognition that the E-Business Suite didn’t yet do everything its customers wanted and maybe never would but that an “80 to 85 percent solution” that could be implemented in five months was better than the 100 percent promised by some vendors that would take three years to put in and still wouldn’t work. Immediately, the chorus went up that Ellison had claimed that the suite was complete and now he was saying that it wasn’t. The following week, when AppsWorld moved to New Orleans, Jarvis persuaded Ellison to ease up on the systems integrators. However, although the speech went down much better than the one in Paris, once again there were unintended consequences. Jarvis says, “A whole new thing came out, which was that he gave everyone the impression that you had to have the entire suite or nothing. I’ve looked at the videos many times. He didn’t actually say that, but that was the impression he created.”

  Ten days later, on the first of March, after the markets had closed, Oracle issued a profit warning for its third quarter. It now expected to miss its earnings-per-share number by 2 cents. Although applications had continued to grow strongly thanks to the buzz that had been generated around 11i, database revenue growth was flat to negative. Ellison explained, “It’s very disappointing. A bunch of deals were approved at the senior vice president level, but once they got to the CFO and CEO level, they were pushed off. We have a lot of nervous executives looking at this economy and being very cautious. They want to wait thirty to sixty days so they can get a better read on this economy.” In after-hours trading, Oracle’s stock fell 21 percent, to $16.88, about half the price it had been as recently as early January.

  When Oracle got around to making its official earnings announcement in mid-March, the mix was different but the overall result was the same. But while database did a little better than predicted, applications revenue was slowing at a worrying speed. From the 75 percent growth projected by management on February 14 to the 50 percent expected on March 1 at the time of the profit warnings, the real figure had turned out to be only 25 percent. It didn’t take a genius to see that not everything that was going on could be explained by the weakening economy and edgy CEOs waiting for “visibility” to return. For anyone who wanted to see, there was mounting evidence that it wasn’t only the economy that prospective Oracle applications customers wanted to see stabilize.

  To complete a miserable few weeks for Ellison, from the moment of the earnings warning legal vultures had been stirring up their clients to file class action suits against Oracle for misleading investors about its numbers and the true state of the E-Business Suite. One law firm, Milberg Weiss, alleged that Ellison was involved in the “largest case of insider dealing in U.S. financial history.” It claimed that he had dumped nearly $900 million worth of shares during the third quarter, knowing that the stock price was likely to fall soon. The suit, on behalf of a small pension fund, accused Ellison of concealing information about the cost of the “massive technical problems” of 11i. In fact, the shares that Ellison had sold, his first disposal for five years, were stock options that were due to expire in August. Oracle said that the suit was “entirely without merit and would be defended vigorously.”

  Adding insult to injury, the fashionable technology consultancy Forrester Research leapt on the speeding anti-Oracle bandwagon with a particularly vicious briefing note, charging Ellison with “stooping to a new low—blaming its 11i unhappy customers for doing too much customization.” The note, by an analyst named Laurie Orlov, attacked the whole proposition of the integrated suite, blithely advising Oracle: “By unhinging apps like CRM from tight links to order management, the individual apps can be coupled with apps from other vendors and can then be revised or released in chunks.” Only three months earlier, Forrester had enthused that its “all-integrated approach will make Oracle outperform any best-of-breed technology vendor in the long run.”

  As usual, Chuck Phillips brought some level-headed perspective. His view of the numbers was philosophical: “Is this the beginning of the end? We don’t think so. Every time Oracle misses a quarter—which is about every five to seven quarters—there are a rash of proclamations about market saturation [database] and fatal strategy missteps [usually applications] that point to eternal doom. Even in the best of times, Oracle has not produced smooth license or product line growth from quarter to quarter. The numbers generally fall where they may without, much to our chagrin, a ton of backlog management. Billions of license revenue gets booked in the last five days of the quarter and it’s difficult to smooth out that business model. Usually, investors throw in the towel on Oracle for three quarters and don’t want to hear the name mentioned for a while. But then, comparisons get easier, customers eat through their licenses and need more capacity, the product footprint expands into higher growth areas, the buying cycle improves with the economy and recently released products get seasoned and get some traction in the installed base. The only cure is time.”

  The mood at Oracle was bloody but unbowed. There was an embarrassed realization that it had fallen down badly on execution in sending a hugely important new product out to customers in an unfit state. But at the same time, there was complete conviction both that the suite message was right and that Oracle was well on the way to dealing with the outstanding quality issues. Talking to me nine months later, Ron Wohl said, “Fundamentally, we tried to do something that I think was much larger in scope than even we realized and was much larger than the scope of any software we had released beforehand. In retrospect, we didn’t do a good enough job up front—our QA [quality assurance] procedures didn’t work because they had outgrown our scale.

  “I would have liked to have done two things differently: we should have changed a number of processes within development to deal with the scale of the release, which we have now done; and two . . . we would have been wiser to withhold the release a little bit. We would have been better off as a company if we had deferred the release by three or four months. The stability actually improved at a very rapid pace. You coul
d see it in the bug numbers, you could see it in customer satisfaction by anecdote. Customers who had started on a later version [11.5.3 and above], they were extremely happy. Customers who started on one of the early versions weren’t happy, and they had a right to be upset with us.”

  Mark Barrenechea thanks God for the recession: “We actually got a little lucky that the economy turned down. It gave us more time to deal with the problems. Of course we’d like the satisfaction of stabbing our competitors before they hang themselves. But the downstep in the economy meant that a lot of them were just dying. It couldn’t have happened at a better time for us. It meant that there were fewer choices among the best-of-breed vendors and customers were more nervous about making those choices. Secondly, in a boom economy, there’s a lot more IT dollars. If something isn’t working the customers can say, ‘Fuck you’ and go to another vendor, spend some more money. So companies weren’t willing to double down on their mistakes and bet the farm, they couldn’t abandon us. So they forced Oracle, and we were forcing ourselves to walk through the issues and fix the problems. To quote John Milton, I was not displeased that there was a downturn in the economy.”

  Ellison is rueful but philosophical. There’s little in the software business left to surprise him. “A little over ten years ago we undertook a massive rework of our database software. It strained our company to the limit, but it resulted in a database product that was so much better than anything offered by the competition that we became the number one database company in the world. We have to make big leaps forward in technology or sink slowly into obsolescence. The pain is worth the gain. It’s very much like childbirth. It’s a lot of pain followed by great joy. The joy helps you forget the pain. When Version 7 of the database came out, customers loved it. The pain of Version 6 was largely forgotten. The applications release 11i birthing process was not nearly as painful as Version 6 of our database, and I think the 11i gain will be even greater. We’ll see. People haven’t forgotten the pain yet. In fact, during the worst of our bad publicity and our sinking stock price, somebody said to me, ‘Oh my God, this is as bad as 1991.’ Jeff Henley and I both laughed. In 1991, we were on the brink of bankruptcy. At the worst point in this recent downturn, we had $6 billion in cash and profit margins of thirty-five percent.”

  Ellison had previously said that the only thing that would make any difference was a quarter that beat expectations. “Most analysts try to foretell the future by understanding and explaining the recent past. If you have a bad quarter, if your sales are going down, they will attribute part of that sales decline to product problems that are creating competitive pressures. If you suddenly have a good quarter, better than the competition, they explain that it’s because your product is getting better. It’s as simple as that—they just look at the numbers. So the best way to improve perception of the E-Business Suite is to sell more of it. Then the analysts will say, ‘Good job, Oracle, it’s a great product you got there.’ ”

  Despite everything, at least the idea of the integrated suite was beginning to win acceptance, not just from customers but even from rivals, like SAP and PeopleSoft. George Roberts soon began to notice that he no longer had to bang the table with suite-versus-kit arguments. “If you have a strong strategy, it forces your competition to respond. Whether the release was premature or not, every major competitor has been forced to respond to the Oracle message and the value proposition that we brought out with 11i.” The initial execution might have left much to be desired, but Oracle is a company that learns from its mistakes. Oracle, and Ellison in particular, have a reputation for arrogance. But during the difficult first six months of 2001, I discovered an impressive determination to face facts and put things right combined with an iron conviction that the E-Business Suite was still destined for greatness. Establishing it would not be the early-round knockout that Ellison had predicted, but as the going got tough, his certainty of ultimate victory, if anything, seemed to increase.

  * * *

  1. LE writes: We were getting a lot of pressure from our customers to deliver the 11i applications. I made a lot of them mad by delaying the release for several months. I should have delayed it even longer.

  2. LE writes: The biggest lesson I learned from the release of 11i was to make sure we used the applications internally before we gave them to customers. That’s how we operate today, and it’s made all the difference. Once we’re using an application successfully at Oracle, it’s much less likely that a customer will run into problems.

  3. LE writes: Not true. Ron explained the risks, and I understood the uncertainty surrounding release of the new order management system. And I made the decision to release it.

  4. LE writes: This accidental discovery that our customers were uniformly more successful and happier with our applications products when we were managing the software for them—upgrading to new versions, fixing bugs, tuning performance—caused us to rethink our approach to applications delivery. E-Business Suite Outsourcing is now the fastest-growing business inside Oracle.

  5. LE writes: Nonsense. Of course Ron knew about the problems with our internal order management implementation. He was personally running the project to put it in and make it work. We planned to be down for a while, and we were.

  6. LE writes: Worse than being overzealous in this presentation, I was ambiguous, imprecise, and easily misunderstood. The problem was that I failed to be clear as to what I meant by “customize.” Of course customers can tailor our applications to their business processes, but they should never modify our software. Customers can add to our applications and connect our applications to non-Oracle applications, but they should never go in and change the application program itself. If they do that, they will create quality and support problems and find it difficult to upgrade to new versions of our applications products.

  11

  TAKING STOCK

  April 2001

  In the weeks immediately after the “lost” third quarter and the trip to China, Ellison has been trying to get a handle on how bad the financial situation really is. But immediately after what could have been a rather tricky board meeting, he shows no sign of stress. The previous day, he flew in from a two-week spring break with his children in Tahiti on the recently renamed Ronin (its previous name—Izanami—spelled “I’m a Nazi” backward, which Ellison didn’t think quite right for its Jewish owner).1 His only grumble is that he strained his neck and sprained his toe while trying to crash tackle eighteen-year-olds during a game of beach football.

  Ellison’s next appointment is a press conference to announce a shift in the marketing strategy behind 11i. To some extent, the change is being driven by the worsening economy. However, he also wants to try to correct some impressions about the E-Business Suite that may be damaging its prospects in the market and for which he feels largely responsible. Ellison has already acknowledged that he made a mistake in sounding too hostile to systems integration—some integration with legacy software is inevitable in all but the newest businesses. It’s modifying code that’s the real crime. But there’s another misapprehension that he thinks could be holding back sales: by spending so much time and money extolling the benefits and uniqueness of 11i as both integrated and complete, Oracle may have frightened off potential customers by making them think that they have to buy and install the whole suite at once—a massive undertaking that implies getting rid of any existing software no matter how recently acquired and fit for the job.

  The new line is that the E-Business Suite can be broken down into bite-size chunks that automate discrete “business flows.” The idea is that you can decide which “flow” of business processes you want to automate first, get that done quickly and relatively cheaply, and then move on to automate others when the first has proven its worth. To dramatize the idea and to differentiate Oracle’s offering from that of its deadly rival, Siebel, Ellison has come up with the idea of installing “global CRM” within ninety days of purchase for a fixed price—any time or cost overruns
“will be on our nickel,” he says.2

  The press briefing is held in the ground floor executive dining room at 500 Oracle Parkway, the headquarters building. There’s a good turnout of about twenty journalists, including reporters from The Wall Street Journal, the Financial Times, BusinessWeek, Fortune, and Forbes. Ellison starts off by saying that the software industry’s preoccupation with offering separate applications for what are really continuous processes has made “the graceful flow of information” hard to achieve. For example, says Ellison, although Siebel provides sales automation, it doesn’t have contract automation. By contrast, the Oracle E-Business Suite automates complete business flows that reflect the way businesses operate in the real world; the alternative is to automate functional departments in separate modules. But unless data is to stay fragmented among different applications, you need to write separate programs that attempt to move the data between each application. It’s an approach that, as Ellison never tires of saying, frequently fails, meaning one group can’t work with another. Even when it doesn’t fail, it’s still very expensive.

  The example he gives of Oracle’s alternative is a flow called “campaign-to-cash”—one of ten other “fast-forward flows” that are being announced with names such as “plan-to-campaign,” “procure-to-pay,” “click-to-order,” “opportunity-to-forecast,” and “call-to-resolution.” Once a decision is made to initiate a sales promotion, the software filters prospects and leads from the customer database and sends thousands of targeted e-mails—all within minutes. When a prospect turns up at the Web store, his or her identity is known. If a sale isn’t made, the details will go to telesales, which will follow leads that didn’t transact. Once a sale is booked, every other part of the process is automated, right through to the moment when the order is fulfilled and the vendor gets the money—hence campaign-to-cash. Ellison boasts that while a Siebel implementation can take nine months to complete in just one country of operation, Oracle guarantees to “go live” globally within ninety days.3 While the cost of the software license fee to the consulting charges for big CRM implementations is typically in the ratio of 1 to 7; Oracle is betting on getting nearly 1 to 1.