Softwar

by Matthew Symonds

  Robert Shaw says, “The consulting business was there first and foremost to drive product adoption and to have experts on site who knew the products and could get the stuff to work better, quicker than outside consulting firms. But what was not there from the beginning was a good interface between consulting and development to get product issues and future requirements from the field back into the development organization. The number of bugs, the number of deficiencies in the applications being uncovered by the field, was just overwhelming. It was almost catastrophic. We were always the ones that found the problems that would embarrass development. We were always on their case because we wanted relief in our budget5 because we were spending so much time building around the product to make it work in a customer site.”

  It didn’t take long for Ray Lane to lose patience with Wohl completely. “I went to Ron one day, and I said that I had finally had it and that I was going to go to Larry and recommend he take him out of the job. I said to him that I’d never stab him in the back, but I would stab in the chest. Ron didn’t say anything except, ‘Okay, fine.’ It was like, ‘I think you’re kind of crazy for doing it because I know Larry loves me.’ ” If that’s what Wohl was thinking, he was right. It wouldn’t be the last time that Lane would try to get Wohl fired. It was the beginning of a war of attrition between the two men that would last for years.

  However, it wasn’t in Lane’s nature to accept what he felt was an intolerable situation. If development under Wohl couldn’t or wouldn’t produce the weapons to make the fight with SAP more equal, Lane’s side of the company would come up with an answer of its own. The idea came from Shaw’s hard-charging consulting operation. Shaw says, “We’d made the first step of training the sales force on how to sell solutions. Then, at the next level, when we started competing for the major accounts with SAP, it became obvious that if, say, you’re going to a major oil company you’d better have somebody who knows how an oil company works and can talk about the issues—the same with a pharmaceutical company or any of the retail process industries. So Larry and Ray gave me the task of building a sales force that could compete more successfully and aggressively by vertical markets with SAP.”

  Apart from manufacturing industries, where Oracle was relatively strong, the verticalization of the sales force6 only rammed home the absence of business functionality in the markets that Oracle was now trying to target. Shaw says, “When we looked at the development plan and how long it would be before we would have a competing product with rich functionality on a global basis, it was measured in years, not months. So that’s why we came up with the strategy of, ‘Can we go out and build partnerships with other applications companies?’ where we’d preintegrate suites made up of Oracle and third-party software for verticals like the CPG [consumer packaged goods] market and the energy business.”

  Ron Wohl acknowledges that he could not give Lane and Shaw what they wanted: “When Ray and his team demanded a CPG solution, development didn’t have any good answer because we couldn’t build it in anything like the time frame that sales wanted. Sometimes, the engineering answer may not be acceptable to sales. They tried to convince Larry that Oracle’s strength was its sales force rather than its products—if we need to license the product from somebody else, let’s go ahead and do it. Robert and Ray really wanted those markets so badly.”

  Faced with apparently insuperable time-to-market problems, Oracle embraced a best-of-breed approach by identifying the vendors that had products that could fill the gaps in its own package and by coming up with a methodology for gluing them together into complete suites that would be a match for SAP. Lane says, “The idea was that we would design these templates that would convince the customer we had the right foundation. We had financials and we had HR, but it had taken us four years to build global financials. To compete against SAP, who had everything, we would work with these other companies to fill the gaps. I endorsed the strategy, and Larry endorsed it.” He also maintains that Ellison was as enthusiastic as he was to go down the best-of-breed road: “Larry bought it hook, line, and sinker.” Ellison, however, claims that he was always “a little uncomfortable about using the Oracle sales force to sell third-party products.” But he was won over both by the arguments that Shaw presented and by the need to take some of the heat off Wohl.7

  In early 1995, board member Joe Costello of Cadence Design had come out in support of Lane’s campaign to oust Wohl. Costello had complained to the board that a presentation that Wohl had given to the board of Oracle’s applications strategy was “uninspiring, uninsightful, with no clear picture of how he was going to beat the competition.” At the subsequent board meeting, he rounded on Ellison, who he believed was protecting Wohl and failed to understand how disastrous things in applications were. Ellison was shaken by the intensity of Costello’s attack: “Joe jumped up and started screaming at me, ‘Larry, you’re fucking up, you’re fucking up the company again, just like last time, you’re fucking it up!’ all because I wouldn’t fire Ron Wohl.”8 After such an outburst, Costello could not remain on the board. His parting shot was to pick SAP for his firm rather than Oracle.

  Ellison had already agreed to the idea of selling some third-party content if it meant Oracle could win deals it would otherwise lose. “Selling ten million dollars’ worth of Oracle applications along with two million dollars’ worth of i2 [a supply-chain specialist] and a custom integration project was just fine. I had no objection to selling third-party applications so long as seventy-five percent of the deal was Oracle product. This phase one of selling third-party products and custom integration worked pretty well, and we won some deals we otherwise would have lost. Phase two was much more ambitious. We’d become the king of best-of-breed by preintegrating our applications with a number of third-party products. We’d develop a best-of-breed suite for a number of different industries. Rather than selling third-party products on an exception basis to win the occasional deal, we’d transform our entire applications business into a best-of-breed suite business. And Ray’s consulting organization would run the new applications engineering and marketing teams.

  “Why did I agree to this? Well, we weren’t doing well in applications and there was tremendous tension in the company as a result. I didn’t think that this was the time to fight Ray over applications because I wasn’t sure I’d win the battle at the board level. And I didn’t have any good answers on how to solve our applications problems quickly. Ray would tell me over and over again how he could do a better job at applications development than Ron. He just wore me down. So I said, ‘Okay, go ahead and build your best-of-breed suites. Maybe you’re right. I don’t think so, but give it your best shot.’9 So Oracle started buying and partnering with lots of applications software companies. We built a consumer packaged goods [CPG] suite, an energy upstream and downstream suite, a retail suite, etc. Very quickly we were in several new applications markets. Unfortunately, prepackaged reusable integrations proved very difficult to do, so every time we sold a best-of-breed suite we had to do lots of custom integration for the customer. We lost money—lots of it—on every one we sold.”

  Jeff Henley says, “Larry sort of said, ‘This is a horse’s ass thing to do, but okay.’ It was a moment of weakness. He felt that we were getting screwed and that the field was mad at him.” It seemed sensible to Ellison to adopt a twin-track strategy. Lane and Shaw would be given their head to work on best-of-breed, while Wohl would continue to slog away, trying to make Oracle’s own apps more competitive.

  Whatever Ellison’s misgivings about allowing the establishment of engineering teams within consulting, once he had given it his blessing he had no choice other than to back it to the hilt. He says, “I wasn’t going to try to undermine Ray’s strategy—that would have been madness. While I didn’t think it would work, I didn’t know for certain it wouldn’t work, and I certainly didn’t think it was unreasonable for us to try. And maybe we’d win more deals if the sales force got behind a strategy they believed in.
”10

  To demonstrate his commitment, Ellison pledged that development would help engineer the vital interfaces between Oracle’s own products and those of its partners. Wohl was opposed to the strategy, believing it would divert resources away from making Oracle applications better and because he wasn’t sure it could be done. Consequently, Ellison moved a young engineer called Nimish Mehta into applications to work with Shaw’s industry groups and provide them with what they needed. Mehta had previously worked on getting Oracle’s core technology to work with Windows, but despite his lack of applications experience and Wohl’s profound skepticism, he was confident that he could pull it off. For Lane, Mehta was a breath of fresh air compared with the nitpicking Wohl.

  If there were uncertainties about the “experiment,” the 1995 annual report gave no hint of them. “With more than 4,000 consultants in 43 countries, Oracle Services can assemble a global team with the skills necessary to deliver the solutions customers require. . . . Oracle Vertical Markets offers comprehensive industry solutions for selected market segments. Under the Certified Application Partners (CAP) program, Oracle integrates data products from its partner companies to deliver ‘best-of-breed,’ turnkey solutions to customers in important targeted industries.” Reading it, anyone might have thought that this was something that Oracle was already successfully delivering. They would have been wrong.

  Nevertheless, the best-of-breed strategy got off to a flying start. Just as Lane and Shaw had hoped, they suddenly had a story that was capable of convincing multinational companies to buy an Oracle solution rather than turn to SAP. Many companies, especially in America, disliked the inflexibility of SAP’s software and were happy to back a U.S. competitor that promised to deliver an alternative based on gluing together the best software around. Nothing summed it up better than a landmark deal Lane negotiated with the breakfast cereal company Kellogg. In late February 1996 Oracle was committed to providing a viable “Consumer Package Goods Industry Supply Chain Management Solution” to Kellogg within three years that would “maintain a best-of-breed level of functionality, which, at minimum, meets or exceeds the functionality of its foremost competitor’s product(s).”

  From the suite, Oracle would provide only one component, the financials. The rest would come from other software firms: order management would come from IMI, logistics from Manugistics, maintenance management from TSW, manufacturing from Datalogix (later acquired by Oracle to prevent it from going bust), and warehouse management from McHugh Freeman. Oracle would serve as the system integrator and would market “the resulting solution.” All of this would be achieved through “an Oracle CPGI defined interface technique that is based on open application transaction and database standards. The complexity of this interface technique will provide for external bolt-on interfacing, which is significantly easier than interfacing to competitors’ solutions.”

  Naturally, Kellogg would also be expecting and getting massive discounts on both license and consulting fees. However, Oracle would be gaining valuable experience in CPG by working with Kellogg and would be able to use it as a reference account in selling to other companies in the same business. Similar deals were put together with other flagship customers to announce Oracle’s arrival as a serious player in a particular sector. Typical was a deal with British Petroleum to provide a complete refinery inventory management system.

  Suddenly Oracle’s revenues from applications went into overdrive. In the last three months of fiscal 1995, Oracle’s sales topped a billion dollars in a quarter for the first time, helped by a 115 percent surge in applications sales. The following year, applications sales showed a 73 percent gain in the final quarter, capping a tenfold rise in profits and share price over the previous four years. Ellison, now chairman of Oracle as well as chief executive officer, singled out Ray Lane for having contributed more to this “unprecedented success” than anyone. Such was Ellison’s confidence in Lane that he also announced his promotion to president and chief operating officer. It was probably the high-water mark in their relationship.

  In early 1997, Ellison was able to boast that Oracle was not only the number two applications company in the world but was rapidly “gaining on number one, because our applications install more quickly, adapt to change more easily, and operate more economically.” It was a nice story, but it wasn’t really true.

  The theory of Oracle’s best-of-breed strategy to take on SAP was elegant and convincing, but the execution turned out to be almost the exact opposite, partly through the absurd commitments that Oracle was prepared to make to win business, partly because what it was trying to do was extremely difficult, if not actually impossible. Systems integration is difficult enough at the best of times, and SAP implementations became a byword for the time, money, and agony that companies would have to go through to come out with a functioning ERP system at the other end. Many of those that bought into the hype over ERP in the 1990s are still trying to figure out the return on their investment, while a good number would rather not even ask the question.

  Although Oracle’s applications were designed to be more flexible than SAP’s, much of the flexibility was lost by trying to integrate with the applications of other vendors. But the real problem was Oracle’s ambition. In most cases, best of breed can be made to work with lots of customization. But even though the same group of vendors’ applications may be used in combination on many different occasions, each customer ends up with a computer system that is essentially unique to itself.

  What Lane, Shaw, and another ex-Booz consultant named Charlie Schneider set out to do through their integration methodology was to make the whole process repeatable. While Oracle might not make much money on a contract like the one with Kellogg, out of it would come a preintegrated suite that could be sold profitably to other CPG companies at a competitive price. In other words, repeatability equaled leverage. That way Oracle could convince itself—and, more important, Lane could convince Ellison—that it was still a software firm and not just a distributor/integrator.

  Ellison remembers, “They had all these wonderful names for the new integration technologies that they were developing: the message backbone, integration glue, and all this other stuff. I tried to get a detailed technical explanation as to how it all worked, but nobody was able to explain it to me. I just couldn’t understand what they were saying. One of two possibilities here: I’m too stupid to understand, or they don’t understand it either. Even though they couldn’t describe how this stuff worked, they just knew it would solve all their product integration problems. Charlie Schneider is a capable guy, but he had never developed software before. I don’t think he ever understood the unbelievable difficulty of building this magical integration technology. Just because they wanted it and needed it, they believed they could build it. It was like, wouldn’t it be nice to have flying cars? All we have to do is interface the wings to this Oldsmobile and it will be phenomenal. We just need really good glue. Brilliant.”

  One of the things that made it so difficult was having to use the proprietary tools of each application partner. Another was getting all the pieces to run on the Oracle database platform. Sohaib Abbasi says, “There were components that had never been built for our database and had to be adapted for it. From a technology point of view, we could never get those applications to fit in with any of our products.” But the killer was the upgrade cycles of the third-party vendors.

  Ellison says, “IMI would finish a product in June, but Manugistics wouldn’t have their new version done until November. The customers were allowed to say which products they wanted, which versions of which product and in what order they wanted them implemented. The customers had so much flexibility that there was no repeatability in the systems integration at all. Each implementation turned out to be unique. The Kellogg contract was a key deal Ray was able to snatch away from SAP at the last minute by making some astoundingly open-ended commitments. The Kellogg contract basically said, ‘Oracle will make its applications software d
o whatever Kellogg needs to run its business, and make the IMI software and all our other partners’ software do whatever Kellogg needs too.’ Needless to say, Kellogg’s software enhancement list was long and our losses were large on that one.”

  Interestingly, even Robert Shaw, the main evangelist of best-of-breed, doesn’t fundamentally disagree with Ellison about the practical difficulties: “At the application layer . . . it rapidly became an engineering nightmare.” Ray Lane, however, is unrepentant. He accepts that repeatability was a goal that was never reached and that Oracle went through a lot of pain, especially in CPG, but he doesn’t see it as the failure that Ellison does today. “It was the right decision. The only alternative was to let it all go to SAP. Take CPG: we generated a lot of CPG clients, not without a lot of pain, but there are maybe thirty-five to fifty clients in CPG alone, who are still Oracle today, including Kellogg. It generated hundreds of millions of dollars of business.”

  Lane argues that one of the reasons for Ellison’s jaundiced view of best of breed is that he still doesn’t understand the applications business. “It’s not true what Larry says about Kellogg. No, it wasn’t the greatest contract ever, but I think we went into it with our eyes open. Larry’s not an applications guy. With the database, Larry’s used to building a product where he can ignore customers. We were building a suite of products, connecting them together and trying to figure out what Kellogg wants it to be at the same time. SAP had hundreds of these projects. Look at the stories of SAP trying to install their suite at Compaq or P&G—a billion dollars, two billion dollars. They lost incredible amounts of money doing this. They had the same problems as we had except that our job was a little tougher because of integrating third-party apps.11 But this is something Larry is measuring relative to the database business, which we totally owned and in which you just sell the software and don’t change it.