No Better Time

by Molly Knight Raskin

  To minimize the chance of failures, Lewin and Leighton immediately put one of the company’s first rules in place: the Akamai office could never be empty. In order to provide customer service twenty-four hours a day, seven days a week, Akamai needed someone in the building at all times to troubleshoot. For at least the first year of business, this meant either Leighton or Lewin had to be there, too. This work commitment wasn’t always easy on their spouses, both of whom were carrying the bulk of the domestic and parenting duties. Sometimes Anne Lewin or Bonnie Berger would arrive at the office in search of Danny or Tom, hoping to get their undivided attention on family matters. In February 1999, for example, a very pregnant Berger showed up in Akamai’s lobby in search of Leighton, who had been too caught up to return any calls that day. When someone who had volunteered to locate Leighton asked her what the message was, Berger said: “I’m in labor!” The couple just made it to the hospital in time, welcoming their daughter, Rachel, into the family.

  The potential perils of Lewin or Leighton being away from the office for any more than a day or few hours, however, were significant. If they weren’t around, the programmers who spent hours translating Akamai’s algorithms into code, which they often pushed out of the system well after midnight, could make an error with far-reaching consequences. “The programmers were fast and talented, but they weren’t crossing all their t’s and dotting their i’s,” said Leighton. “There were always bugs in the code, which was the price we paid for the [programmers’] speed. Either Danny or I had to be here to check [the code], or it would all blow up, and we’d have to go in and fix it.” At such an early stage, they couldn’t afford for things to go wrong. They’d spent the summer and start of fall of 1999 securing a non-binding commitment from big-name companies to try FreeFlow when it launched. The deal they offered was no cost to the companies—by signing on they agreed to test the service on a small portion of their content, ideally an image as tiny as a few pixels and embedded deeply on their Web site. From its headquarters in Cambridge, Akamai would then simulate peak loads of traffic to the select image long enough to prove the value of FreeFlow.

  With the help of venture capital contacts and media mogul Gil Friesen, Lewin, Leighton, Seelig, and Kaplan, who worked out of California, were able to convince an impressive roster of ten beta testers to sign on. They included CNN, Disney, Yahoo, Geocities, MGM, Paramount, Warner Brothers, Universal, and Playboy. None of them were testing any real content with Akamai; as stated above, most started with one trial image hidden deep within their Web sites. But they were still some of the biggest names on the Web, and Akamai needed to secure a commitment from them beyond the trial stage. As COO, Paul Sagan swore everyone at Akamai to secrecy—cautioning them not to reveal the names of those who were still deciding whether they trusted the technology and the company. In addition, Akamai needed to maximize its public relations buzz. They had only one shot to announce an inaugural commercial launch. In an e-mail to the company on March 2, 1999, Sagan wrote, “If in doubt about what to tell people outside of Akamai, say less, not more. And refer them to me. Keeping a veil of secrecy about ourselves remains crucial. We don’t want to tip off our competitors too early.”

  Akamai had no room for error. If the head of engineering for any one of these FreeFlow customers happened to log onto their Web site at 3:00 a.m. and notice the Akamaized content frozen or failing to load, they could easily cut ties.

  As it happened, the first major glitch in Akamai’s fledgling network had nothing to do with code. Sometime in January of 1999, the servers started to fail. Not just in one place but in many data centers, and no one at Akamai knew why. They considered outsourcing the repairs to Exodus, but this would take time. Instead, Sagan said, they decided to go at it alone: “We had to do it ourselves; we had to basically try to repair the network before the whole thing got disconnected.” One of their first stops was an Exodus location at Boeing Field in Seattle. It was close to midnight when Lewin and Sagan arrived at the colossal data center, a windowless, bunker-like building filled with racks of servers stretching from one end to the other. Lying on the cold floor surrounded by a few tools and myriad electronic parts, they pulled out the servers to diagnose the problem. It didn’t take long. Almost as soon as they pulled the devices apart, they began finding loose screws. “They pretty much all had the same sloppiness,” said Sagan. “I don’t know if Danny even knew what he was doing, but if he didn’t, he pretended to.” Lewin and Sagan worked through the night to painstakingly disassemble each server, tighten its insides, and reinstall it into the racks. They later learned that, in a rush to fill Akamai’s order for four hundred servers, the small California-based company True Solutions had hired inexperienced high school interns to put them together. The company shipped off the shoddily assembled servers, and once they were placed in the data centers, which were kept at low temperatures to protect the equipment, the loose screws had wiggled free, separating the components and causing the machines to crash.

  The job of fixing the ailing servers was tedious and time-consuming, but it was a necessity. By the end of January, the servers were securely in place and Akamai was ready for the official debut of FreeFlow.

  As a final test of the network’s strength before launching, Akamai’s engineers ran a load of data approximately six times the size of the heaviest traffic the largest news Web sites were struggling with. It barely hiccupped.

  Image might have been the last thing on the minds of everyone at Akamai as FreeFlow took off. They were academics, and their work wasn’t at all well-suited to the slick, fast-talking arena of publicity. But in the dot-com craze, generating buzz was a necessary element of success. Buzz made business moguls out of college students and helped build the astronomical market valuations of companies with little more than a cool-sounding business plan on paper. Akamai needed to create the impression that everyone who was anyone on the Internet was using its service. For this, Akamai turned to Marco Greenberg. In addition to his work helping Lewin, Leighton, and Seelig get Akamai off the ground, Greenberg had spent over a year building his own business, NYPR (New York Public Relations). He’d already garnered some big clients in the worlds of media, technology, and government, and was eager to officially take on Akamai. Before he was formally hired, Greenberg was interviewed by advisor Todd Dagres of Battery, who was drawing as much talent as he could into the company. Greenberg didn’t have to prepare much; he knew Akamai inside and out and possessed stellar experience, including a few years as a manager in the corporate practice of global PR giant Burson–Marsteller, running big accounts including the Government of Israel. To Dagres, it was clear that Greenberg was much more than Lewin’s friend—he was the best man for the job. Greenberg would remain in New York to manage his firm but travel to and from Cambridge once or twice a week for meetings and press duties at Akamai.

  For Greenberg, it was a thrill to publicize Akamai. At the time, the media were eagerly pouncing on Internet startups with compelling background stories, and Greenberg knew just how to spin Akamai’s. “The story was that these MIT scientists had found a way to make the World Wide Wait a thing of the past,” explained Greenberg. “It was the right place and the right time.” Greenberg said he never even tried to understand the nuances of the technology like an employee of Akamai or a techie; that wasn’t his job. “I understood it just enough—I knew they were setting up these servers in various locations and that they would provide a richer experience for Internet users. My goal was to express it in layman’s terms so that your grandmother could understand it. And my grandmother could.”

  In the first official press release, issued by Greenberg on January 14, 1999, FreeFlow, which Greenberg had named, was touted as “the world’s largest fault-tolerant network for distributing Web content.”{35} Greenberg distributed it to every possible outlet, and it began getting traction almost immediately. The media strategy, he said, was to go for the “big stuff” right away, well before plans were in place for an Initial Public O
ffering (IPO), and to emphasize the company’s MIT connection.

  Because of his age, charisma, and background, Lewin was the most obvious candidate for good press, but he wasn’t always keen to engage in publicity. Some say he shied away from the spotlight, but according to Greenberg, he was just busy and focused on the day-to-day grind of running the company. Leighton felt likewise; he was happy to speak with the press, but preferred not to. Because of this, Greenberg often put Sagan out front, particularly when the press involved television. Because of his career in journalism, Sagan had inbuilt media savvy. He understood the impact of good press, and he also understood how to generate it. Most importantly, he was skilled at translating what Akamai did into something that sounded simple and impressive to a mainstream audience. When a sound bite was needed, he could serve one up. Even so, the media often got the facts muddled. On January 15, Sagan appeared on CNN for an interview with senior correspondent Steve Young, a veteran technology reporter. Sagan spent two full minutes explaining Akamai, after which Young turned to him with a perplexed look and said: “Paul, I’m not sure I fully get the concept. Are you selling equipment? A service? Both?” Patiently, Sagan started over.

  The fact was, Akamai hadn’t sold anything yet. The FreeFlow trial was going smoothly, but the ten trial customers weren’t ready to trust a small company out of MIT with the bulk of their Web content. Akamai’s first order of business, beyond the publicity blitz, was the hiring of a sales force. By early February, the company had a small team in place, led by John Sconyers and Earl Galleher, who both came from DIGEX, a Maryland-based Web hosting giant. Both Sconyers and Galleher were first introduced to Lewin, Leighton, and Seelig in the fall of 1998. At DIGEX, they were living with the problems of flash crowds and slow Web traffic on a daily basis, losing droves of customers frustrated by the company’s inability to keep their sites up when traffic spiked. Galleher, who had been president of the Web Site division of DIGEX, said he didn’t understand the computer science behind Akamai, but he did understand their business model and believed in it from the first time Lewin presented it to him on a whiteboard at MIT. “Danny got into this zone when he was just laying it out on the board and you could just see it all come to life,” Galleher recalled. “You just knew it was going to work how he described it.” At the time, Galleher offered his support and later contributed $25,000 to Akamai’s first round of financing. “Danny knew that, technologically, I’m dumb as a brick—he could do his hashing algorithm talk for two days, and I wouldn’t be any further along,” remarked Galleher. “But it didn’t take a rocket scientist to understand the pain corporations felt when they suffered network outages, and I knew it well.”

  At age thirty-nine, Galleher became Akamai’s vice president of sales. One of Galleher’s first tasks was to help Akamai set a pricing plan. At the time, the content delivery network (CDN) business model was so new that there was no industry standard. Akamai had a rough plan in place, one that would charge customers around $800 per megabit of data delivered per second. Galleher insisted the price was far too low. “They were looking at it mathematically,” said Galleher. “But I was looking at it in terms of what the market would bear.” Galleher suggested more than doubling the price to what he believed was the magic number: $1,995 per megabit per second. “Everyone scoffed at me,” recalled Galleher. “And I told them, ‘You know what? I’ll fuckin’ prove to you that I can sell it at this price.” And he did. When the company finally landed a full-price contract with Discovery Channel, it came in at $1,995. From that moment on, Akamai had in place a straightforward, service-based revenue plan, charging clients an even $2,000 per megabit per second per month. The pricing was based on each client’s peak usage, which meant if it reached 5 megabits per second at any time during one month, Akamai would receive $10,000 for that month.

  Galleher was a colorful character, known for his candid but sometimes brash demeanor. Lewin liked his passion, and the fact that he could always be counted on for a dose of humor. One story in what became the “Legend of Galleher” involved a debate within Akamai over whether or not to work with Playboy Corp., based out of Chicago, as a customer. “We’d done some soul searching, and at first many people on the board [of directors] were against the idea of it because to them it meant we were carrying porn,” said Dagres, who was involved in early talks with the company. “But we could also justify it with the fact that the U.S. Postal Service was Playboy’s biggest distributor.” Dagres recalled one meeting in particular when he asked the representatives from Playboy about their Internet service needs, and one of them mentioned their worries about site security. “Are you concerned someone might hack into [it] and put porn on it?” Dagres joked. The line fell flat. “I swear you could hear a pin drop,” he said. It took a particularly rough quarter for sales, however, to iron out any internal conflict over Playboy. Dagres related that, during a board meeting focused on how Akamai would make its sales mark for the period, Galleher commanded the room. “I have a solution,” he said. “Porn.” Galleher said this in jest, but there was some truth in his words. Pornography companies were early adapters of Internet technology and arguably the first e-commerce businesses. And, because of market demand, they were willing to pay for top-tier content delivery. With this in mind, executives at Akamai agreed. The company agreed to limit the customer base in that sector to sites they deemed acceptable, which included Playboy. Still, the decision became fodder for company-wide jokes like the name “Naughty NOCC,” referring to the Akamai’s “Network Operations Control Center,” where someone at the company was delegated to monitor the Playboy Web site.

  John Sconyers, a graduate of MIT’s Sloan School, had spent a few years managing all of the New England accounts for DIGEX. At age twenty-nine, he became Akamai’s twenty-third full-time employee, joining the company a month before Galleher as senior account manager. Sconyers knew that, to gain momentum, Akamai would have to sign on at least two-dozen big names as customers—and do so quickly. But selling a service that few people fully understood wasn’t going to be easy. Competitors like Sandpiper and Inktomi both offered caching and mirroring services that were intelligent enough to pose a threat and already had market traction. But the greatest threat, both Sconyers and Galleher knew, didn’t come from the competition. It came from what they called “do-it-yourself” (DIY). This meant the companies so desperate for a way to manage escalating Web traffic that they were building their own solutions with huge amounts of money, massive infrastructure and highly skilled engineers. The idea of swooping in on a company like Yahoo, for example, and suggesting that Akamai’s solution—one rooted in obscure theoretical math—was better than its own, risked sounding more like bravado than brilliance. “At that time, what we were doing was not obvious, and most people thought we were totally nuts,” Sagan said. “It wasn’t so much that we had skeptics. It was that we didn’t really have believers.”

  To win over a critical mass of customers, Akamai’s tiny sales force set out with almost evangelical zeal. They understood the virtuous circle, or network effect, of their business model—the more customers they won over, the higher performing Akamai’s service would be because it would be handling more data, further proving its excellence. Galleher and Sconyers set out almost immediately, flying all over the country to meet with content providers, usually with a few recent MIT graduates in tow—newly minted engineers wearing baseball hats and backpacks. “I was out on the road and taking these brilliant kids with me,” Sconyers recalled. “It was certainly the first business trip they’d ever been on, and there we were meeting with CTOs and other top executives of these media giants, who set the vision and controlled budgets of tens of millions. It wasn’t as if we were going out to meet with a pizza joint or a plumbing shop. It was Disney, it was Yahoo, it was CNN, and it was Universal. And all of us were jumping on planes every week, flying around to get this all started. It was just thrilling.”

  When the meetings didn’t go well, or there was an intractable technology
expert to convert, Lewin, or one of Akamai’s other executives, would deliver the pitch. It quickly became apparent that, when it came to sales, Lewin was the company’s most powerful weapon, capable of turning skeptics into true believers. “A lot of meetings would begin with some decision maker saying, ‘Thanks for coming to Seattle, but I’ve only got twenty minutes. My boss told me I had to take this meeting, so sit down and tell me what you have to say,’” Sconyers remembered. “Two hours later, Danny would still be at the whiteboard in full throttle, with a room full of technology staff in rapt attention.” Galleher, who traveled thousands of miles with Lewin to make sales calls, recalled: “Danny was so focused on getting everyone in the room to experience this euphoric passion, one that made them believe ‘I don’t know exactly what this is, but I need to have this, and I need to have it now.’ Once he got the customer to this frothy pitch level of excitement, he’d basically leave the room, and I’d close the deal.”

  Galleher added, “The fact was, these customers had a need, and they weren’t aware that it was possible to solve it our way because nothing [else] like Akamai existed. The persona of Danny rapidly created a market perception that we were intensely smart and what we were doing was highly relevant.”

  While the sales team traversed the country in the early months of 1999, Akamai’s staff in Cambridge, which was growing steadily, continued to build the network’s infrastructure. At the time, they had no more than a few hundred servers, all but one set in the US. The more customers they signed, the more points of service they would need. There was also the threat posed by the competition. To beat them, they needed to be the first to install their servers in data centers. “That way, if anyone came on and said they wanted space to do what Akamai was doing, we’d already be there,” reasoned Sagan. “We saw it as a land grab for beachfront property.”