Case Study: When Your Website Crashes Constantly
It was an August evening in Columbus, Ohio, and silence filled the Nissan Pathfinder. Lev Barinskiy and Seth Kravitz sat quietly, each considering the shambles of their company, InsuranceAgents.com. The pair had built the e-commerce site, which lets consumers compare quotes from insurance agents, as college students working in Barinskiy's parents' basement. Five years later, the company had revenue of $12 million and had just landed at No. 24 on the 2009 Inc. 500.
Just weeks after learning they had made the list, the two entrepreneurs launched their new technology infrastructure for the site. It had been in the works for nine months and was designed to handle growing traffic while introducing a slew of new features. But the technology failed spectacularly. The company's website had ceased to function. Business was at a standstill.
Sitting in the SUV, Kravitz and Barinskiy were thinking the same thing: Could the new technology system be saved? Should they scrap it and revert to the old one? Who should be held accountable for the disaster? And how long could InsuranceAgents.com hold on without any cash coming in?
Kravitz and Barinskiy had met at a fraternity party at Ohio State University. They became friends and soon discovered that they had complementary skills. Kravitz was a code wizard, and Barinskiy, a born salesman, was already running his own Allstate agency when he wasn't hawking snacks at OSU football games.
Even before graduation, Barinskiy had big plans. As an insurance agent, he was constantly dissatisfied by the quality of customer leads that were available. His goal was to start an online business that could more helpfully connect agents with customers. With Kravitz's tech chops, that goal seemed within reach. In 2003, Barinskiy founded a company called Cyber Technology. A year later, Kravitz joined him.
Barinskiy bankrolled the start-up with earnings from his other ventures, which still commanded much of his attention. Meanwhile, Kravitz developed a series of websites designed to make money by driving insurance shoppers to lead-generation websites. On those sites, a shopper entered his or her personal information, which was then sold to agents vying for business in that shopper's area.
But by 2007, Barinskiy and Kravitz realized that they could make more money selling customer information directly to agents, rather than acting as a middleman. Barinskiy handed off his insurance agency to his brother and made the start-up his top priority.
A new website, InsuranceLeadz.com, along with the new business model, debuted in May 2007. And the founders' instincts proved correct. Thanks to Barinskiy's experience running an agency, the site had all sorts of agent-friendly features, such as responsive customer service and flexible payment plans. Sam Goldsmith, who runs an insurance agency in Indianapolis, spends $3,000 a month on leads from Barinskiy and Kravitz's website. "I rely on their business to produce leads and feed my family, and I run my business off of their marketing," he says.
Over the next 12 months or so, annual revenue rocketed from $700,000 to nearly $12 million, and payroll expanded from six to 35 employees. The number of leads sold each month tripled, to 150,000, and the website was adding 300 new agents per month. "We definitely felt a little invincible because of how fast the company was growing," Kravitz says.
Unfortunately, the company's website began to strain beneath the weight of all this new traffic. Crashes occurred regularly. Kravitz and Barinskiy also wanted to add new features -- such as live-transfer leads, which connect agents with clients immediately by phone rather than via e-mail. But to do that, the business needed an entirely new technology infrastructure. Work on a new system began in October 2008, with an in-house team of four developers overseen by the two founders and Nate Kropp, the company's chief operating officer.
The company introduced the system -- and announced that it was changing its name to InsuranceAgents.com -- on July 1, 2009. But it immediately became clear that the system was a mess. Customer data were vanishing. Some agents were not receiving any leads at all; others were finding them in their spam folders. The tech team jumped on the case, but efforts to fix the problems only revealed new issues. The company had been accustomed to selling more than 60,000 leads a week. In that first week of July, sales dropped some 60 percent.
The new system, it turned out, was extremely complicated and the product of poor communication among the coders. Barinskiy and Kravitz discovered that each coder had worked on his own portion of code without anyone worrying about how it would interact with the rest of the site. This lack of collaboration had gone unnoticed because the system was tested in conditions that were highly controlled, rather than being subjected to an open beta test. "Everyone in upper management had been told that everything was tested and working correctly," says Kropp.
The co-founders estimated that the company would burn through its available cash, more than $1 million, in a matter of weeks. After a week of waffling, they had to act. And they had to act fast.
The Decision Kravitz and Barinskiy briefly considered reverting to the old system. But it had been buggy as well. And the tech team wanted to press forward. Two weeks after the launch, the system had been patched together and the site was selling leads again, but not nearly enough to cover expenses. The co-founders knew they had to make some serious changes. It was obvious that the tech department was a mess. But Kravitz and Barinskiy couldn't just give the techies the boot; after all, despite their shortcomings, they were the ones most familiar with the system. "We had to bring in an experienced CIO," Kravitz says. They found one, who quickly reorganized the department, creating systems to improve communication between the coders and the executive team and keeping everyone calm and focused amid the tension of the crash. (That executive is no longer with the company; Kropp has since taken over CIO duties.)
But technology wasn't the only problem. Revenue, which had been coming in at a rate of $1.2 million to $1.8 million a month, was a mere $400,000 that July. The business had relationships with two banks, but both institutions had pretty much stopped making commercial loans. The founders put out feelers to some venture capitalists who had approached them in the past, but they were no longer eager to talk.
Barinskiy and Kravitz decided to make one last-ditch effort and ask friends and family for enough money to buy them the time to repair the system's problems. Kravitz began making the rounds, but not without qualms. Asking family and friends for money "changes the dynamic of the relationship," he says. "You can't just have casual conversations with that person anymore."
Fortunately, Kravitz's sixth call was to a well-off college acquaintance. Twenty minutes later, he had a $200,000 loan. The interest rate was steep -- nearly 19 percent -- but the cash was available immediately, and the friend did not want any equity as part of the deal. The funds bought Kravitz and Barinskiy a month and change to get InsuranceAgents.com back on track.
That time was spent slogging through long days of writing and reviewing new software and scrambling to keep frustrated customers from abandoning the site, which continued to suffer from occasional slowdowns and crashes. By September, the company was breaking even; in October, InsuranceAgents.com was profitable again. When the new year rolled around, the founders were feeling confident enough to begin thinking about growth again. The company has since brought on a chief financial officer and added a number of more new features to the website.
The debacle, Kravitz says, "has turned out to be the biggest learning experience of our lives." It was also the most humbling. "There was definitely a time where we felt like we were walking on water," Kravitz says. "We thought, We must have this magic formula, and we know what we're doing. We were quite wrong about that."
The Experts Weigh In
An Eye on Cash Flow
There is good growth and bad growth. Good growth is profitable, sustainable, and capital efficient. Bad growth is growth for its own sake. Barinskiy and Kravitz should have been focusing on their cash flow and their cash balance. Start-ups often have a sales obsession rather than a profit orientation. You can run out of cash very quickly, and forecasting can help you avoid a crisis.
Founder, Management Resources Group
The Value of a Manager
They really needed somebody besides the three top executives to own and manage the technology project. The executives have so many responsibilities on their plate that they can't devote the attention needed for something this specific. There's a perception that project management creates unnecessary overhead, so there's this pressure to not do it. But when you don't do it, you basically end up with situations like the one these guys faced.
Prepare for the Next Crisis
The best time to raise money is when you don't need it. When you're growing from $700,000 to $12 million, most people are thinking, I'm growing so quickly, I don't need to raise money. But during those times, you have the best story to tell. If Kravitz and Barinskiy can get back on track to do $12 million a year, that's when they should approach a VC. They're going to have to give up some equity, but if they raise $5 million to $10 million, that will allow them to grow and get the benefit of the VC's experience and connections.
South Jordan, Utah