In 2015, FreshBooks founder Mike McDerment celebrated when a new competitor called BillSpring released a product better than his own. When, nine months later, BillSpring showed substantial sales, McDerment was ecstatic.

That's because McDerment was also the founder of BillSpring. He launched the business as a way to directly test a new from-the-ground-up successor for the invoice and accounting software sold by his 300-employee company. At a time when experiments are in vogue as a way to drive innovation, the creation of BillSpring stands as an ambitious and ingenious example.

FreshBooks' origins were prosaic. In 2003, McDerment was running a four-person design firm in Toronto, when he accidentally saved over an invoice. That small mistake inspired him to develop software that let clients view invoices online. It was meant to be extremely simple, perfect for tiny firms like his. When clients responded favorably, he tried putting it on the market, signing up 10 customers who paid $9.95 a month. Encouraged, he worked on the business in his parents' basement for three and a half years, emerging with thousands of clients and a staff of six.

By 2014, FreshBooks, with more than 10 million users, was the market-leading cloud accounting software for self-employed professionals. But even as sales climbed, "I looked into that crystal ball of three to five years out, and I thought, 'Will we be leading then as well?'" McDerment says. "I did not think we could get from here to there with the constraints our prior platform placed on us."

Simply put, the product had aged. "We wanted to simplify it and make it more modern, because it reflected the aesthetics of 10 years ago," McDerment says. In addition, the platform--built largely without the help of professional programmers--made improvements and fixes difficult to execute, a major problem in an era of swift technological change.

Just rewriting the code was a nonstarter. "That always takes longer than you think and costs more," McDerment says. "There's also the sophomore jinx: There is no guarantee that the new product will be better."

McDerment's goal was to build an entirely new product that, if it had problems or proved unpopular, would not tarnish the FreshBooks brand. "Think about the risks a company takes to build a platform," he says. "Millions of people use our site every month. If we launched something that was really broken, that comes back to bite us."

McDerment also hoped to change the way his development team worked, ramping up speed and making programmers comfortable with risk. Finally, he wanted to operate in secret so as not to tip off competitors.

Reinventing everything

McDerment debated different approaches to developing and testing the product with his executive team. But the answer snapped into focus one weekend while he was doing chores. "I came in on Monday and said, 'What if we launched our own competitor?'"

The timing for an ambitious new development project was propitious. In 2014, FreshBooks had raised $30 million in its first round of institutional funding. Early the next year, McDerment incorporated BillSpring in Delaware: an entirely new business with a new brand, a new URL, a new logo, and a new end-user agreement written by a new lawyer. To encourage innovation and risk-taking, he made the product--which was still largely in prototype form--available for free.

The team used keyword advertising on Google to target mostly first-time customers. McDerment still needed most of the traffic going to FreshBooks, which was paying the bills, so he marketed the original company more heavily. But he also wanted to see how many people opted to try BillSpring, and then use their feedback to improve it. 

As the team worked, it experimented with new development, research, design, and engineering processes, versions of which were incorporated into FreshBooks' standard operating procedures. It also grew the development staff by about 50 percent. "We became like a giant petri dish," McDerment says. "Now we ship software five or 10 times faster than we used to."

After about nine months, BillSpring switched to a paid model, allowing the team to assess conversion rates. By early 2016, they had enough data to be confident the new product would be not just accepted but also embraced by FreshBooks' users. "One person called us to cancel FreshBooks to tell us they were going to this new company," McDerment says. "That was a good day."

Soon after, McDerment closed down BillSpring and routed all customers to a new FreshBooks website, running the reinvented product, which also debuted for the FreshBooks base. The company had spent about $7 million on the venture. McDerment won't give sales figures for the new product, but says the company has been pleased with its financial performance and satisfied with the experiment.

"At Facebook, they have this idea of move fast and break things," he says. "We didn't have the luxury to do that under the FreshBooks banner. Under the banner of the disposable brand, we did."