Fearing Obsolescence, a Company Charts Its Reinvention; When a Web marketing and communications company discovered its hardware system was out-of-date, it set out on the giant — and ultimately successful — task of remaking the company

August 14, 2013

Fearing Obsolescence, a Company Charts Its Reinvention

By STACY PERMAN

It may be the most terrifying moment a business owner can face: the realization that what you have been doing successfully, possibly for years, no longer works. For Clint Smith, co-founder and chief executive of Emma, a Web-based marketing and communications company, that moment came three years ago when his team attended the annual South by Southwest Music and Media conference in Austin, Tex. Started in 2002 and based in Nashville, Emma had grown quickly. By 2010, it had 90 full-time employees and 30,000 clients. It had recently passed $10 million in sales, but an awareness had begun to set in that its hardware system — built before the cloud even existed — was showing signs of strain. Capacity was running low and programmers had to navigate several layers of the system to update existing features or introduce new ones. These concerns crystallized at the conference when the Emma executives listened to Google employees discuss their plans for Gmail.Hearing how innovative and agile Google’s software was and how many programmers the company could deploy, the Emma team began to question everything about the company’s ability to compete. “It became fairly obvious that we had to do something,” Mr. Smith said. But how were they going to rethink and rebuild the company while continuing to serve existing customers?

The first step, Mr. Smith said, was to “declare war on our technical-to-nontechnical staffing ratio.” Within a year Emma had ramped up its roster of developers from 20 to 41 — or nearly half of its full-time employees. “If we were to make this a world-class software company,” Mr. Smith said, “we needed to bolster our resources and look and staff up more like a software company.”

Next, the company considered a plan, outlined by consultants already on retainer, that was intended to extend the life of Emma’s existing system. But with a price tag of $250,000, the plan offered little more than a costly Band-Aid. As Mr. Smith and his team considered their options, they found themselves returning to the same question: What would Emma do if it were starting fresh?

Re-engineering both their platform and their products would take time, an anticipated 12 months, and it would cost money. They did not even try to put a price on the project. “It felt more like a commitment to a new way of operating than a one-time project,” Mr. Smith said.

They also knew that any transition process would bring glitches and delays and the real possibility that clients would flee. For an extended period, the company would not be able to adjust its system or processes; it would be stuck with all of the worst aspects of the old platform.

But if Emma did not rework its system, the company would lose long-term competitiveness, leading to a loss not only of existing clients but of future ones as well. In the end, Mr. Smith determined it would be riskier not to start over, noting, “Big fear beats little fear every time.”

When Mr. Smith looked for other small companies that had rebuilt their technology infrastructure, he found few templates. “I will say none of them had taken on the whole thing and certainly not in such a relatively short period of time,” he said.

One that came close is Stratose, a company offering health care cost-containment services in Atlanta. Ten years ago, it completely rewrote its own system, which its president, Tina Ellex, called “the Big Bang theory,” a tough process that required huge blocks of time and a dedicated team. Ideally, Ms. Ellex said, she would have preferred a modular approach. “Anybody would,” she said. “I’ve not seen a lot of companies do a complete rewrite.”

At Emma, the project took 18 months, six months longer than anticipated. And it cost $4.5 million. Preliminary research and design began in the summer of 2010. The company relied primarily on its own engineers and developers, hiring a small outside team to build one set of e-mail tools.

The principal operating guideline was to design a flexible system that would remove the need to do anything like this again. “We didn’t know what the marketplace would look like in five years,” Mr. Smith said. “Also, we didn’t know how databases would evolve in five years either. We couldn’t design for the future, but we could design something that could adapt to what the future will bring.”

Emma chose an application program interface design and a service-oriented architecture, which enabled it to produce and execute updates and new features. The design allows internal services to connect with each other as well as with outside services, which it had little ability to do previously. Emma also re-engineered its database structure to take advantage of cloud-based storage like Amazon’s S3 and new technologies like Redis and CouchDB.

Throughout the process, the company tried to keep disruption affecting its customers to a minimum. “In the beginning our clients didn’t know we were refactoring so many things,” Mr. Smith said. “Increasingly, as we got further into the process we also increased the transparency around it. We shared the reasons for what we were doing.”

Mr. Smith conceded that bumps along the way contributed to the loss of some clients. “Our churn rates did go up, especially at the end of the project,” he said, but the company retained most of its top customers. That, he said, was a result of redoubling customer-service efforts, including the creation of a concierge team of relationship managers that held clients’ hands throughout the transition.

Released in January 2011, the end result was a new suite of agile products and services. Emma’s new campaign editing software, for example, lets customers drag and drop content and images into a host of marketing materials like newsletters, updates and invitations. One client, Jason Carey, vice president for marketing and communications at the Brooklyn Public Library, said he was impressed with the ability to split-test a new survey that the library had begun sending to users, trying out changes to its design.

Another client, Lisa Nelson, Web site editor at the Dana-Farber Cancer Institute in Boston, found the transition process “almost seamless,” describing Emma as being clear about the expected improvements and responsive during glitches. She recalled trying to send out the institute’s e-newsletter to 375,000 of its supporters and donors, only to have problems logging on. “I panicked,” she said. After several futile efforts, she called Emma’s customer service team. “This newsletter is the biggest we send out in the year and it had to go out that day,” she said. “Within minutes they resolved the issue.”

Late in the process, Mr. Smith said, he realized they had underestimated the amount of testing some of the new features required. “When we launched, some applications were buggy,” he said. “We found there was a lot of follow-up and cleanup work to do.”

He also concluded that they could have broken the project into more manageable chunks. “We realized we were terrible at project management and scoping out time frames,” he said. “This doesn’t bode well when you take on the biggest project of your life.” In retrospect, he said, they should have “launched one service, tested it, seen its execution, and learned from it before going on to the next piece of the process.”

But they did get better as the project progressed. Ultimately, during the 18-month process, the company added 250 to 350 new clients a month. Emma now has 47,000 customers and is on track to bring in roughly $16 million in revenue this year.