(FORTUNE Small Business) – I started thinking about hiring a second-in-command when I realized I was the limiting factor in the growth of my Reston, Va., company, Brittenford Systems, which develops software. As our revenues grew from $2.6 million in 2004 to $4 million last year, I had less time to do the things I'm really good at--strategy, client relationships, rainmaking. I needed to delegate what I'm less good at--operations and project management. I needed a strong No. 2.

This is where it gets embarrassing. I went through several mishires before I found the right fit. The first was someone I met at a business club who was between jobs. Within 60 days I knew he wasn't going to work. He had last managed a staff of 600; now it was six. He was far too senior and out of touch with our needs in terms of technology and clients. We parted on good terms after five months.

My second No. 2 was a prospective client I met on a sales call. He had been a product manager at a large software company--relevant experience, I thought. But in his last job, he never interacted with his software's users. We custom-build software for clients, and work closely with them. It is very much a people job, and that wasn't his forte.

After hiring two very senior candidates, I figured I needed more of a platoon leader. Following a modest search, I promoted an employee I knew to be incredibly intelligent. I also knew his style was abrasive, but I figured you have to break eggs to make an omelet. It only took a few weeks before we both knew it wasn't working. He had been an independent contractor for many years, and after we parted, he returned to working for himself.

The trauma of those bad choices had left my team demoralized. So I took a big step back and, with advice from other CEOs and experts, revamped our hiring process. The first step sounds absurdly basic, but I'll guarantee the vast majority of companies are not doing it: clearly defining the position. Not by writing a job description, but by gathering the staff to create a detailed list of the top six or eight things the manager would be accountable for in the coming year. We also talked about the skills, education, and behavioral profile the ideal candidate would have. For the last, we used a personality-assessment tool from PI Worldwide in Wellesley, Mass. Then we posted the ad on Monster.com and in a local industry newsletter, and generally put out the word.

About 250 applicants responded. We could eliminate 80% based on their résumés or the results of PI Worldwide's online behavioral test, which we asked them to take. Then we designed a set of ten phone-screening questions--for example, "Can you describe your ideal next job?"--that could be conducted by an administrative person. That ruled out another 15% of the candidates. I then interviewed the remaining group by phone and in person, using methods recommended by Bradford Smart, author of Topgrading, a book on best practices for HR.

Next came the newly restructured team interview. Rather than each team member asking the same generic questions and getting the same good answers, we assigned a different topic to each. The candidates, I thought, should do 70% of the talking. After interviewing the candidate one-on-one, each team member e-mailed me with a yes-no vote based on his criteria only.

Finally, I asked the two remaining candidates to write a one-page strategic plan for the company. That may sound like overkill, but it surfaced misplaced expectations and evidenced their original thinking. We chose our favorite, who accepted the offer. Within 90 days, he has increased our pipeline, ramped up recruiting, and brought a new sense of optimism to the team. The search lasted four months, but that time seemed insignificant compared with what it cost us to make the wrong hires.

Owner to Owner is written by small-business owners and relates to a lesson learned. Mahoney can be reached at brittenford.com. Please send feedback or column ideas to fsb_mail@timeinc.com.