In October 2025, our agency was a small team. Just a few agents we'd recruited from online communities. By early 2026, we had over 70 agents. This is exactly how that happened, including the parts that didn't go well.
The first thing to understand is that recruiting in insurance is a volume game with a conversion problem. There are thousands of people who are curious about selling insurance. They've seen the TikToks, they've heard you can make six figures from home. The hard part isn't getting applicants. It's getting people who will actually show up, get licensed, and write business.
We started with Agent Surge, which is our recruiting automation tool. It posts to job boards, captures applicants, and drops them into a pipeline. The key insight was that speed mattered more than anything. When someone applies to an insurance opportunity, they've probably applied to five others at the same time. The agency that gets them on a call first wins.
So we set up Agent Surge to post on Indeed, ZipRecruiter, and a few insurance-specific job boards. When an application came in, the system would call them within two minutes. Not a text. Not an email. An actual call. We had built the AI calling functionality for lead follow-up, and we realized the same technology worked for recruiting.
The script was straightforward: introduce the agency, ask about their insurance background, explain the opportunity, and book a Zoom call with our team. If they were already licensed, they went into the fast track. If they weren't, we explained the licensing process and booked a follow-up.
October was our proof of concept. We brought on 25 applicants that month. About half were already licensed, half needed to get licensed. After vetting and the ones who ghosted, we ended October around 12 after the first wave came through. November hit 40. By January we'd crossed 70.
Those numbers include churn. We probably onboarded over 100 people in that span. Some never finished licensing. Some finished licensing and never wrote a policy. Some wrote a few policies and decided insurance wasn't for them. That's normal. The 70+ are people who were actively producing.
The onboarding process was the real breakthrough. Before Closd, onboarding a new agent took about three weeks. They'd need to get contracted with carriers, learn the products, set up their tools, shadow some calls, and eventually start dialing on their own. Three weeks before they even had a chance to make money.
We built an onboarding checklist inside Closd that cut that to six days. Day one: account setup, carrier contracting initiated, assigned a mentor. Day two: product training modules (we recorded eight short videos covering the main products). Day three: PitchLab practice sessions. Day four: live shadow calls. Day five: supervised solo calls. Day six: fully independent with live leads.
The secret wasn't that we compressed the timeline by cutting corners. We compressed it by removing the dead time. In a traditional onboarding, an agent might do 30 minutes of training, then sit around for two days waiting for their carrier appointments to come through. We pre-contracted with our main carriers so appointments processed in 24 to 48 hours, and we filled the gaps with PitchLab practice so agents were never just waiting.
The leaderboard was the other game-changer. Once we had more than 10 agents, we put a leaderboard in the dashboard. Weekly production numbers, visible to everyone. Top three agents got called out in our group meetings. We ran monthly contests with small cash bonuses.
People respond to competition in this industry. Having 50 other agents see your name at the top (or the bottom) of a leaderboard is incredibly motivating. Our per-agent production actually went up as we added more agents, which is the opposite of what most agencies experience. Usually, quality drops as you scale. For us, the competitive environment pushed everyone to produce more.
Now, the mistakes. We made plenty.
The biggest one was not vetting hard enough in November. We were so excited by the volume that we were basically onboarding anyone with a pulse and an insurance license. That led to a few bad actors: one agent who was misrepresenting policies to clients, another who was submitting falsified applications. We caught both within two weeks, but it was a wake-up call. After that, we added a background check step and a three-reference requirement to the pipeline.
The second mistake was geographic concentration. We had too many agents in Arizona and not enough geographic spread. When you have 15 agents all working the same leads in the Phoenix metro area, they start stepping on each other. We fixed this by being intentional about where we recruited and assigning lead territories.
The third thing that didn't work: hiring unlicensed people and expecting them to get licensed on their own. The licensing pass rate for people who self-study is terrible. Around 30% in our experience. We eventually built a study group with scheduled practice exams, which pushed our licensing rate above 70%. If someone wasn't willing to commit to the study schedule, we knew they probably weren't going to make it.
Lessons we'd give to any agency owner trying to scale:
Speed of response is everything. In both lead follow-up and recruiting, the first person to make contact wins. Automate this or you'll lose to someone who does.
Onboarding speed directly correlates with retention. The longer it takes a new agent to make their first sale, the more likely they are to quit. Compress that timeline by any means necessary.
Build competition into your culture from day one. Leaderboards, contests, public recognition. Agents who feel like they're part of something competitive and transparent produce more than agents who feel like they're working in a vacuum.
And vet your people. Growth is exciting, but one bad agent can create compliance headaches that take months to clean up. It's worth the extra day in the pipeline to make sure you're bringing on people who'll represent your agency well.
Scaling to 70+ agents sounds like a headline. The reality was a lot of late nights, a lot of Zoom calls, and a lot of learning what doesn't work. But the playbook is repeatable, and Closd made it possible to do at a speed that would've been unthinkable with our old stack.