Growing a support team from 14 agents to 90 while scaling a customer base from 60,000 to one million subscribers is a success story. But it's also a story full of painful months, hard lessons, and a few experiments that went straight in the bin.

I spent five years as Head of Customer Support at Cleo, and during that time I learned more about scaling support operations than I ever expected to, and the lessons weren't always comfortable ones. So here's what actually happened: what worked, what didn't, and what I'd do differently.

The numbers that actually matter

Before getting into the mechanics, some context. I find it genuinely helpful when other customer support leaders put real numbers to what they're talking about:

  • When I joined Cleo in 2020, we were handling around 15,000 to 16,000 conversations per month
  • By the end of my tenure in January 2026, that number had climbed to 75,000 per month
  • Our subscriber base grew from 60,000 to one million in that same window

Those two growth curves look similar on a chart. But here's the thing that mattered most: the conversation line eventually moved below the subscriber line. That's the goal. That's what you're working toward.

The metric that drove that result was contact rate, the percentage of your monthly active users who reach out to your support team in any given month.

  • When I joined, we were sitting at around 12%
  • Over time, we brought that down to somewhere between three and four percent

Think about what that means in practice. If we'd kept a 10% contact rate as we scaled to one million subscribers, we'd have been handling 300,000 to 400,000 conversations per month. That's a completely different operation, and a vastly more expensive one.

What growth actually feels like

Growth looks clean in retrospect. In the moment, it's a different experience entirely.

There was a month, around August 2023, where our first response times went from 10 to 15 minutes to over 300 minutes. That happened because I took my eye off capacity planning for one month. I was behind on customer onboarding, and it took six months to recover.

If you're not accurately forecasting conversation volume and planning your team's capacity around it, this is what happens. One bad month compounds into six. It's a painful lesson to learn, and I had to learn it the hard way.

We also went 24/7 earlier than most teams would plan to, and that wasn't a strategic decision either. One of my agents went backpacking in Australia in February 2020, got locked down during COVID, and asked if she could keep working from there. We said yes, and suddenly we were a 24/7 operation. We built a team around her. By the time I left, we had:

  • 41 agents in Colombia
  • Seven agents in the US
  • Overnight coverage from the UK, Kenya, and the Philippines

Sometimes you throw your cap over the wall and figure out how to climb after it. We learned a lot of lessons that way.