Contact center operations

What I have done in this field

I have built service units from nothing and taken over existing ones. From a site with forty seats to the service center of a private university that took on first-level support for three faculties.

Building

Site selection, recruiting lead time, technology, first client. When a build goes wrong, it almost always goes wrong in the lead time, not in the technology. If you want to open in January, you start hiring in October, and the knowledge base is in place before the first call.

Quality and pay

A quality system nobody understands changes nothing. I build scorecards a team manager can explain in a conversation, and I only tie pay to them once the measurement is stable.

When I introduced a scorecard for agents, there was a great deal of scepticism, and the works council was at the table. We went through the metrics together until it was clear what they reward and what they cannot measure. In the end the system worked, and it convinced the people who had been against it. Since then I start with the question of what a metric rewards, and only then with whether it can be measured.

Multilingual teams

You do not lead a team from nine countries with one rulebook. Expectations about response time, directness and escalation differ by market, and a single service level across all markets produces arguments about interpretation.

Knowledge is the real obstacle

When you build a first-level function, the technology is finished in weeks. The question of who gives which answer and who maintains it takes months. That is exactly where most overflow projects fail.

What AI has changed

I introduced an AI solution for inbound calls. It took load off the operation, and it set off knock-on effects I had not accounted for. Automating the simple cases also changes the remaining ones: how long they take, how hard they are, and what a good call even means afterward. That calculation belongs before the rollout, not after it.