Support that scales is a system, not a headcount

When support volume grows, the instinct is to add people. People do help. But headcount is the lever you reach for when you don't have a system — and it stops working long before the users stop arriving.

Two experiences shaped how I think about this. The first was administering customer support at Picsart across 86 million monthly active users, where nothing forgives an improvised stack: the Help Center, the chatbot, and the analytics all had to operate as one coherent thing, or sheer volume buried the team. The second was the opposite situation — standing up a support operation from nothing on Jira Service Management, with no queue, no routing, and no repeatable workflow. Just tickets arriving and improvisation after that.

Both pushed me to the same conclusion: support is a pipeline, and you design it like one. Intake, routing, workflow states, SLAs, and the reporting that tells you where it's clogged. The goal is consistent handling from queue to done that doesn't live in one person's head.

The measurable part matters as much as the mechanical part. If you can't see deflection, response time, and where work piles up, you're flying blind — and “hire more agents” is the only lever you've got. Once it's measured, better levers appear: a Help Center article that kills an entire category of tickets, a bot that absorbs the repetitive third of contacts, a routing rule that stops bouncing tickets between teams.

Headcount scales support linearly. A system scales it the way the product scales. That difference is the whole game.