CS System Administration + Zendesk
End-to-end administration of the customer-support stack at consumer scale — Zendesk Help Center, chatbot deployment, and the analytics layer that made support measurable across 86 million monthly active users.
What was broken
Supporting 86 million monthly active users means support can't be ad hoc. The Help Center, the automation, and the reporting all had to operate as one coherent system — and stay measurable — or sheer volume would bury the team. Nothing about that scale forgives an improvised stack.
What I built
I administered the Zendesk stack end to end: structured the Help Center so users could self-serve, deployed chatbot automation to deflect the repetitive contacts, and built the analytics layer that turned raw support activity into numbers leadership could actually act on. The pieces were designed to work together, not as bolt-ons.
What changed
A support operation that scaled with the product rather than against it — meaningful self-serve deflection through the Help Center and bot, and support performance that was finally measurable across the full 86M-MAU base. Decisions ran on data instead of anecdote.