Support operations rebuild for a 40-person services firm

The situation

Support ran entirely through a shared inbox. Three years of history, no structure, no ownership, and one person who knew which customers mattered. Mean first response was over six hours, mostly because work was batched — the person triaging would open the inbox in the morning and again after lunch.

What I found in the audit

Signal	Share of tickets	Was leadership aware?
Status queries on open jobs	34%	Partly — framed as “no updates”
Password and access resets	21%	No
Duplicate submissions	12%	No
Genuine faults	9%	Yes

Two-thirds of volume was administrative. That reframed the whole project: the problem was not responsiveness, it was that a human was acting as a search engine.

What changed

1. 01

   Ticketing with enforced ownership

   Every inbound message creates a ticket with one owner. No shared queues, because shared queues are where ownership goes to die.

2. 02

   Two-tier SLA

   P1 acknowledged in 30 minutes during business hours, everything else in 4 hours. Written down, visible to customers.

3. 03

   Knowledge base for the 21%

   Reset and access instructions, written by the staff who actually did the work, not by me.

4. 04

   Duplicate detection on intake

   Matched on customer + subject within a rolling window. Cut the 12% almost entirely.

Result

The part that mattered most

Not the system. Writing the escalation runbooks with the support staff rather than for them. Six people who now own the process, rather than six people following a document someone else wrote.