IT Manager

Keep your IT queue managed instead of worked, so one person never quietly becomes the whole function.

The moment the person running IT starts taking tickets, the queue stops being managed and starts being worked, and a small IT function quietly becomes one overloaded person. Your IT Manager holds the shape of the thing instead: the queue's health, the incident, the roadmap, and the question nobody asks until somebody is on holiday.

Queue health is age, not count. Forty open tickets where the oldest is two days is healthy. Twelve where three are three weeks old is not, and a count cannot tell those apart. It reads the age distribution, what is about to breach its window, what nobody owns, and what has been reassigned more than twice — because a ticket bouncing between people is a routing problem wearing a ticket's clothes.

It watches where the work lands. When one person resolves everything in a category, that is capacity risk now and knowledge risk later, and it appears in both reports.

Incident command keeps the roles apart. Severity comes from impact, not from who is loudest. Someone leads the technical work, someone communicates, someone keeps the timeline — and the commander is never also the one fixing the fault, because a person deep in a fault cannot see the shape of the response. When severity is ambiguous it declares high: standing down five minutes later costs nothing, declaring an hour late costs the hour.

Updates go out on the cadence even when nothing has changed. Silence during an outage generates more traffic than any update does. No cause is communicated before it is established, because an early theory becomes the official story and is very hard to retract.

The postmortem finds the cause and the escape. The cause is what broke; the escape is why nothing caught it before staff or customers did. Both matter and only the first usually gets written down. It measures time-to-notice explicitly, because most incident duration lives there rather than in the fix. No cause may be a person's name, and it produces at most three actions with owners and dates — three that happen beat twelve that do not.

The roadmap is sequenced by consequence, not by cost. Every item carries what it costs to do, what it costs to defer, and what breaks if nothing happens for six months. What is deferred is named as deferred, so the deferral is a decision the owner made rather than something that happened.

And once a month it asks what only one person knows. Which systems have a single administrator, which category only one person can resolve, which recurring task nobody else has ever done. The fix is a runbook a second person has actually followed, because one nobody has followed is a document.

It approves no spend, signs nothing, makes no production change, and never decides who is disciplined after an incident.

Connects to your internal queue, knowledge base, mailbox and estate records. Works alongside your helpdesk technician, systems and network administrators, and whoever owns identity.