Running an agency service desk: tiers, service catalog, and metrics

Reference article · 5 min read

A service desk is the part of an IT organization the rest of the agency sees. Staff judge technology by how a locked account or a failed permitting-system login gets handled, not by the network refresh. A desk that runs on habit looks fine for years, then struggles when a key person retires or a system fails during a council meeting.

The working parts are few: a catalog people can read, intake channels that all end in a ticket, tiers with written escalation rules, priority tied to response targets, knowledge articles that cut repeat work, and a few measures reported the same way monthly. Each part can be built without replacing the ticketing system first.

Write the service catalog in plain language

The catalog says what the desk supports and what a requester gets. Write it for the person filling out the form, naming items the way staff do: new employee setup, phone for a field vehicle, access to the permitting system. For each entry:

• What you get. The device, account, or change, described concretely.
• Who can ask. Any employee, a supervisor, or the department's system owner.
• What approval is needed. Name the approver role, not a person, so the entry survives turnover.
• A typical turnaround. A range, plus what makes it longer, such as a purchase or a background check.

Keep the first version short and add entries as patterns appear. Separate requests (something new) from incidents (something broken); they route, prioritize, and get measured differently.

Intake channels and who triages

Decide which channels you support: a self-service portal, a phone line, email that becomes tickets, a walk-up counter, an after-hours number.

• Every channel ends in a ticket. Work that arrives by hallway conversation still gets logged, or the desk's workload is invisible when staffing is questioned.
• Name the triage role and a backup. Triage means categorizing, setting priority, spotting duplicates or a wider outage, and assigning. It is a skill, not a leftover duty.
• Make intake usable. The portal should work with assistive technology, and field staff without a computer need a phone path that reaches a person.

Define tiers by authority, not seniority

Tiers describe what a person is equipped and permissioned to do.

• Tier 0. Self-service: password reset, knowledge articles, request forms, a status page.
• Tier 1. First contact. Resolves what it is authorized to resolve, and owns the ticket until it closes.
• Tier 2. Specialists by domain: network, identity, application administration, field devices.
• Tier 3. Engineering, the department system owner, or the vendor.

Write the escalation rules down: time triggers, such as a ticket sitting untouched past its target, and content triggers, where suspected compromise, a public safety system, or a filing deadline moves the ticket up at once. Keep ownership with the assigning tier so nothing is handed into silence, and decide in advance who may open a top-severity case with each vendor.

Priority, severity, and response targets

Severity describes technical impact. Priority decides order of work, usually as a grid of impact (how many people, which systems) against urgency (whether a deadline or a safety issue is involved).

• Publish response targets by priority. Be more cautious with resolution targets, since repair often depends on a vendor, a part, or someone else's outage.
• Anchor urgency to the agency calendar: payroll, agenda deadlines, court filing windows, billing cycles, elections.
• Give dispatch, radio, jail, and similar public safety systems their own category, their own after-hours path, and a named authority to declare a major incident.
• Internal targets are commitments to colleagues, not contract terms; where a vendor delivers the service, the contract governs.

Knowledge articles that prevent the next ticket

The cheapest ticket is the one that never arrives. Write the article when the ticket closes, from the steps that actually worked, and attach it to the ticket category so it surfaces during triage.

• Keep two shelves: staff-facing self-help in plain language, and agent-facing procedures with exact steps and permissions.
• Give every article an owner and a review date. Stale instructions cost more than missing ones.
• Keep credentials and personal, health, or investigative details out of ticket bodies. Tickets are agency records, subject to retention schedules and sometimes to public records requests.

The few measures worth reporting

Report a small set consistently:

• Time to first response, broken out by priority.
• Backlog and aging, including tickets past a defined age.
• Reopen rate, which exposes tickets closed rather than resolved.
• First contact resolution, as a check on whether Tier 1 is properly equipped.

Publish definitions with the numbers so a measure means the same thing next year, report trend rather than a single month, and break volume out by category and department. Ticket count used as a productivity score produces more, smaller tickets.

How to start, and what to ask a provider

1. Sort several months of ticket history into categories.
2. Write catalog entries for the categories generating the most volume.
3. Build the priority grid and response targets, and review them with department directors.
4. Write escalation rules, post the after-hours path, and start reporting monthly.

If part of the desk is contracted out, ask how coverage works after hours and on holidays, who the named escalation contacts are, and whether staff reach a person rather than a queue. Require in the contract that ticket history and knowledge content belong to the agency and stay exportable, that reporting uses your definitions, that any staff-facing portal meets your accessibility standards, and that subcontractors are disclosed. If desk staff may encounter criminal justice information, federal policy imposes screening and training requirements that change over time; confirm current expectations with your state's CJIS systems agency.

HQT works with state and local agencies on enterprise IT operations, including service desk and ticketing design.

Frequently asked questions