IT Support Tiers Explained: Tier 0 to Tier 3 (and What to Automate)
What each IT support tier actually covers — tier 0 self-service, tier 1 triage, tier 2 technical, tier 3 engineering — with example tickets per tier, how tiers map to a small IT team, and which tiers automation can absorb.
IT support tiers are how the industry describes escalation: the routine handled cheaply and fast at the bottom, the rare and hard handled by experts at the top. The vocabulary comes from enterprise ITSM, but it’s just as useful at a 50-person company — not as an org chart, but as a map of which work needs a human and which doesn’t.
The tiers at a glance
| Tier | What it is | Example requests | Who handles it |
|---|---|---|---|
| Tier 0 | Self-service — no human involved | KB articles, FAQ, self-service password reset, AI agent answers | The employee (plus the systems you built) |
| Tier 1 | Triage + known issues with documented fixes | Password resets, access requests, “how do I connect to VPN”, printer setup | Help desk generalists |
| Tier 2 | Technical investigation of unknown issues | Recurring VPN drops on one laptop, mailbox migration gone wrong, app misconfiguration | Sysadmins / senior IT |
| Tier 3 | Deep expertise: engineering, architecture, vendors | Network redesign, identity architecture, vendor escalations, incident response | Specialists, engineers, or the vendor itself |
Some orgs add a “tier 4” for external vendor support; most fold that into tier 3. The boundaries matter less than the gradient: as you go up, tickets get rarer, harder, and more expensive per resolution.
What each tier looks like in practice
Tier 0 — self-service
The cheapest resolution is the one that never becomes a ticket. Tier 0 is your knowledge base, your FAQ page, your self-service password reset, and increasingly an AI agent that answers questions from your documentation the moment they’re asked. Tier 0 only works if the underlying knowledge exists and is findable — which is why it fails as a static portal and works as an answer delivered in the channel where the question was asked.
Tier 1 — triage and known issues
High volume, low variance. Nearly everything at this tier has a documented fix or a repeatable procedure — that’s the definition of the tier. It’s also the tier that burns out IT generalists, because the work is interruptive and repetitive at the same time.
Tier 2 — investigation
The issue is real, reproducible, and not in the runbook. Tier 2 needs admin access, technical judgment, and time to diagnose. This is where you want your skilled humans spending their day.
Tier 3 — engineering and vendors
Architecture changes, security incidents, bugs only the vendor can fix. Rare, expensive, and impossible to automate — which is fine, because it’s a tiny share of volume.
The SMB reality: one person is all the tiers
At 25–500 employees there is no tiered org — there’s an IT manager and maybe a sysadmin, and they are tier 0 through 3 depending on the hour. The tier model is still the most useful planning lens they have, because volume is not distributed evenly: the overwhelming majority of requests are tier 0–1 shaped (documented answers and routine actions), while the work that actually requires their expertise is tier 2–3. Every hour spent on tier 1 is an hour taken from the work no one else can do.
What to automate: tiers 0–1, almost entirely
The tier structure tells you exactly where automation pays. Tier 0–1 work is defined by having a documented answer or a repeatable procedure — the two things an AI agent handles well:
- Questions with documented answers → the agent retrieves from your knowledge base and answers with a citation, in Slack, in seconds.
- Routine procedures → password resets with MFA verification, account lookups, access request intake — executed by the agent, audited, with humans handling only exceptions.
- Triage itself → when the agent can’t resolve, it escalates a ticket with the conversation attached and routes it to the right person — so what reaches your humans arrives pre-triaged with context.
What’s left for people is the work that deserves them: tier 2 investigation, tier 3 engineering, and improving the systems that keep tier 0–1 automated. That trade — machines absorb the repetitive majority, humans keep the interesting minority — is the entire economic case for an AI IT help desk.
Frequently asked questions
What is tier 1 IT support?
Tier 1 is the first human line of support: triaging incoming requests, resolving common documented issues (password resets, access requests, how-do-I questions, basic troubleshooting), and escalating anything deeper. Tier 1 work is high-volume and procedural — which is why it is the tier AI agents can absorb most completely.
What is the difference between tier 1 and tier 2 support?
Tier 1 resolves known issues with documented fixes; tier 2 investigates unknown ones. A tier 1 ticket is “reset my password” or “how do I join the VPN”; a tier 2 ticket is “the VPN disconnects every 20 minutes on this laptop” — it needs technical skill, admin access, and diagnosis rather than a runbook.
What is tier 0 support?
Tier 0 is self-service — the employee resolves the issue with no human involved: knowledge base articles, FAQ pages, self-service password reset, or an AI agent that answers from documentation. Every ticket resolved at tier 0 is a ticket tier 1 never sees.
How many support tiers does a small company need?
Formally, usually none — in a company of 25–500 people, one or two IT generalists are all tiers at once. The model still matters as a lens: it shows which portion of the workload (tier 0–1, typically the majority of volume) can be automated so the humans can spend their time on tier 2–3 work that actually needs them.