- List the request types you actually receive and decide who handles each one before you buy any tools
- Support tiers describe who has the skills; priority describes impact and urgency. Set response targets by priority, not by tier
- Treat password resets, account recovery and MFA changes as high-risk requests: verify identity through a trusted channel, then approve and log every change
- Go live with a minimum setup: one intake channel, categories, assignment rules, an identity check and simple SLA tracking
- Separate response, restoration, resolution and update targets, and match them to the capacity you have staffed or contracted
This guide covers support tiers, identity verification, ticketing and SLAs. For what a service desk does, how it compares with an MSP or IT management, and when to outsource, see our IT service desk guide.
Define your support tiers before buying any tools
Before you evaluate any software or service, list the kinds of IT requests your organisation actually raises and decide who should handle each type. This costs nothing and prevents you from over-engineering a support model for problems you do not have.
A three-tier model is common, but it is optional. A small firm can run well with general first-line support, self-service for standard requests (such as installing software from an approved catalogue) and a clear route to external specialists. What matters is that for each request type you know who receives it, who carries out the work, who tracks it and who escalates it when it stalls.
- Tier 1: First-line support. Logging and categorising tickets, printer issues, software installation questions, email configuration, basic hardware troubleshooting, and anything resolvable with a knowledge base article or a short remote session. Password resets and account unlocks often start here too, but only through a verified identity procedure (see the next section). Tier 1 staff need patience, a good knowledge base, remote support tools and clear escalation rules.
- Tier 2: Technical support. Network configuration, application errors that require log analysis, creating or deploying device images, VPN setup, software licensing problems, and anything requiring administrator-level access. Tier 2 engineers need genuine technical skills and the ability to diagnose problems without a script.
- Tier 3: Specialist support. Security incidents, infrastructure architecture, custom integrations, compliance-related system changes, and anything that needs expertise your Tier 2 team does not have. Tier 3 may involve external specialists, a managed security provider, or a vendor's own support team.
How tickets split across tiers varies widely between organisations, so do not plan staffing on rule-of-thumb percentages. Analyse the first 60 to 90 days of ticket data (volumes by category, time to resolve, how often tickets are escalated, and peaks by day and hour) before deciding how much internal or outsourced capacity you need. You may find that many routine requests can move to a self-service portal, or that a handful of complex systems generate most of the effort.
A tier describes who has the skills to handle a ticket. Priority describes business impact and urgency. A critical incident escalated to Tier 3 can need an immediate response, while a low-priority Tier 1 request can safely wait. Do not set response targets by tier alone, and remember that escalating a ticket to a specialist does not change its business priority.
Password resets and MFA changes are high-risk requests
Password resets, account recovery and changes to multi-factor authentication (MFA) methods look like routine Tier 1 work, but they are sensitive identity functions. Attackers know this. Calling the help desk while posing as an employee and asking for a password reset or a new MFA device is a documented route to account takeover: CISA's advisory on the Scattered Spider group describes exactly this technique. Voice cloning and other deepfake tools make impersonation easier still.
- Verify identity securely. Do not rely on details an attacker can find or guess, such as name, job title, manager or employee number. Use a call-back to a number already on record, confirmation from the user's manager through a known channel, an in-person check, or the verification features of your identity platform.
- Require approval and logging. Record who asked for the change, how identity was verified, who approved it and who carried it out. Notify the user through a separate channel after every reset.
- Add controls for privileged and executive accounts. Administrator, finance and executive accounts, and any change to MFA methods, may need a second approver, a specialist queue and a short delay before the change takes effect.
- Keep secrets out of tickets. Never write passwords, recovery codes, MFA secrets or API keys in ticket comments, emails or attachments. Use self-service reset or a secure one-time channel instead.
For more on how attackers get around MFA, read how attackers get around MFA, and see our guide to deepfakes and identity fraud for voice and video impersonation.
Ticketing and tracking: the minimum viable setup
The foundational rule of a service desk is this: if it is not in the system, it does not exist. Every request needs a ticket. Every ticket needs a state (open, in progress, waiting on user, resolved). Every resolved ticket needs a record of what was done. This is not bureaucracy for its own sake. It is the only way to manage a queue that more than one person is working on, and the only way to identify patterns when the same problem keeps coming back.
"A service desk without a ticketing system is a queue that nobody can see."
For small and medium businesses, the tool choice matters less than the discipline of using it. Freshservice, Zoho Desk, Jira Service Management and Help Scout all work at this scale. Freshservice is built for IT service management, with asset tracking, a self-service portal and automation rules. Zoho Desk integrates with other Zoho products if you already use that ecosystem. Jira Service Management suits teams already on Atlassian. A basic configuration of any of them can be running quickly, but an operational rollout takes longer: you need named owners, permissions for support staff, workflows for incidents and requests, SLA rules, reporting and guidance for users. The timeline depends on the complexity of your environment and the integrations you need, such as your identity directory, device management and monitoring alerts.
The minimum setup you need before going live is: one intake channel (a form or email address that creates tickets automatically), ticket categories that map to your request types, an assignment rule so tickets land with the right person, an identity verification step for account and MFA requests, and a simple SLA clock so you can see what is overdue. Everything else (asset management, automation, satisfaction surveys) can be added once the basics are running smoothly. Trying to configure everything at once is how service desk rollouts stall.
SLAs: what to commit to and what to avoid
A service level agreement (SLA) is a commitment about how quickly support will respond to, update on and resolve different types of requests. SLAs are useful because they set expectations, create accountability, and give you a way to measure whether your service desk is performing. They become a problem when they are set unrealistically, because you then spend more energy managing SLA breaches than fixing problems.
The most common mistake growing businesses make is setting the same SLA for everything. A single "respond within one hour" rule sounds professional, but a network routing problem that needs a qualified engineer will not be fixed in an hour, and a request for a new keyboard does not need a one-hour response. Committing to one timeline for everything creates a list of SLA violations that nobody trusts and everyone ignores.
A more durable approach is to set targets by priority. Priority is driven by impact (how many people, services or processes are affected) and urgency (how quickly the impact grows or how close a deadline is). Then separate the clocks that people often lump together:
- Response time: how quickly a person acknowledges the ticket and starts work (an automated email does not count).
- Restoration target: how quickly the service is restored or a workaround is in place, so users can work again.
- Resolution target: how quickly the underlying fault is fixed and the ticket can be closed.
- Update interval: how often the user or business owner hears about progress while the ticket is open.
- Escalation time: how long a ticket can sit at one level before it moves to a specialist or a manager. For example, a critical incident might go to the IT lead immediately, and a high-priority ticket after one hour without progress.
The table below is an illustrative example only, not a recommended baseline for any particular business size. It assumes support during business hours. If you need 24/7 coverage for critical services, that must be contracted and staffed explicitly, and it usually costs considerably more.
| Priority | Example | Response | Restoration | Resolution | Updates |
|---|---|---|---|---|---|
| P1 Critical | Core system down, many users or a key process stopped | 30 minutes | 4 hours | Root cause tracked as a problem | Every 30 to 60 minutes |
| P2 High | A team or key user blocked, no workaround | 1 hour | 8 business hours | 2 business days | Every 2 to 4 hours |
| P3 Medium | Degraded service, workaround available | 4 business hours | 1 business day | 5 business days | Daily |
| P4 Low | Question, minor issue or standard request | 1 business day | Not applicable | 5 to 10 business days | On status change |
Illustrative example, business hours coverage. These are contract examples; ITIL does not set response or resolution targets. Replace every figure with targets that match your own capacity and contracts.
Whatever numbers you choose must match the capacity you have actually staffed or contracted, and should be tested against your real ticket history before you publish them. It is better to set conservative targets and meet them consistently than to set aggressive ones and miss them. If you work with a managed provider, the SLAs and coverage hours in your contract should match what you publish internally. Mismatches create gaps that users always notice.
Avoid SLA targets based on what sounds good rather than what your capacity supports. If your Tier 2 engineer is one person covering multiple responsibilities, a 4-hour restoration target on complex issues will breach constantly. Start with what you can reliably deliver and tighten targets as your capacity grows.