What to Look for in an IT Support SLA

A Service Level Agreement is the part of an MSP contract that turns promises into commitments. It defines what level of service you will actually receive, how it is measured, and what happens when the provider falls short. The sales pitch tells you what a provider hopes to deliver; the SLA tells you what they will be held to. That makes it the single most important document to read closely before you sign. This guide walks through every component worth checking, defines the metrics in plain language, flags the pitfalls that hide in the fine print, and shows what real contract terms look like.

Uptime guarantees

Uptime is the percentage of time your services are operational, and it is usually the headline number in an SLA. The percentages sound similar but the difference in allowed downtime is large:

• 99.9% uptime: roughly 8.76 hours of downtime per year.
• 99.99% uptime: about 52.56 minutes per year.
• 99.999% uptime ("five nines"): under 5.26 minutes per year.

The number alone is not enough. Check whether the guarantee applies 24/7 or only during stated service hours, and how scheduled maintenance windows are treated, because planned downtime is often excluded from the figure. A 99.9% guarantee that quietly excludes nights, weekends, and maintenance is a weaker promise than it appears.

It also matters which systems the guarantee covers. An uptime figure that applies to the provider's monitoring platform but not to your email, your line-of-business application, or your internet connection is measuring the wrong thing. Ask exactly what the percentage is calculated against, who measures it, and how an outage is recorded. The right question is not "what is your uptime," it is "uptime of what, measured by whom, over what window." A provider who can answer that crisply is one who has thought about delivery, not just marketing.

Response time vs resolution time

These two metrics sound alike and mean very different things. Response time is how long the provider takes to acknowledge your issue and begin working it. Resolution time is how long until the issue is actually fixed. A provider can have a fast response time and a slow resolution time, so look at both. Confirm how the SLA defines resolution: a fully fixed problem, or merely a temporary workaround that lets them stop the clock. Look for tiered support, where issues are categorized by severity with different commitments for each, and check that the timeframes are realistic for the complexity of your environment.

Pay close attention to who assigns the severity tier, because that decision controls which commitment applies. If the provider unilaterally classifies your urgent problem as routine, your one-hour response quietly becomes a next-business-day one. A fair SLA either defines severity with concrete examples or gives you a say in the initial classification, with an agreed process for disputing it. Also ask whether the response-time clock runs only during business hours or around the clock, since a "two hour response" that pauses overnight can mean tomorrow morning for an issue reported at 5 p.m.

Escalation procedures

Some issues need to move up to senior engineers, and the SLA should say how that happens. Look for defined escalation triggers, so you know what kind of problem gets escalated and when, and a named point of contact who owns escalated issues. A clear escalation path is what keeps a critical problem from sitting in a general queue while your business is down.

Reporting and transparency

Reporting is how you verify the provider is meeting its commitments rather than taking their word for it. A strong SLA specifies how often you receive reports, what they contain (performance, incidents, uptime), and which key performance indicators are tracked. Regular reporting turns your review meetings into data-driven conversations instead of impressions, and it is the paper trail you need if you ever have to enforce the agreement.

Penalties for missed commitments

An SLA with targets but no consequences is not really an agreement. Find out what happens when the provider misses. The most common remedy is service credits, a discount on future invoices when uptime or response targets are not met. A typical schedule might credit a small percentage of the monthly fee for each tier of missed uptime, scaling up as the miss gets worse. Read how the credit is triggered and how you claim it, because some agreements put the burden entirely on you to detect and report the miss within a tight window, which makes the credit hard to ever collect.

Stronger agreements include contractual penalties or a right to terminate for repeated failures, which is the protection that actually matters when a relationship is going badly. A single missed target is noise; a pattern of them is a reason to leave, and your SLA should let you. Penalties matter for two reasons: they motivate the provider to deliver, and they give your business recourse when it does not. A provider who resists any penalty clause is telling you how confident they are in their own delivery. Our guide to MSP contract red flags covers the clauses that quietly remove this protection.

Contract length and flexibility

Contracts typically run one to three years. Longer terms are not automatically bad, but they should come with flexibility: the ability to renegotiate as you grow, and termination clauses you can actually use without crippling penalties. A provider confident in their service does not need to lock you in for years with no exit.

Reviews and performance assessments

The SLA should commit both sides to regular reviews, whether monthly, quarterly, or annually, and define what those reviews cover: KPIs, incident history, and overall satisfaction. These check-ins are where misaligned expectations get corrected before they become reasons to leave, and where the agreement gets adjusted as your needs change.

SLA metric definitions: a plain-language glossary

SLAs are full of shorthand. Here is what the common terms actually mean:

• Uptime / availability: the percentage of time a service is operational, measured over a defined window.
• Response time: how long until the provider acknowledges your ticket and begins work, not how long until it is fixed.
• Resolution time: how long until the issue is resolved, by the SLA's own definition of "resolved."
• MTTR (mean time to resolve): the average resolution time across incidents, a useful track record number to ask for.
• Severity tier: the priority level assigned to an issue, which determines its response and resolution commitments.
• Service credit: a billing discount issued when the provider misses a committed target.
• Maintenance window: scheduled time for planned work, usually excluded from uptime calculations.
• KPI: a key performance indicator the provider commits to measuring and reporting.

Knowing these terms lets you read an SLA without taking the salesperson's interpretation of it.

Common SLA pitfalls to watch for

The weaknesses in an SLA hide in definitions and exclusions, not in the headline numbers. Watch for these:

• Uptime that excludes the hours you need. A 99.9% figure means little if it only applies during business hours and excludes all maintenance.
• Resolution defined as a workaround. If "resolved" means a temporary patch, the provider can close tickets without fixing root causes.
• Targets with no penalty. Commitments that carry no service credit or remedy are aspirations, not obligations.
• Vague severity definitions. If the provider decides what counts as critical, your emergency can be downgraded to routine.
• No reporting requirement. Without committed reports, you cannot prove a missed target when you need to.
• Security left out entirely. An SLA silent on security and incident handling leaves your biggest risk unmanaged. The CISA cybersecurity best practices and the FTC's small-business guidance both treat these baseline protections as non-negotiable.

Sample SLA contract terms

It helps to see what reasonable terms look like in practice. The example below is illustrative, not a template to copy, but it shows the shape of a fair tiered agreement for a typical small business:

Alongside the tiers, a fair agreement would state an uptime guarantee with its measurement window, a service-credit schedule for misses, monthly reporting, and a quarterly review. Use these examples as a yardstick: if a provider's proposed SLA is far weaker, ask why, and if it is far stronger, make sure the price reflects realistic delivery rather than a number designed to win the deal.

Putting your SLA to work

A good SLA protects your business and sets the tone for the whole relationship. Read it against your real needs, define what each metric means in writing, and make sure the penalties give you genuine recourse. To see how the SLA fits into the larger evaluation, read our guide on how to choose a managed IT service provider, and if you are still deciding between models, our comparison of managed IT vs break-fix shows why an SLA only exists in the managed model. To estimate what this level of service should cost, use our MSP cost calculator, and download our free MSP Buyer's Guide for the full contract-review checklist.