EngineSpire Technologies Inc., home

DGDigital InfrastructureGuide

Managed IT Services: What a Useful SLA Should Contain

A service level agreement should state exactly what is supported, how quickly the provider responds, how performance is measured and what happens when targets are missed. This guide explains the clauses that matter and the vague wording to avoid.

Published
Updated
Reading
3 minutes
Prepared by
EngineSpire Technologies

Key points

  • Define scope precisely: users, devices, sites, applications and hours covered.
  • Priority levels need clear definitions and separate response and resolution targets.
  • Require regular reports against targets, with access to the underlying ticket data.
  • Make security duties, privileged access and incident notification explicit; attackers target providers.
  • Plan the exit from the start: documentation, credentials and data handover.

Scope

  • Services included: help desk, monitoring, patching, backup, security operations, network and cloud administration.
  • Assets covered, backed by an inventory that both parties can see.
  • Users and sites, including branches and remote workers.
  • Support hours and time zone, and what happens outside them.
  • Remote and on-site support, with travel areas and charges.
  • Exclusions: projects, hardware purchases and third-party applications.
  • How new users, devices and sites are added, and at what price.

Priority definitions and targets

PriorityDefinition exampleWhat to agree
CriticalA business-wide outage or an active security incidentResponse time, round-the-clock coverage and continuous work until restored
HighA major function degraded or many users affectedResponse and restoration targets, and escalation after a set time
NormalA single user or non-critical function affectedResponse and resolution targets during business hours
LowRequests and questionsFulfilment target for standard requests
  • Define "response" as a qualified engineer starting work, not an automated acknowledgement.
  • Separate targets for restoring service or providing a workaround from targets for permanent fixes.
  • Define when the clock pauses, for example while waiting for your staff, and when it restarts.
  • Name escalation contacts at both organizations.

Proactive services

  • Monitoring coverage and how alerts are handled.
  • Patching schedules, maintenance windows and compliance targets.
  • Backup schedules and regular restore tests with results shared.
  • Vulnerability scanning and remediation timelines.
  • Lifecycle and capacity reports that feed your budget planning.

Reporting and reviews

  • Monthly reports: tickets by priority, targets met, trends and recurring problems.
  • Patch compliance, backup success, restore tests and security events.
  • Quarterly service reviews with improvement actions.
  • Access to the ticketing system so you can verify the reports.

Security responsibilities

CISA and partner agencies from several countries have warned that managed service providers are attractive targets because access to one provider can lead to many customers. The agreement should require:

  • MFA on all provider access to your systems.
  • Named individual accounts, never shared logins.
  • Least privilege and time-limited administrative access where possible.
  • Logging of provider actions, with logs available to you.
  • Separation between your environment and other customers' environments.
  • Secure configuration and monitoring of remote management tools.
  • Notification of security incidents affecting you within an agreed time.
  • Disclosure and approval of subcontractors.

Data protection

  • A data processing agreement setting out how the provider handles personal data on your behalf, as the Data Privacy Act's rules on outsourcing require.
  • Confidentiality obligations and where data may be stored.
  • Cooperation during personal data breaches, including information needed for notifications.

Remedies

  • Service credits for missed targets, calculated simply.
  • The right to terminate for repeated or serious failures.
  • Credits should not be the only remedy for significant breaches.

Pricing and change

  • A clear pricing model: per user, per device, fixed fee or a combination.
  • Rates for out-of-scope work.
  • How and when prices can change.

Exit and transition

  • Handover of documentation: diagrams, configurations, asset lists and procedures.
  • Transfer of administrator credentials, licences and vendor contracts held in the provider's name.
  • Return or deletion of your data, with confirmation.
  • Transition assistance for a defined period and fee.

Warning signs in a proposed agreement

  • "Best efforts" in place of measurable targets.
  • Response times with no definition of response.
  • No reporting obligations.
  • Shared administrator accounts.
  • No exit or transition clause.

Common questions

What is the difference between response time and resolution time?

Response time is how quickly a qualified person starts working on an issue. Resolution time is how long it takes to restore service or provide a workaround. A useful SLA defines and reports both for each priority level.

Are service credits worth negotiating?

They show commitment, but credits are usually small compared with the cost of an outage. Clear definitions, reporting, escalation and the right to terminate for repeated failures matter more.

How should a managed service provider access our systems?

Through named individual accounts with MFA, least-privilege permissions and logged sessions that you can review or revoke. Avoid shared accounts and uncontrolled, always-on remote access tools.

Sources and further reading

  1. [1]Protecting Against Cyber Threats to Managed Service Providers and their Customers — CISA and partner agencies
  2. [2]SP 800-161 Rev. 1: Cybersecurity Supply Chain Risk Management Practices — NIST
  3. [3]Republic Act No. 10173: Data Privacy Act of 2012 — National Privacy Commission

Need this applied to your environment?

We can turn this guide into a scoped assessment of your own sites, systems and constraints.