Service level agreements: how to make service commitments enforceable

Almost every business now depends on services it does not control. Cloud hosting, payroll software, IT support, logistics, security and maintenance contracts all keep the lights on. When one of those services fails, the question is not only “how quickly can it be fixed?” but also “what can we actually do about it?”

The answer usually sits in the service level agreement (SLA). Many SLAs promise impressive uptime and response times, but turn out to be difficult to measure, easy for the supplier to dispute and almost impossible to enforce. Others contain service credits so small that they do not change the supplier’s behaviour at all.

This article explains how SLAs work under South African law, where they commonly go wrong, and how to draft service commitments that you can rely on when it matters.

What an SLA is, legally

An SLA is simply part of a contract. It may be a schedule to a master services agreement, a standalone document or a set of standard terms on a supplier’s website. In each case, it is governed mainly by the common law of contract. The service levels are contractual obligations, and failing to meet them is a breach of contract unless the agreement says otherwise.

That last qualification is important. Many suppliers draft their SLAs so that missing a target is not a breach, but only triggers a modest service credit as the customer’s “sole and exclusive remedy”. If you sign that, you may have given up your right to claim damages or cancel, even where the failures are serious and repeated.

Service levels must be measurable

A service level is only useful if both parties can tell, objectively, whether it was met. Vague commitments such as “best efforts”, “industry standard” or “prompt response” invite argument. Good service levels define:

  • the metric, for example system availability, response time, resolution time or delivery accuracy;
  • the target, such as 99.5% availability per calendar month;
  • how it is measured, including which tools are used, who measures, and what counts as downtime;
  • exclusions, such as planned maintenance windows, customer-caused faults and force majeure events, defined narrowly; and
  • the measurement period and reporting obligations, so that performance is visible every month and not only when a dispute arises.

Pay attention to the arithmetic. A target of 99.5% monthly availability allows roughly three and a half hours of downtime a month. If your business cannot tolerate that, the target is too low, no matter how good it sounds.

Service credits and the Conventional Penalties Act

Service credits are the most common SLA remedy. They usually take the form of a percentage reduction in the monthly fee when a service level is missed. Commercially, they are a quick, predictable way to compensate the customer without proving loss.

Legally, a service credit that becomes payable because of a failure to perform will often be a penalty stipulation under the Conventional Penalties Act 15 of 1962. Three consequences follow:

  • Penalty stipulations are generally enforceable. The Act was intended to make them so.
  • Under section 3, a court may reduce a penalty to the extent that it is out of proportion to the prejudice suffered by the creditor. Credits that are excessive relative to the actual harm are therefore vulnerable to reduction.
  • Under section 4, a creditor who claims a penalty may not also recover damages for the same breach unless the contract expressly allows it. If you want both service credits and the right to claim damages above the credit amount, the agreement must say so clearly.

Suppliers often add a cap on service credits, for example 10% or 15% of the monthly fee. Combined with a “sole remedy” clause and a general limitation of liability, this can mean that a prolonged outage costs the supplier very little. Customers should read the SLA, the limitation of liability clause and any exclusion of consequential loss together, not in isolation.

Escalation, persistent failure and the right to cancel

Credits compensate. They do not fix the problem. A well-drafted SLA should also include:

  • Incident classification, so that a critical outage receives a faster response than a minor issue.
  • Escalation steps, naming the roles that become involved as an incident persists, up to senior management.
  • Root cause analysis and remediation plans for serious or repeated failures.
  • A persistent failure trigger, for example missing the same critical service level in three months out of any six, which entitles the customer to terminate without penalty.
  • Step-in or transition assistance, so that you can move to another provider without losing data or continuity.

Under the common law, a party can cancel for breach only if the breach is material or the contract contains a cancellation clause (a lex commissoria). Because a single missed target is rarely material, an express persistent failure clause is often the only practical route out of a poorly performing contract.

Data, security and regulatory overlay

Where the supplier processes personal information on your behalf, it is an operator under the Protection of Personal Information Act 4 of 2013 (POPIA). Section 21 requires a written contract obliging the operator to maintain appropriate security safeguards and to notify you immediately if there are reasonable grounds to believe personal information has been accessed by an unauthorised person. The SLA is a natural place to set response times and reporting duties for security incidents.

Regulated businesses may face further requirements. For example, financial institutions are subject to the cybersecurity and cyber resilience joint standard issued by the Prudential Authority and the Financial Sector Conduct Authority, which deals with the management of third-party service providers. Their SLAs need to reflect those obligations.

If your customer is a consumer or a small juristic person that falls within the thresholds of the Consumer Protection Act 68 of 2008, terms that unfairly limit the supplier’s liability may also be challenged under that Act.

Checklist: making your SLA enforceable

  1. Identify the services your business cannot function without, and set targets that reflect your actual tolerance for downtime.
  2. Define each metric, how it is measured and who measures it. Insist on regular written reports.
  3. Keep exclusions narrow and require advance notice of planned maintenance.
  4. Decide whether service credits are your sole remedy, or whether you also want damages. If both, say so expressly.
  5. Check that credits are meaningful but not out of proportion to likely loss.
  6. Read the SLA together with the limitation of liability and exclusion clauses.
  7. Include escalation steps, root cause analysis and a persistent failure termination right.
  8. Provide for exit assistance and the return of your data.
  9. Align security incident reporting with POPIA and any sector-specific rules.
  10. Diarise a review of the SLA at renewal, and keep records of every service failure in the meantime.

How O’Reilly Law can help

We draft, review and negotiate master services agreements, SLAs, SaaS and outsourcing contracts for both customers and suppliers. We help clients set realistic service levels, structure service credits and liability caps, include practical exit rights and align the contract with POPIA and sector-specific requirements. Where a supplier is not performing, we advise on remedies, cancellation and dispute resolution.

Contact us to review or negotiate your service level agreements.

Related reading: Limitation of liability and indemnity clauses: what do they really protect? · Dispute resolution clauses: choosing how your contract disputes will be decided

This article is general information and not legal advice. Correct as at November 2026.