Failed Projects, Contract Traps and Official Sources
ERP SLAs and support plans: uptime percentages, response times and service credits explained
What ERP and SaaS service level agreements promise, how uptime percentages translate into hours of downtime, how small service credits are, and what to ask for.
A service level agreement (SLA) sets what the ERP vendor promises on availability (uptime), support response and resolution times, planned maintenance and what you get if it misses, usually service credits off future invoices. Read it carefully: a 99.5% monthly uptime promise still allows more than three and a half hours of unplanned downtime a month, planned maintenance is often excluded, and service credits are typically a small percentage of one month's fee, far less than the cost of a day without your system. This guide translates SLA terms into practical numbers and lists what to ask for.
This guide is part of why ERP projects fail and how to avoid it.
What uptime percentages mean
Using an average month of about 730 hours:
| Monthly uptime promise | Downtime allowed per month | Per year (approx.) |
|---|---|---|
| 99.0% | about 7.3 hours | about 88 hours |
| 99.5% | about 3.7 hours | about 44 hours |
| 99.9% | about 44 minutes | about 8.8 hours |
| 99.95% | about 22 minutes | about 4.4 hours |
The calculation: downtime = (1 − uptime) × hours in the period. For 99.5%: 0.005 × 730 = 3.65 hours.
Check how the SLA measures uptime:
- Is planned maintenance excluded, and when does it happen in Australian time? A maintenance window set for a US weekend may fall in an Australian business morning.
- Is uptime measured for the whole service or per component (for example, the API or the reporting engine)?
- Who measures it, and can you see the reports?
- Is it measured monthly or quarterly? Quarterly averages hide bad months.
Support response and resolution
| Severity (typical definitions) | Example | What to ask for |
|---|---|---|
| Critical | System down, payroll cannot run, no orders can be processed | Response within an hour, 24/7, with continuous work until resolved |
| High | A key function fails with no workaround | Response within a few business hours |
| Medium | Function impaired with a workaround | Response within one business day |
| Low | Questions, cosmetic issues | Response within a few business days |
Response time is not resolution time. Ask for both, and ask which time zone and business hours apply. A vendor whose "business hours" are in another country may not answer during your working day.
Also clarify who supports you: the vendor for platform issues, the partner for configuration and integrations, a third party for add-ons. A single point of contact reduces finger-pointing.
Service credits: why they rarely cover your loss
Illustrative example: an SLA promises 99.9% uptime and offers a credit of 10% of the monthly fee if missed. The subscription is $4,000 a month. The system is down for a full business day.
- Service credit: $4,000 × 10% = $400
- Cost to the business: 25 staff unable to process orders or pay suppliers for 8 hours at an illustrative loaded cost of $60 an hour = 25 × 8 × $60 = $12,000, plus lost sales and overtime to catch up.
The credit is about 3% of the internal cost. Service credits are often stated as your sole remedy for SLA failures, so check the liability clause too. Use SLAs to measure performance and trigger rights, not as insurance.
What to negotiate
- Uptime of at least 99.5%, ideally 99.9%, for production, measured monthly.
- Planned maintenance outside Australian business hours, with advance notice.
- Response times for critical issues around the clock, especially for payroll days.
- Escalating service credits for repeated breaches.
- A right to terminate without penalty if the SLA is missed persistently (for example, three months in any six).
- Status page and incident reports after major outages.
- Support in Australian business hours from people who understand Australian payroll and GST.
- Recovery point and recovery time objectives for disaster recovery: how much data could be lost and how quickly service is restored.
Record answers with the ERP buying checklist and check the related terms in the contract traps guide.
Support plan pricing
Vendors often offer tiers: standard support included, premium support at an extra cost (commonly priced as a percentage of subscription fees), and a named customer success manager at higher levels. Partners sell support retainers by blocks of hours. Before buying premium support, compare what the higher tier actually changes: response times, hours of coverage, a named contact. See pricing models for how support fits into the annual cost, and model its effect on renewal uplifts.
Your own resilience
Even with a strong SLA:
- keep regular data exports (exit planning);
- have a manual fallback for critical processes such as payroll and dispatch;
- protect user accounts with multi-factor authentication;
- know who to call and how to escalate.
Frequently asked questions
Is 99.9% uptime good?
It is a common standard for business SaaS. It allows about 44 minutes of unplanned downtime a month. Check what is excluded.
Are planned outages counted?
Usually not, if notified in advance. Ask when they occur in Australian time.
Can I claim compensation for an outage?
Usually only the service credits in the SLA, unless the contract says otherwise or the law gives other rights. Check the liability clause and get advice for large losses.
Does the partner have an SLA too?
Ask for one. Partner support for configuration and integrations is often separate from the vendor SLA.
Sources
- Australian Cyber Security Centre — business continuity and cloud security guidance
- ACCC — consumer guarantees for services and unfair contract terms
- business.gov.au — business continuity planning
