1. Home
  2. /
  3. Blog
  4. /
  5. SLAs
SLAs 7 min read

SLA Comparison Checklist: 15 Service-Level Terms That Hide Real Cost

A complete checklist for comparing SLAs across vendors. The 15 terms most teams overlook, why they matter, and how to spot SLA red flags during negotiation.

Server racks in a modern data centre — uptime SLAs at scale
Photo by Manuel Geissinger on Pexels

Two vendors quote 99.9% uptime. One pays a 10% service credit when they miss it. The other pays nothing because their SLA carves out scheduled maintenance, third-party outages, force majeure, and anything they label "investigation pending". Headline uptime is the easy number to compare — the cost lives in the carve-outs. Here are the 15 SLA terms every comparison should pull out.

The uptime metric

1. Headline uptime percentage. 99.9% = 8.76 hours of downtime per year. 99.99% = 52.6 minutes. The difference is dramatic — make sure the number is comparable across vendors.

2. Measurement window. Monthly, quarterly or annual? Monthly windows are stricter for customers because a single bad day can breach the SLA. Annual windows let vendors smooth over bad weeks.

3. Outage definition. Does an "outage" require 100% unavailability, or counts degraded service? "Functional but ≥50% slower than baseline" should count.

Carve-outs that swallow the SLA

4. Scheduled maintenance. How much downtime is excluded? 4 hours/month is normal; 8+ hours starts to gut the SLA.

5. Emergency maintenance. Unbounded "we'll let you know" emergency maintenance windows are a red flag — push for a cap (e.g. 2 hours/quarter).

6. Third-party dependencies. If the vendor blames AWS for an outage, does the SLA pay out? It should — the customer doesn't care which sub-component failed.

7. Force majeure scope. Earthquakes, sure. "Cyberattacks of any nature" — push back, that's the vendor's job to defend against.

Response and resolution times

8. Priority tiering. How are P1/P2/P3 tickets defined? The vendor's definition of P1 might be much narrower than yours.

9. Time-to-first-response. The vendor's commitment to acknowledge a ticket. 15 minutes for P1 is good; 4 hours is poor for enterprise contracts.

10. Time-to-resolution. The vendor's commitment to actually fix the problem. Many SLAs don't include this; demand it.

11. Escalation path. Who do you call after the first level fails? Phone numbers in the contract, not just a portal.

Penalty mechanics

12. Service credit calculation. How is the credit calculated? % of monthly fees? % of the affected service tier? Round-down or round-up?

13. Credit claim process. Some vendors require the customer to file the claim within X days, with specific evidence. Miss the deadline, no credit.

14. Maximum credit cap. Most SLAs cap credits at 10-30% of the monthly fee. The remaining liability lives in the main contract liability cap.

15. Termination right. What level of repeated SLA breach gives the customer a termination right? Three breaches in 12 months is a fair starting point.

Comparing SLAs across vendors

A spreadsheet-based SLA comparison is brutal because every vendor structures their SLA exhibit differently. AI SLA comparison tools normalise the vocabulary — vendor A's "Service Credit Calculation" and vendor B's "Penalty Schedule" map to the same row. Free plan covers 4 comparisons per month, enough to validate the workflow with your most recent SLA negotiations.

Related reading

Run an AI vendor comparison in 60 seconds

Compare vendor proposals, RFPs and contracts with AI. Free plan: 4 comparisons / month.

Start free

Related articles