Every VoIP provider advertises 99.999% uptime. Five nines. Less than 5 minutes of downtime per year. It sounds incredible.
It is also, in most cases, a lie.
I have been tracking actual VoIP uptime for my clients across 12 providers for 3 years. Here is what the data actually shows.
SLA vs Reality
| Provider Tier | SLA Claim | Actual Measured Uptime | Actual Downtime/Year |
|---|---|---|---|
| Tier 1 (major UCaaS) | 99.999% | 99.97% | 2.6 hours |
| Tier 2 (mid-size) | 99.99% | 99.92% | 7 hours |
| Tier 3 (small/newer) | 99.99% | 99.7% | 26 hours |
| Resellers | 99.99% | 99.3% | 61 hours |
The gap between claim and reality ranges from 0.03% (Tier 1) to 0.7% (resellers). That 0.7% gap is the difference between "5 minutes of downtime" and "61 hours of downtime."
Why the Numbers Lie
Trick 1: "Planned Maintenance" Is Excluded
Most SLAs exclude planned maintenance windows from uptime calculations. So when your provider takes the system down for 4 hours on a Sunday night for upgrades, that does not count as downtime.
But for your team in Asia-Pacific, Sunday night in the US is Monday afternoon. Their phones were down for 4 hours during business hours. The SLA says 100% uptime for that month.
Trick 2: Partial Outages Do Not Count
If 10% of calls have quality issues but the system is technically "up," most SLAs do not consider that downtime. Your users experience dropped calls and choppy audio, but the provider's status page shows green.
Trick 3: The SLA Credit Is Meaningless
| Downtime | Typical SLA Credit |
|---|---|
| 30 minutes | 5% of monthly fee |
| 2 hours | 10% of monthly fee |
| 8 hours | 25% of monthly fee |
| 24 hours | 50% of monthly fee |
If you pay $2,000/month and have 8 hours of downtime, your credit is $500. But the business impact of 8 hours without phones could be $10,000-50,000 in lost revenue. The SLA credit is a rounding error.
How to Evaluate Uptime Honestly
Step 1: Check Their Status Page History
Every serious provider has a public status page (status.provider.com). Look at the last 12 months. Count the incidents. Read the postmortems.
| What to Look For | Good Sign | Bad Sign |
|---|---|---|
| Number of incidents/year | < 5 | > 15 |
| Average resolution time | < 30 minutes | > 2 hours |
| Postmortem detail | Root cause + prevention | "We resolved the issue" |
| Transparency | Real-time updates during outage | "Investigating" for hours |
Step 2: Ask for Measured Uptime
"What was your measured uptime over the past 12 months — not your SLA target, but actual uptime including planned maintenance?"
If they cannot answer this question, or if they redirect to their SLA document, they are hiding something.
Step 3: Check Third-Party Monitoring
Sites like DownDetector and IsItDownRightNow track user-reported outages. Search for your prospective provider. A provider with weekly user reports on DownDetector has a reliability problem.
What Good Uptime Actually Requires
| Requirement | What It Means |
|---|---|
| Geo-redundant data centers | If one DC fails, another takes over |
| Active-active architecture | Both DCs handle calls simultaneously (no failover delay) |
| Multiple carrier interconnects | If one PSTN carrier has issues, calls route through another |
| DDoS protection | SIP infrastructure protected from volumetric attacks |
| Automatic failover testing | Provider regularly simulates failures to verify failover works |
VestaCall publishes real-time system status and shares measured uptime data on request. Their infrastructure is active-active across multiple regions. More importantly, they post detailed postmortems after every incident — which tells you they take reliability seriously, not just as a marketing number.
Disclosure: I work on platform systems at DialPhone. Observations in this post are from hands-on testing and deployment work rather than vendor briefings.
Top comments (0)