Skip to content
Answer 3 min read

What Is a Good Uptime Percentage?

A

99.9% ("three nines") is a good practical uptime target for most websites — it allows under nine hours of downtime a year, which most outages fit inside easily. 99.99% is a stricter, enterprise-grade target; anything below 99% starts to mean multiple days of downtime a year, which is noticeable to real users.

How much downtime each percentage actually allows

Uptime percentages sound close together but the allowed downtime shrinks fast. Using a 365.25-day year (525,960 minutes, accounting for leap years):

The nines table

99% uptime = about 3.65 days of downtime per year.

99.9% uptime ("three nines") = about 8 hours 46 minutes per year.

99.95% uptime = about 4 hours 23 minutes per year.

99.99% uptime ("four nines") = about 52 minutes 36 seconds per year.

The same numbers, per month

Yearly figures can feel abstract, so here's the same math averaged per month: 99% is about 7 hours 18 minutes of downtime a month; 99.9% is about 44 minutes a month; 99.95% is about 22 minutes a month; 99.99% is about 4 minutes 23 seconds a month. Framed monthly, the jump from 99% to 99.9% is the difference between "a bad afternoon" and "barely noticeable" — which is why 99.9% is such a common practical target.

Picking a target that fits your site

For most websites and small SaaS products, 99.9% is a realistic and honest bar — it tolerates the occasional bad deploy or brief host hiccup without requiring redundant infrastructure spread across multiple locations or providers. Chasing 99.99% or higher usually means real engineering cost (failover, redundancy, on-call rotations) that only pays off once downtime has a direct, measurable cost — a checkout flow, a paid API, a service other companies depend on.

The number only means anything if it's measured honestly, from outside your own infrastructure, on a consistent schedule — not self-reported from inside a system that might be the thing that's down.

What a higher target actually costs to promise

Publishing an uptime target — in a status page footer, a contract, or an SLA — is a commitment to measure it consistently and often to credit customers if you miss it. Each additional nine typically means real spend: redundant servers in more than one location, automated failover that's actually tested, and someone on call to respond at 3 a.m. None of that is free, which is why most SLAs cluster around 99.9% rather than 99.99% — it's the point where the cost of the next nine stops being worth it for most businesses.

A target you can't operationally back up is worse than a lower one you consistently hit — an unmet 99.99% promise damages trust more than an honestly-stated 99.9% that's actually true every year.

What uptime percentage should a small website aim for?

99.9% is a realistic, honest target for most small sites and SaaS products — it allows under nine hours of downtime a year and doesn't require the redundant infrastructure that higher targets typically need.

How much downtime does 99.9% uptime allow per year?

About 8 hours 46 minutes per year, based on a 365.25-day year (525,960 minutes) — roughly 43 minutes per month on average.

Is 99% uptime bad?

99% allows about 3.65 days of downtime a year, which is noticeable to regular users if it happens in a few concentrated incidents rather than spread evenly. It's an acceptable floor for low-stakes sites, not a target to aim for on anything customer-facing.

What's the difference between 99.9% and 99.99% uptime?

99.9% allows about 8 hours 46 minutes of downtime a year; 99.99% allows about 52 minutes 36 seconds — roughly a 10x reduction, which usually requires meaningfully more engineering investment to achieve.

How do I know my actual uptime percentage?

You need outside-in monitoring that checks your site on a consistent schedule and records every failure — uptime percentage is only meaningful when it's measured independently of the system it describes.

Holter watches this for you: outside-in monitors plus dead-man heartbeats for silent failures. Free plan: 5 monitors, 5-minute checks, no credit card.

Start measuring your uptime — free