Skip to content
itsorted.online
Maintenance guide

99.9% Uptime Explained: What SLA Percentages Really Mean

Hosting companies, email providers and IT support contracts all promise uptime as a percentage: 99.5%, 99.9%, 99.99%. The numbers look almost the same, but the downtime they allow differs by hours a month. The fine print matters even more: what counts as downtime, what's excluded, and what you get when the promise is broken, which is usually a small credit you must claim yourself. This guide converts SLA percentages into minutes, explains how credits work using Google Workspace's published SLA as an example, and shows why independent monitoring is your evidence.

By IT Sorted teamReviewed 7 min readHow we write guides
in

Key takeaways

  • 99.9% uptime = up to 43.2 minutes of downtime in a 30-day month, or 8.76 hours a year.
  • Each extra 9 cuts allowed downtime by 10 times: 99.99% is 4.3 minutes a month.
  • SLAs define 'downtime' narrowly and exclude scheduled maintenance and problems on your side.
  • Google Workspace gives 3 extra days of service if monthly uptime falls below 99.9% but stays at or above 99%.
  • Credits usually must be claimed within a deadline, e.g. 30 days for Google Workspace.
In this guide · 7 sections
  1. 01Uptime percentages converted to downtime
  2. 02What counts as downtime (and what doesn't)
  3. 03Service credits: what you get when the SLA is missed
  4. 04Why your own monitoring is your evidence
  5. 05What uptime to ask for in contracts
  6. 06Do it yourself or get help?
  7. 07Frequently asked questions

Short answer

99.9% uptime allows about 43 minutes of downtime in a 30-day month, or about 8 hours 46 minutes a year. 99.99% allows about 4 minutes a month; 99.5% about 3.6 hours. A missed SLA usually earns a service credit you must request, not a refund, so keep your own uptime monitoring records.

Uptime percentages converted to downtime

The arithmetic is simple: take the percentage that isn't uptime and apply it to the minutes in the period. A 30-day month has 43,200 minutes and a 365-day year has 525,600.

Uptime SLAAllowed downtime per 30-day monthAllowed downtime per year
99%7 hours 12 minutes87.6 hours (about 3.65 days)
99.5%3 hours 36 minutes43.8 hours
99.9% ("three nines")43.2 minutes8.76 hours
99.95%21.6 minutes4.38 hours
99.99% ("four nines")4.3 minutesabout 53 minutes
Calculated from 43,200 minutes per 30-day month and 525,600 minutes per year.

Most SLAs measure per calendar month, so a 31-day month allows slightly more. The difference between 99.5% and 99.9% is about three hours a month, which matters for a booking site or online shop, and much less for a brochure website.

What counts as downtime (and what doesn't)

The percentage is only as good as the definition behind it. Providers define downtime narrowly, and that definition decides whether an outage you experienced counts at all.

  • How it's measured. Google's Workspace SLA defines downtime as a period when the user web interface has more than a five percent user error rate, measured on Google's servers. A slow but working service may not count.
  • Scheduled maintenance. Many hosting SLAs exclude planned maintenance windows announced in advance, sometimes at business hours in your time zone.
  • Your side. Problems caused by your code, plugins, DNS changes, expired domains or exceeding plan limits are usually excluded.
  • Third parties. Internet outages, attacks and 'force majeure' events are commonly excluded.
  • Partial outages. One feature failing, such as email sending but not receiving, may not count unless the SLA covers that component.

Service credits: what you get when the SLA is missed

An SLA rarely promises compensation for your lost sales. It promises a credit against future bills, scaled to how badly the target was missed. Google Workspace's SLA is a clear published example.

Monthly uptimeGoogle Workspace credit
Below 99.9% but at least 99.0%3 days of service
Below 99.0% but at least 95.0%7 days of service
Below 95.0%15 days of service
Source: Google Workspace Service Level Agreement, checked October 2026. Credits are extra service days or a monetary credit on a future invoice.

Google also says customers must request the credit by opening a support case within 30 days of becoming eligible, or they lose it, and caps the total credit per month. Most hosting and SaaS SLAs work the same way: you claim, you prove, and the payout is a fraction of the monthly fee.

Why your own monitoring is your evidence

Providers measure uptime with their own tools. To claim a credit, or to judge whether a provider is good enough, you need independent records: when the site went down, for how long, and what error it showed.

  1. Monitor from outside

    Use an external uptime service that checks your site every few minutes from the internet, not from the host's network.

  2. Check the pages that make money

    Monitor the homepage plus checkout, booking or contact, with a keyword check so an error page counts as down.

  3. Keep the incident log

    Export or screenshot outage records with start and end times. That's what a credit claim needs.

  4. Claim within the deadline

    Diary the claim window when an outage happens; 30 days is common.

  5. Review the yearly total

    Compare your measured uptime with the SLA once a year before renewing hosting or support contracts.

UptimeRobot's pricing page lists a free plan with 50 monitors checked every 5 minutes, and paid plans with checks every 60 seconds or faster (as of October 2026, US$). Our uptime monitoring tools guide compares free options and what to alert on.

What uptime to ask for in contracts

Match the SLA to what downtime costs you, not to the highest number on offer.

SLA checklist before you sign

Uptime is only one part of a support agreement. Our IT support contract SLA checklist covers response and resolution times, and our managed vs shared hosting guide explains how hosting type affects reliability.

Do it yourself or get help?

Setting up a free uptime monitor takes ten minutes and is worth doing today. The work that slips is watching the alerts, keeping records and acting on them, including claiming credits and deciding when a host isn't good enough.

Our Monitor plan (S$29 a month) checks your website every 5 minutes, alerts on SSL and domain expiry, watches email blacklists and sends a monthly health report, so you have the uptime record when you need it.

Frequently asked questions

About 43 minutes in a 30-day month (43.2 minutes), or about 8 hours 46 minutes over a year. A 31-day month allows slightly more. Most SLAs measure uptime per calendar month, so a provider can have one 40-minute outage every month and still meet a 99.9% promise.

A factor of ten. 99.9% allows about 43 minutes of downtime a month; 99.99% allows about 4.3 minutes. Each extra nine cuts the allowed downtime by ten times and usually costs more. For most small business websites, 99.9% with fast support matters more than chasing four nines.

Usually you can claim a service credit against future bills, scaled to how far uptime fell. Google Workspace, for example, adds 3, 7 or 15 days of service depending on the shortfall. You normally have to request it within a deadline, such as 30 days, and credits rarely cover your real losses.

Usually not. Many hosting and SaaS SLAs exclude planned maintenance announced in advance, along with problems caused by the customer, third-party networks and force majeure events. Check the exclusions and when maintenance windows happen in Singapore time before you rely on the uptime figure.

Keep independent monitoring records. An external uptime service that checks your site every few minutes logs when it went down, for how long and what error it returned. Export or screenshot the incident, open a support case with the provider within its claim deadline, and attach the record.

Rather have it done for you?

Fixed price, set up in your own accounts.

Order online, get a written scope, and keep full ownership of everything we set up. Questions first? Ask us in writing.