Performance

How to Spot Fake Uptime Guarantees in Web Hosting

Most "99.99% uptime" promises are marketing. Here is how to read the SLA, test the real numbers, and pick a host that actually stays online.

Priya ShahBy Priya Shah·June 25, 2026·How we test

What an uptime guarantee actually promises

Almost every shared and VPS host advertises 99.9% or 99.99% uptime on the pricing page, but the number on the homepage is rarely the number in the contract. The real promise lives in the Service Level Agreement (SLA), and it usually carves out planned maintenance, DNS issues, third-party network problems, and anything the host labels force majeure. Once you subtract those, the effective guarantee can drop to a much weaker figure.

The other half of the promise is the remedy. A guarantee with no compensation is just a slogan. Reputable hosts spell out a service credit — for example, 5% of the monthly fee for each hour of downtime beyond the threshold, capped at a month of hosting. If the SLA only offers a vague apology or requires you to prove the downtime yourself with their tools, treat the guarantee as marketing copy, not a contract.

Read the SLA before you read the price

Open the host''s legal page and search for the words uptime, SLA, and service credit. You are looking for three things: the exact uptime percentage, the list of exclusions, and the refund mechanism. 99.9% sounds tight, but it allows 43 minutes of downtime per month. 99.99% allows just over 4 minutes. The gap between those two numbers is the difference between a host that invests in redundancy and one that does not.

Pay close attention to how downtime is measured. Some hosts only count an outage if it lasts longer than 15 minutes in a single block, which means a flaky server that drops for 10 minutes every hour technically never triggers the SLA. A serious provider measures in 1-minute increments from their own monitoring, and lets you submit third-party reports as evidence.

Verify the claim with independent monitoring

Never trust the host''s own status page as the only source of truth — it is written by the people who benefit from understating outages. Set up an external uptime monitor (UptimeRobot, Better Stack, and Hyperping all have free tiers) against your real domain, not just the server IP, so you also catch DNS and CDN failures. Run it from at least two geographic regions to separate global outages from regional routing problems.

Give the monitor 30 to 60 days before you draw conclusions. A single bad afternoon does not mean the host is unreliable, and a clean week does not mean it is. Look at the monthly summary: total downtime, longest single incident, and how the host communicated during the outage. A provider that posts a clear root-cause analysis within 48 hours is doing the job; one that goes silent is telling you what the next outage will look like.

Red flags that the guarantee is hollow

Watch for guarantees that exceed what the underlying infrastructure can deliver. A single-server shared plan on a single data center cannot honestly promise 99.999% uptime — that requires multi-zone redundancy most budget hosts do not run. If the marketing number is dramatically higher than what comparable hosts at the same price point offer, the difference is almost always in the exclusions.

Other warning signs: the SLA is buried three clicks deep, the service credit is paid in account credit only (never cash refund), or the company reserves the right to amend the SLA at any time without notice. Any of those, on their own, is a reason to negotiate or move on. Together, they mean the guarantee is decorative.

Frequently asked

Questions readers ask about this topic

Is 99.9% uptime good enough for a small business site?

For a brochure or blog, yes — 99.9% allows about 43 minutes of downtime per month, which most visitors will never notice. For an e-commerce store or a site that drives leads, aim for 99.99% (around 4 minutes per month) and make sure the SLA actually pays out when missed.

Does planned maintenance count against the uptime guarantee?

Almost never. Nearly every host excludes scheduled maintenance from the SLA, as long as they give advance notice (usually 24 to 72 hours). Check how often the host schedules maintenance and whether it happens during your peak traffic hours.

What is the difference between uptime and availability?

In practice they are used interchangeably, but technically uptime measures whether the server is running, while availability measures whether visitors can actually reach your site. A server can be up while DNS, the CDN, or the network in front of it is down — which is why you should monitor the public URL, not just the IP.

How do I claim a service credit when downtime happens?

Open a support ticket within the window specified in the SLA (often 30 days), include the outage start and end times, and attach independent monitoring logs. If the host stalls or refuses without citing a specific exclusion, that is your signal to start migrating.
Keep exploring · Performance

Where to go next on Hostilo

Newsletter

One email a month. Hosting deals, new reviews, no fluff.

Related reading