What we publish, how it's produced, what it doesn't cover, and what we do when we get something wrong.
Last updated: 2 November 2026
Every speed figure comes from our own automated tests. The system connects to each server, runs the same Ookla speed test, records the result and moves on. It uses the same tool, settings and timeouts for every provider. We don't use figures supplied by VPN providers. Our methodology has the full details, including its limits.
This site is funded by affiliate commissions: if you buy a VPN through one of our links, we may be paid at no extra cost to you. We currently test only Surfshark and NordVPN, and we have affiliate links for both, so we have a financial interest whichever one you choose.
The commission a provider pays isn't an input to any result. The test code has no record of affiliate agreements, and rankings are calculated only from measured results. Affiliate links are labelled, and they're marked in the page code as sponsored.
On our comparison pages, the winner for each location is whichever provider measured better on the metric you've selected. The overall pick is whichever provider wins more of the locations where both were tested. If a location or overall result is exactly tied, we show Surfshark because it's the cheaper plan. Where only one provider has been measured, we show that provider and say so.
Providers can't pay to change a result, ranking or conclusion, and they don't get advance notice of tests or access to our systems. If a provider sends a technical correction, we handle it the same way as anyone else's: we check it against our data and only change what the evidence supports.
We measure speed and connection reliability. We don't audit a provider's privacy practices, logging claims, security or customer support, and we don't test whether a VPN works with particular streaming services. Weigh those separately when choosing a VPN.
Our live data pages, including server and country pages, are generated from our test results. Their explanatory text comes from templates that insert the measured figures, so it updates as new results arrive and isn't written individually for each page. Articles and explainers are written and checked before publication, and any figures in them come from our own data unless we name another source.
If we find a mistake in an article or policy, we correct it. If a test fault produces bad data, we fix the fault and say so on the affected page where it changes what the page tells you. To report an error, contact us. Prices and deals are checked periodically, but they change often, so confirm them on the provider's website before you buy.