How VPNs actually work
The questions people ask before they buy one, answered with our own measurements rather than with adjectives.
15 pages, each built from the same daily collection: the server locations seven providers publish across 155 countries and territories, the money-back windows and jurisdictions they state, and speed we measured ourselves across 14 cities and 46 paired runs.
What that buys you is a count for the specific country you need rather than a server total, and a speed figure with a sample size attached rather than a claim. Where we have not measured something, these pages say so instead of filling the gap.
Guides
- Browser extensions vs the full app — One application, not the device
- Can you be identified through a VPN? — What the provider holds, and which law it holds it under
- Choosing a VPN for travel — Set it up before you leave, not after
- Free VPN or paid VPN? — What the difference actually costs you
- How to hide your IP address — Four methods, and what each actually covers
- Is it legal to use a VPN? — Four categories, and where each country sits
- Living abroad and keeping access to home — The country you left is the one to check
- No-logs VPNs, and who checked — Auditors named, jurisdictions listed
- Public Wi-Fi and what a VPN actually fixes — The honest version, including what it does not cover
- Reaching AI services from restricted countries — Blocked at both ends, which changes the answer
- VPNs and crypto trading — What the provider keeps, and under whose law
- VPNs and gaming — What helps, what hurts, and by how much
- What happens when you turn it off — Immediately, and what stays behind
- What is a VPN? — The mechanism in plain terms, with measurements
- Why is my VPN slow? — 40% to 90% of the line survived, depending on the city
Where the numbers come from
Provider coverage is read from each company's own published server list once a day and counted per country — 155 countries and territories, seven providers. Speed is measured by hand rather than by script: 14 cities, 46 paired runs, comparing two services against the same raw line in the same session, with a baseline measurement of the bare line in 11 of those cities.
We only name a faster provider in a city where we have five or more paired runs and a gap of at least 10%. That threshold is met in 4 of the 14. In the other ten we publish the numbers and say the sample is too small.
Every file behind those numbers is published as open data under CC BY 4.0, and the full method — including what the figures do not prove — is on how we test. If you find an error in the data, we would rather have the correction than the link.