How We Test VPNs

We run speed tests from three geographic regions (UK, US, Germany) at three times of day using a 1 Gbps fibre connection as baseline. We test each VPN's own recommended servers plus the nearest servers to each test location. The final speed score is the median of 40+ individual test runs, expressed as a percentage of our baseline.

We read every provider's privacy policy in full and assess: jurisdiction, data retention policy, log types collected (if any), third-party sharing, and whether an independent no-logs audit has been conducted. Audited policies score higher. Policies with vague language about 'connection metadata' are penalised.

We test each VPN against Netflix (US, UK, JP), BBC iPlayer, Disney+, HBO Max, Amazon Prime, and Hulu from a location outside the target country. Tests are run monthly. Consistent unblocking scores higher than sporadic unblocking.

Using ipleak.net and custom DNS leak tools, we verify that no DNS or WebRTC leaks occur during normal use and during a simulated connection drop. We also confirm that the VPN's kill switch functions correctly by killing the network process mid-session.

We install and use each VPN's apps on Windows, macOS, iOS, and Android. We assess: onboarding flow, server selection UX, protocol switching, kill switch accessibility, and support quality.

We compare the effective monthly cost of each VPN's best-value plan (usually 2-year) against the features delivered. Premium pricing is acceptable when justified by test performance; it is penalised when it isn't.