Quick version: a VPN that "works" in marketing (the app says Connected) and a VPN that actually hides you are two different things. The gap is bigger than most users realize. This is the universal method to verify your VPN is doing its job, regardless of provider.
The four signals every working VPN must produce
Whatever VPN you use - NordVPN, Mullvad, your own self-hosted OpenVPN box - a working tunnel produces the same four observable signals. Our checker measures all four in your browser, in about 2 seconds:
- Visible IP changed. Websites you visit see an IP address. With a real tunnel, that IP belongs to your VPN provider, not your ISP. Turn the VPN off, note the IP, turn it back on, refresh - the number must change.
- Country matches what your VPN claims. If your VPN app shows "connected to Tokyo" but our checker geolocates you in Frankfurt, something is wrong. The location must agree.
- WebRTC doesn't leak your real IP. Browsers expose all network interfaces via WebRTC. A good VPN blocks this at the firewall level; a leaky VPN doesn't, and a webpage with 5 lines of JavaScript can see through your tunnel.
- IPv6 isn't escaping the tunnel. Many VPNs only route IPv4. If your ISP supports IPv6, websites with IPv6 endpoints see your real IPv6 address even while your IPv4 is safely tunneled.
What VPN providers actually differ on
Once you understand the four signals, the question for each VPN becomes: does the default configuration handle all four? Some do, some don't. The differences are real and stack with secondary factors:
- Kill switch behavior. When the tunnel drops momentarily, does traffic stop or fall back to the ISP? Reputable VPNs default to "stop". Free VPNs and corporate "security suite" VPNs often default to "fall back" - which is the worst possible behavior for privacy.
- WebRTC blocking method. OS firewall block (best - works at the network layer) vs. browser config (good but only inside the browser) vs. nothing (most free VPNs).
- IPv6 handling. Tunnel IPv6 through the VPN (Mullvad, ProtonVPN) or block IPv6 entirely (NordVPN, Surfshark, ExpressVPN) or ignore IPv6 (long tail of cheaper providers - leak risk).
- DNS routing. Provider's own resolvers vs. system DNS. Windows multi-homed name resolution leaks DNS even when IPv4 traffic is tunneled correctly; reputable providers patch this on Windows install.
- Audit history. Some providers have audited no-logs claims (NordVPN, ExpressVPN, Surfshark, ProtonVPN, Mullvad, IPVanish, PIA, TunnelBear). Some have only marketing claims. Some have actual logging incidents in their history (HMA 2011, IPVanish 2018, PureVPN 2017).
- Jurisdiction. Where the company is registered determines which governments can compel data. Switzerland (ProtonVPN, PrivadoVPN) and Sweden (Mullvad) are outside Five/Nine/Fourteen Eyes. US (PIA, IPVanish, Hotspot Shield, Atlas) is inside Five Eyes. This matters for high-threat-model use.
- Streaming optimization. Whether the provider actively rotates IPs to defeat Netflix/BBC/Hulu detection. Premium tier (NordVPN, ExpressVPN, Surfshark, CyberGhost) does this; budget and privacy-first options often don't bother.