On average, Proton dropped about 15 percent of my unprotected speed, but that number needs some context. In a location like Atlanta, Georgia, midday on a Thursday, I experienced a drop of only around 3 percent. In Columbus, Ohio, in the evening on a Friday, that grew to a 25 percent drop. This type of variation is normal. Providers like Surfshark and NordVPN see similar variations and have similar speed drops on average.
The difference for Proton is that I’ve yet to stumble upon a real stinker of a server. I’m sure they exist—with some 15,000+ servers, you’re bound to find one at some point—but I haven’t seen them after weeks of use. Windscribe and ExpressVPN are competitive with Proton on average, but they also have some locations where I saw anywhere from a 40 to 60 percent drop in speed. Those results aren’t indicative of the speed overall (you just swap to a different server), but Proton gets you there faster.
That edge is likely due to Proton’s VPN Accelerator. I’ll admit, it sounded like nonsense. In the Proton VPN app, you’ll find a toggle for VPN Accelerator, which boldly claims to increase speed by up to 400 percent; not likely. Despite the speedup, I don’t think VPN Accelerator will reach anywhere near that quoted number, at least in the vast majority of cases.
Still, there are some advantages, most notably, BBR. Bottleneck Bandwidth and Round-trip propagation time, or BBR, is a congestion control algorithm developed by Google that’s been deployed on YouTube and Google itself. Rather than limiting packet transfer when packets are lost, as most congestion control algorithms work, BBR models the network and estimates available bandwidth. It doesn’t need to see lost packets to kick in.
Proton’s speeds aren’t entirely attributable to BBR, but I suspect it helps when connecting to servers over long distances. Connecting in the UK, for example, I saw an average speed loss of around 20 percent, which is much closer to my US results than it has any right to be.
