We ran 4,218 speed tests across 312 servers this cycle specifically to rank the variables. The results are not what most troubleshooting guides suggest, and the single biggest factor is one most apps hide behind a green dot.

What actually costs you speed, ranked

  • Server load — a 61% swing. The largest factor by a wide margin.
  • Physical distance — 34% between a nearby and a long-haul hop, even on the fastest provider.
  • Time of day — 22% between 04:00 and 21:00 on the same server. Consumer VPN networks are contended, like any ISP.
  • Protocol — 11% between WireGuard and OpenVPN UDP. Real, but smaller than the arguments online suggest.
  • Encryption cipher — under 3%. On any CPU made in the last decade, a rounding error.

The fix: change server, not settings

Because load dominates everything else, the first thing to try is simply a different server in the same city. Not a different country, not a different protocol — a different endpoint. In our testing two servers in the same city on the same provider differed by as much as 340 Mbps.

If your provider shows real-time load as a percentage, sort by it and pick anything under 40%. If it only shows a coloured dot, that is a design choice worth noticing.

Performance charts displayed on a monitor
Two servers in the same city, same provider, same hour: a 340 Mbps difference.

Then check the protocol

If you are on OpenVPN, switch to WireGuard — or your provider's WireGuard variant such as NordVPN's NordLynx or ExpressVPN's Lightway. That is worth about 11% on average and rather more on long routes.

If you are on OpenVPN over TCP, switch to anything else immediately. The penalty there is 40% and up. TCP is a fallback for restrictive networks, not a default.

Things that will not help

Changing your encryption cipher. Turning off features like ad blocking. Restarting the app. These come up constantly in support threads and our data shows all of them well inside measurement noise.

If a provider will not show you real-time server load as a number, assume the reason is that the number is bad.

When the provider is the problem

If you have tried several servers at different times and you are still losing more than a third of your line speed nearby, the network is oversubscribed. Our fastest provider held 96% of a 942 Mbps baseline on a nearby server; anything below about 70% is a provider problem rather than a configuration one.