How to Evaluate VPN Speed (and Read Speed Claims Critically)

An Honest Disclosure About These Numbers

Let's be clear about what this page is and isn't. We did not run a controlled, first-hand VPN speed lab for this edition, so we do not publish a VPNTex Mbps score for any provider — not a headline figure, not a per-server table. We think that is the honest position when we cannot show you a reproducible test rig behind the numbers, and it is a deliberate contrast with sites that print a precise Mbps table you have no way to verify. Where speed matters in our reviews, we describe the protocol a provider ships (a modern WireGuard-class protocol is what matters most on current hardware) and the provider's own stated position, and we point you here to test it for yourself. The rest of this page is the genuinely useful part: what determines VPN speed, how to measure it on your own connection, and how to judge anyone else's speed claim.

What Actually Determines VPN Speed

A VPN can only ever slow your connection down a little — it cannot make your line faster than your ISP delivers. How much it slows down comes from a handful of factors. Protocol is the big one: a modern WireGuard-class protocol (WireGuard itself, NordLynx, or Lightway) is far more efficient than legacy OpenVPN and is what you should use by default. Distance to the server matters next — a server in your own country adds very little overhead, while a transatlantic or transpacific hop adds latency and reduces throughput for everyone, on every VPN. Server load is a live variable: a congested server at peak evening hours is slower than the same server at 3am. Your own baseline sets the ceiling — a VPN on a 1 Gbps fibre line and the same VPN on a congested hotel Wi-Fi will look nothing alike. And your hardware and encryption acceleration play a role on older or low-power devices. This is why a single Mbps number, stripped of the connection, server, protocol and time it was taken on, tells you almost nothing.

How to Test VPN Speed Yourself, Properly

The most trustworthy speed number for you is the one you take on your own line, because it reflects your ISP, hardware and location — not someone else's. Do it like this. First, measure your baseline with the VPN off, using a tool like Ookla's Speedtest or Cloudflare's speed test, and note the download, upload and ping. Second, connect the VPN to the nearest server and re-run the same tool on the same device, wired over Ethernet if you can — Wi-Fi adds its own variability that has nothing to do with the VPN. Third, run each measurement three times and take the average; a single run is noise. Fourth, repeat on a distant server (say, another continent) so you can see how the provider holds up over distance, not just next door. Fifth, test at a realistic time — your peak evening hours, not the quietest moment of the night. Compare the VPN-on figures to your baseline: retaining most of your speed on a nearby server is what a good modern VPN does, and a big drop everywhere usually points to the wrong protocol (switch to WireGuard/NordLynx/Lightway) or a congested server (switch servers).

The Mistakes That Make Most Speed Numbers Meaningless

Once you know the failure modes, you can spot a bad speed claim in seconds — your own or anyone else's. Mistake 1 — Testing on Wi-Fi: Wi-Fi speeds fluctuate with interference, distance from the router, and congestion, so a test run over Wi-Fi cannot isolate the VPN's impact from Wi-Fi variability. Test on a wired connection where you can. Mistake 2 — Single test runs: one speed test is essentially noise, because network conditions vary second to second; running a single test and publishing the result is like measuring the temperature once and calling it the daily average. Mistake 3 — Ignoring time of day: internet speeds vary dramatically through the day, so a VPN measured at 3am and compared to one measured at 8pm will differ for reasons that have nothing to do with the VPN. Mistake 4 — Different baseline connections: comparing VPN speeds taken on different internet lines is invalid — a VPN on a 100 Mbps connection cannot be meaningfully compared to one on a 1 Gbps connection. Any comparison worth trusting holds the baseline, device, protocol and time roughly constant.

What a 'Speed Loss' Percentage Really Means

The most portable way to talk about VPN speed is not raw Mbps but the percentage of your baseline the VPN retains. If your line does 500 Mbps without a VPN and 450 Mbps with it connected to a nearby server, that is 90% retained — a 10% loss. This framing travels between connections in a way a bare Mbps figure never can: 90% retention means roughly the same thing whether your baseline is 100 Mbps or 1 Gbps, whereas '450 Mbps' is meaningless unless you also know the baseline it came from. As a rough guide, a good modern VPN on a nearby server retains most of your speed, the figure falls as distance grows (true of every VPN), and any provider that drops your speed dramatically even next door is either using the wrong protocol or pointing you at a congested server. When you read a review, translate its numbers into retained-percentage in your head — if it doesn't give you the baseline to do that, treat the number with suspicion.

How to Read a Published Speed Benchmark Critically

Whenever a site shows you a VPN speed number, run it through this checklist before you believe it. Does it state the baseline connection the test was taken on? Without it, the Mbps figure is unanchored. Does it name the specific server or at least the distance category, since 'nearby' and 'transatlantic' are wildly different? Does it say which protocol was used, because a WireGuard result and an OpenVPN result on the same server are not comparable? Does it give a date — VPN performance shifts as providers add servers and change routing? Does it describe how many runs were averaged and at what time of day? And crucially, could you reproduce it under the stated conditions? A benchmark that answers those questions is worth reading. A precise-looking table that answers none of them — a tidy grid of Mbps figures with no baseline, no protocol, no date, no method — is exactly the kind of number this whole industry over-produces and under-substantiates. Precision is not the same thing as accuracy.

Provider Speed Claims vs Independent Tests

There are two kinds of speed claim in the wild, and they are not equal. The first is a provider's own marketing figure — useful as a statement of intent (this is the throughput the service is engineered for) but self-interested by nature. The second is an independent test run on hardware and a connection you can relate to. When you see a Mbps figure attributed to a VPN, the first question is simply: whose number is this? If it is the provider's own claim, treat it as an upper bound under ideal conditions. If it is an independent test, apply the checklist above. Either way, the most relevant speed data point is the one you generate yourself on your own line, which is why we would rather teach you to test than hand you a number to trust.

Why We Don't Publish Our Own Mbps Score This Edition

We would rather show you a gap than fill it with a figure we cannot stand behind. Running speed numbers that genuinely mean something requires a controlled, documented rig — a fixed baseline, wired connection, consistent server set, repeated runs across times of day, and published raw conditions — maintained over time. We are not operating that lab for this edition, so we do not print a VPNTex speed score, and we would rather say so plainly than dress up unverifiable numbers as first-hand testing. Where speed is decision-relevant, our reviews describe the protocol each provider ships and its own stated position, and this page gives you the method to measure it yourself. If we ever stand up a reproducible speed lab, we will document the exact conditions here and label those results clearly as first-hand — and until then, we will not imply testing we did not do.

Frequently Asked Questions