In short: To measure VPN speed correctly, first record a baseline with the VPN off, then test a nearby server and a distant one, repeat each run 2–3 times, and compare not only download but also ping, jitter and packet loss. On a modern nearby server a 5–15% download loss is normal. A bigger drop usually means a distant or overloaded server, an outdated protocol or a weak device — not a bad VPN. Test on Speedtest or Fast.com under identical conditions and at the same time of day.
What VPN speed actually means
Speed is not a single number. A VPN affects several parameters at once, and you have to read them together or you will draw the wrong conclusion.
- Download — how fast data reaches you; drives downloads and video quality.
- Upload — how fast you send data; matters for video calls, cloud sync and file uploads.
- Ping (latency) — response time in milliseconds; critical for games, calls and how snappy sites feel.
- Jitter — the variation in ping between packets; it makes voice wobble and video stutter.
- Packet loss — the share of data that never arrives; even 1–2% breaks calls and streams.
- Loaded latency — how much ping rises when the link is busy downloading (bufferbloat); often more important than idle ping.
Download almost always drops the most, but ping and jitter decide whether the connection feels comfortable.
A step-by-step testing protocol
A correct test is a comparison under identical conditions. A single with-VPN run without a baseline tells you nothing. Follow this order:
- Remove interference. Close torrents, cloud sync, auto-updates and streaming on every device — background traffic drags the result down.
- Stabilise the link. Test over cable or next to the router: weak Wi-Fi adds its own loss and jitter that you would wrongly blame on the VPN.
- Take a baseline with the VPN off. Open Speedtest and record download, upload and ping. This is your reference — not the number printed on your ISP plan.
- Test a nearby server. Turn the VPN on, pick a server in your country or a neighbouring one, and repeat the test on the same tool.
- Test a distant server. Connect to a server on another continent and measure again to see the cost of distance.
- Repeat and take the median. Run each point 2–3 times and use the median, not the best result.
- Fix the conditions. Test at the same time of day and against the same Speedtest node — evening load and a changed node distort the comparison.
One nuance: the speed test picks the measurement node closest to you (or to the VPN server). For a fair comparison, manually lock the same test node across all runs.
Metrics and their normal range on a nearby server
| Metric | What it shows | Target on a good nearby server |
|---|---|---|
| Download | Incoming data rate | ~5–15% below baseline |
| Upload | Outgoing data rate | ~5–15% loss, sometimes a bit more |
| Ping | Response time | +5–30 ms over baseline |
| Jitter | Latency variation | usually a few ms |
| Packet loss | Data that never arrives | close to 0% |
| Loaded latency | Ping rise under a busy link | moderate, no jumps of hundreds of ms |
These are reference points, not guarantees: the exact numbers depend on your ISP, device and protocol.
Testing speed on a phone: what is different
LiMP VPN runs on iPhone and Android, and on a phone several factors that do not exist on a PC reduce test accuracy. Account for them or the result is random:
- Battery saver caps the CPU clock and therefore encryption speed; turn it off for a fair run.
- Background app refresh quietly spends bandwidth during the test — pause it while measuring.
- Overheating. Over a long series the phone warms up and throttles its clock, so speed drops for reasons unrelated to the VPN.
- Signal quality. On mobile data, speed swings with signal bars and cell load — measure near a window and in one spot.
- Data saver in the system or the app can throttle background traffic and understate the result.
The practice is simple: take the baseline and the VPN run on the same network, in the same place, with battery saver off and a cool device. Do not confuse a speed test with a protection check — to confirm the tunnel really works, see how to test and fix VPN leaks.
What speed loss is normal
On a clean, nearby, modern tunnel, download loss usually stays within 5–15%. The reason is simple: part of the link goes to encryption and to the extra hop to the VPN server. If you lose more, look for the cause instead of blaming the VPN.
The main driver of latency is distance. Data cannot travel faster than the speed of light in fibre, so a server on another continent physically adds 150–300 ms even when the raw megabit figure looks fine. That is critical for games and calls and almost irrelevant for downloading a file. Pick the closest node from the LiMP VPN server list — it is the fastest way to restore responsiveness.
Common mistakes when measuring
- One run instead of a series. Speed fluctuates; a single point proves nothing.
- No baseline. Without a VPN-off measurement you are comparing against a marketing figure, not your real line speed.
- Testing a distant server. Connecting to another continent just in case and being shocked by the drop — that is expected.
- An overloaded free server. Public and free nodes are jammed in the evening; see free vs paid VPN compared.
- Weak Wi-Fi. Testing at the edge of coverage adds loss that has nothing to do with the VPN.
- A browser extension instead of a system VPN. An extension proxies only one tab and measures the wrong thing.
- Different times of day. A morning and an evening run are not comparable because of network load.
What to test with
- Speedtest (Ookla) — download, upload, ping and jitter, with a selectable node — lock it.
- Fast.com (Netflix) — a simple download test that shows real speed to Netflix's CDN; handy for streaming checks.
- Ping and traceroute — reveal latency and the route to the server, helping you see where time is lost.
- Built-in counters in the VPN app and router — useful for monitoring, but confirm the final number with an independent speed test.
Remember that Tor is far slower than a VPN by design — see VPN vs proxy vs Tor if you are comparing tools.
What to do if the speed is low
A slow result is almost always fixable. Quick steps:
- Switch to the closest server.
- Change the protocol to WireGuard — it uses the ChaCha20-Poly1305 cipher and is usually faster than OpenVPN; see how OpenVPN works and how it differs from WireGuard.
- Keep heavy apps out of the tunnel with split tunnelling, as covered in why a bank app fails over VPN.
- Try another server in the same location — a neighbouring node may be less busy.
- Still slow everywhere? A quality provider matters — see how to choose a VPN in 2026.
Checklist: a correct VPN speed test
- Close background downloads and streaming on every device.
- Test over cable or right next to the router.
- Take a baseline with the VPN off: download, upload, ping.
- Measure a nearby and a distant server separately.
- Lock a single speed-test node for all runs.
- Repeat each run 2–3 times and take the median.
- Compare not just download but ping, jitter and loss.
- Test at the same time of day.
Frequently asked questions
Why is my speed sometimes higher with the VPN on?
It happens when your ISP throttles or badly routes certain traffic: the VPN changes the route and bypasses that limit. Better peering between the VPN server and the target service can help too.
Speedtest or Fast.com — which should I trust?
Speedtest is more accurate for the full picture: upload, ping, jitter and node choice. Fast.com shows the speed specifically to Netflix's infrastructure and works as a quick streaming check. Use both and compare against your baseline.
Is a nearby but overloaded server better than a distant one?
Not always. A congested close node can perform worse than a free, slightly more distant one. That is exactly why you should try several servers rather than rely on one.
Can antivirus or a firewall lower VPN speed?
Yes. Traffic inspection by antivirus and firewall software adds latency and CPU load. For the test, check the speed with their standard settings and without aggressive inspection.
Wi-Fi and mobile speeds differ — how do I test them?
Test each network separately and do not compare them: Wi-Fi and LTE/5G have different loss, jitter and load. Take the baseline on the same network where you will actually use the VPN.
Should I reboot the router before testing?
Not mandatory, but useful before a series of runs: it clears stuck sessions and overfull buffers that make results drift. The key is identical conditions for the baseline and the VPN runs.
Why does ping matter more than raw speed for calls and games?
Voice, video calls and games send small packets but need a low, stable response time. High ping, jitter or loss break the connection even when the megabit figure is large.