Tuesday, 3 November 2026Global tech news & guides — what changed, and what it means for you.
AuTechWire
Original editorial illustration of a speedometer gauge, a laptop, and a glowing encrypted tunnel carrying data packets, in dark blue and amber

Does a VPN Slow Down Your Internet Speed? What the Tests Show

by

In brief

  • Independent tests of 54 VPNs average a 33% download slowdown, but the fastest lose under 10%. What decides your loss, and a 5-minute test to measure your own setup.
  • Across 267 tests with full results, All About Cookies found VPNs slowed downloads by 33.2% on average — but only about one in five tested VPNs kept its average slowdown under 20%, and the fastest averaged just 1.6% slower.
  • Distance matters more than encryption: the typical VPN lost about 20% on its nearest server versus about 35% on servers in other countries, and the encryption overhead itself is usually negligible with modern protocols.
  • WireGuard typically runs 15–30% faster in throughput than OpenVPN, and on connections under 100 Mbps the protocol gap barely matters because the internet plan itself is the bottleneck.
  • Test it yourself: baseline, three runs, one variable at a time, same time of day — then use the decision tree below to tell a VPN fault from a server fault from a connection fault.

The short answer, with numbers

The encryption itself is cheaper than most people expect. Surfshark’s own engineering summary puts it plainly: with modern protocols, the performance overhead of encryption is usually negligible, and the real impact comes from routing — your traffic takes a detour through the VPN server before reaching its destination. Cloudflare’s explainer makes the same point from the network side: the “trombone effect,” where data travels a long distance to the VPN server before reaching its final destination, is what introduces most of the latency, alongside server load.

What you actually feel is larger than the encryption cost, because independent testing measures the whole trip, not just the cipher. All About Cookies has tested 54 VPNs since March 2024, logging 270 speed-test results and retesting the top VPNs about every quarter. Across the 267 tests with full results, VPNs slowed download speeds by 33.2% on average. But the average hides the spread that matters: among VPNs tested at least four times, average speed loss ranged from 1.6% to 63.7%, and only about one in five kept its average slowdown under 20%. The fastest in its testing, NordVPN, averaged just 1.6% slower across eight rounds; Surfshark averaged 7.6% slower (last tested October 2026), ExpressVPN 11.9%, and Proton VPN 13.1%. For the wider market picture, including how VPN use itself is growing, see our VPN statistics hub.

Two practical takeaways fall out of the numbers immediately. First, stay close to home: the typical VPN in that testing lost about 20% on its nearest server, compared with about 35% on servers in other countries — and three VPNs were actually faster than the no-VPN baseline on nearby servers. Second, the gap between providers is the story, not the average: the difference between the best and worst tested setups is roughly fortyfold, so “a VPN slows you down by X%” without naming the provider, protocol, and server distance is not a usable answer. Treat every figure in this section as indicative, not a promise: your connection, your device, and your server choice move these numbers around, and even the best VPNs had off days in repeated testing.

The five things that decide your speed loss

1. Protocol. A current vendor comparison by SymlexVPN, updated in October 2026, reports that in independent testing over the past year WireGuard typically ran about 15 to 30 percent faster in throughput than OpenVPN, while noting that OpenVPN’s newer kernel-level offload has closed much of the old gap. WireGuard is the lightweight: roughly 4,000 lines of code and a fixed modern cipher suite. OpenVPN over UDP is the middleweight; OpenVPN over TCP is the heavyweight, because TCP guarantees delivery by re-sending lost packets, and when your traffic already runs inside another TCP session the two layers’ retransmissions stack up. Use TCP only when you need it to get through a restrictive network. For the full comparison, see our VPN protocols compared.

2. Server distance and load. Every kilometre between you and the VPN server adds latency, and a crowded server adds queueing delay — Cloudflare calls the distance penalty the trombone effect: even a data center a few miles from your home can require a round trip to a far-away VPN server and back. The nearest server is usually the fastest — but “nearest” means nearest low-load server, not nearest city. If your app shows server load, prefer a slightly farther server at 20% load over a near one at 95%.

3. Your device’s processor. Encryption and decryption happen on your hardware. A modern laptop or phone barely notices; an old phone, a budget tablet, or a router running the VPN for the whole house can become the bottleneck. If speeds are fine on your laptop but terrible on your old phone with the same server, the phone’s CPU is the suspect, not the VPN.

4. Wi-Fi vs wired. This one is not caused by the VPN, but it contaminates every speed test people run. Wi-Fi adds interference, signal drops, and shared-bandwidth slowdowns that get blamed on the VPN. If you are diagnosing speed, plug in with Ethernet for the test, then go back to Wi-Fi for daily use. If the VPN will not connect at all on public Wi-Fi, that is a different problem — see our guide to VPNs not connecting on public Wi-Fi.

5. Your baseline connection. A VPN cannot make your connection faster than your ISP allows — with one famous exception. If your ISP throttles specific traffic (some throttle video streaming at peak times), a VPN can hide what the traffic is, and speeds can actually improve, because the encryption prevents the provider from identifying and slowing down the throttled service — the one case Cloudflare and Surfshark both name. It is the exception, not the rule, and if someone’s VPN consistently doubles their speed, their ISP was shaping their traffic.

How to test it yourself (the repeatable method)

Do not trust a single speed test. Run this protocol — it takes about five minutes and gives you a number you can compare fairly.

Step 1: Baseline. Disconnect the VPN completely. Close every other app that uses the network. Go to a speed test site (Ookla’s Speedtest or fast.com) and run three tests, one minute apart. Write down the middle download figure, not the best one. This is your baseline.

Step 2: VPN on, nearest server. Connect to the nearest server your VPN offers. Wait 30 seconds for the connection to settle. Run three tests again, one minute apart. The middle figure is your VPN speed.

Step 3: Compute the loss. Subtract the VPN speed from the baseline, divide by the baseline. That is your percentage loss for that server and protocol.

Step 4: Change one variable at a time. Switch the protocol (WireGuard first, then IKEv2, then OpenVPN UDP) and repeat step 2–3. Then try a different server in the same city. Keep notes — you are looking for the combination with the smallest loss, and the differences are often dramatic: it is common to find one protocol much faster than another on the same server.

Rules for a fair test: test at the same time of day (networks congest in the evening), use the same test server on the speed-test site, and use Ethernet if you are comparing protocols rather than judging daily-use speed. Wi-Fi noise can easily swamp a small protocol difference. Separately, confirm the connection is actually private once you have your number — our guide to testing VPN leaks walks through checking for IP, DNS, and WebRTC leaks.

When slowness is the VPN vs the server vs your connection

Use this decision tree before changing providers:

  • Slow on every server, every protocol, but fast with the VPN off? The VPN app or your device is the bottleneck. Check CPU usage during a test, try a different device on the same network, and check whether your VPN app has a speed-focused protocol option you have not tried.
  • Slow on one server, fast on another in the same city? Server load. Switch servers; this is the single most common cause of “my VPN got slow this week.”
  • Slow only in the evening? Congestion — either your ISP’s or the VPN server’s. Test at an off-peak hour to separate the two.
  • Slow with the VPN off too? Not a VPN problem at all. Test your bare connection and talk to your ISP.
  • Fast on the VPN, slow off it? Your ISP is likely throttling something the VPN hides. Congratulations — this is the exception working in your favour.

Settings that recover speed

In order of impact: switch to WireGuard (or your app’s WireGuard-based protocol) if you are on OpenVPN; pick a nearer, less-loaded server; enable split tunnelling if your app offers it, so only traffic that needs the VPN goes through the tunnel — your video call to a local server does not need to detour through another country; use UDP, not TCP, unless the network forces TCP; and keep the app updated, because providers steadily optimise their protocols.

One more honest note: no settings change fixes a bad server network. If every server from one provider is slow at every hour, the provider’s capacity — not your settings — is the problem, and that is what the refund window is for.

What this means for you

  • Expect roughly a third slower on average across all tested VPNs — but under 10% with the fastest providers on nearby servers, and sometimes no loss at all close to home.
  • WireGuard (or equivalent modern protocols) first; OpenVPN TCP only to get through stubborn networks.
  • Nearest low-load server beats nearest city; server load is the most common cause of sudden slowness.
  • Test properly: baseline, three runs, one variable at a time, same time of day.
  • If the VPN is fast and your bare connection is slow on specific traffic, your ISP was throttling — the VPN is not the problem.
  • Persistent slowness on every server at every hour is a provider capacity problem, not a settings problem.

Sources

  • All About Cookies, “VPN Speed Test Results: 54 VPNs Tested Since 2024” — 54 VPNs, 270 tests; 33.2% average download slowdown; range 1.6%–63.7%; NordVPN 1.6%, Surfshark 7.6%, ExpressVPN 11.9%, Proton VPN 13.1%; nearest-server typical ~20% vs ~35% distant. HTTP 200, read firsthand 2026-11-01. https://allaboutcookies.org/vpn-speed-test
  • SymlexVPN, “WireGuard vs OpenVPN” comparison (updated Oct 27, 2026) — WireGuard typically 15–30% faster throughput than OpenVPN; under-100 Mbps bottleneck note; vendor comparison, attributed as such. HTTP 200, read firsthand 2026-11-01. https://symlexvpn.com/wireguard-vs-openvpn-comparison/
  • Surfshark, “Does a VPN slow down the internet?” — encryption overhead usually negligible; distance/routing dominates; 10 Gbps server ports; free VPNs slower through overcrowding. Vendor page, attributed as vendor claims. HTTP 200, read firsthand 2026-11-01. https://surfshark.com/blog/do-vpns-slow-down-internet
  • Cloudflare, “How VPNs affect Internet speed” — trombone effect, server load, encryption/latency tradeoff, ISP-throttling exception. HTTP 200, read firsthand 2026-11-01. https://www.cloudflare.com/en-ca/learning/access-management/vpn-speed/

Shoaib Hassan

Shoaib Hassan is the founder and editor of AuTechWire, an independent technology publication covering broadband, mobile networks, privacy, startups, and consumer tech.

Related Stories

Leave a Comment