All articles

How to choose a VPN: the 8 things that actually matter

VPN marketing pages all sound the same. Here are the 8 things that actually separate a good VPN from a bad one — logging policy, protocol, kill switch, server network, speed, platform support, price, and trust signals.

By JohnAugust 16, 20266 min read

Every VPN's homepage says "military-grade encryption," "blazing speeds," and "strict no-logs policy." That's not a comparison — it's the industry's shared marketing copy. Underneath it, VPNs actually differ a lot. Here's what to check instead of the homepage.

1. Logging policy — and whether it's been verified

The single most important thing a VPN sells you is trust: you're routing all your traffic through their servers instead of your ISP's. What they log (or don't) determines what they could theoretically hand over to a court, a hacker, or a data broker.

Look past the "no-logs" headline for:

  • What's actually excluded. Connection timestamps and bandwidth totals are sometimes kept even by providers who call themselves "no-logs" — that's different from logging which sites you visited.
  • Independent audits. A no-logs claim backed by a named third-party audit firm's published report is worth far more than the claim alone.
  • A history of being tested for real. A provider whose servers were seized or subpoenaed and had nothing useful to hand over is stronger evidence than any audit. This has actually happened to a handful of VPN providers — it's public record, worth searching for by name.

See what "no-logs" actually means for how to verify a specific provider's claim.

2. Jurisdiction

Where a VPN company is legally headquartered determines what governments can compel it to do. Countries in intelligence-sharing alliances (the "Five Eyes," "Nine Eyes," "Fourteen Eyes") can, in some circumstances, request data from companies operating there. Providers based outside these alliances — Switzerland and Panama are common choices — face different legal pressure.

This matters more for people with real threat models (journalists, activists, people in restrictive regions) than for someone who just wants to watch region-locked shows. Weigh it accordingly.

3. Protocol support

The tunneling protocol determines speed, security, and how well the VPN handles switching networks (Wi-Fi to cellular) or restrictive firewalls. WireGuard is the modern standard — fast, simple, well-audited. OpenVPN is older but more configurable and works on more restrictive networks. See the full protocol comparison for the tradeoffs.

A VPN that only offers one legacy protocol is a yellow flag in 2026.

4. Kill switch

If the VPN connection drops mid-session, does your traffic silently fall back to your real IP, or does the app cut your internet entirely until the tunnel reconnects? The second behavior is a kill switch, and it should be on by default or a one-tap setting — not something buried three menus deep or missing entirely.

5. Server network — size and honesty

A big server count is only useful if the servers are real. Some providers use "virtual servers" — software-defined locations that don't physically sit where they claim. That's not necessarily a problem (it can improve reliability), but a provider should disclose it plainly rather than let you assume every listed city has hardware in it.

What matters more than raw count: coverage in the regions you'll actually use, and whether the provider publishes real-time server load so you're not stuck on a congested one.

6. Speed

Every VPN adds overhead — encryption, an extra network hop, sometimes a longer physical path to the destination. Expect some slowdown; the question is how much. A well-run WireGuard-based network on nearby servers should feel close to your normal connection. A noticeably sluggish VPN even on a fast home connection is a sign of an overloaded or poorly run network.

There's no way to know this without testing — trial periods and money-back guarantees exist specifically so you can check speed on your own connection before committing.

7. Device and platform support

Check that the provider has a real, actively maintained app for every device you'll use — not just "works via manual OpenVPN config." Also check the simultaneous-device limit; some providers cap it low (5), others allow unlimited connections under one account, which matters a lot for a household.

Router-level support (running the VPN on your home router itself) is worth checking if you want whole-home coverage without installing apps on every device.

8. Price and payment options

VPN pricing is built around long-term commitments — the advertised price is almost always the multi-year rate, not the monthly one. See how VPN pricing actually works before comparing sticker prices across providers.

Also worth checking: refund window length (30 days is common, some offer more), whether they accept privacy-preserving payment methods (cash by mail, cryptocurrency) if that matters to you, and whether the free trial requires a card upfront.

Which of these matters most, by use case

If you mainly want...Weight these highest
Everyday privacy on public Wi-FiKill switch, protocol, price
To avoid being tracked by your ISPLogging policy, jurisdiction
Streaming from other regionsServer network size, speed
Real anonymity (activism, journalism)Logging policy + audits, jurisdiction, payment privacy
Whole-home coverageRouter support, simultaneous-device limit

Quick FAQ

Is a more expensive VPN always better? No. Price correlates weakly with quality — some of the most respected providers on logging and audits aren't the priciest. Spend the money on the criteria above, not the price tag itself.

Do I need a VPN with servers in every country? Only if you specifically want to appear in a lot of different locations. For most people, good coverage in the handful of regions they actually use matters more than a huge global count.

How long should I trial a VPN before committing to annual billing? Use the full refund window if one exists. Test on your actual devices, on your actual connection, doing what you actually do — not just a one-time speed test.

Does a VPN's app rating in the app store tell me anything useful? A little — very low ratings or a pattern of complaints about billing or connection reliability are worth reading. But ratings are easy to game and don't tell you anything about logging or audits.

TL;DR

Ignore the marketing copy. Check: logging policy (and whether it's been audited or tested), jurisdiction, protocol support, kill switch, honest server network, real-world speed, device support, and actual (not advertised) pricing. Everything else on a VPN's homepage is noise.