VPNs (scraping and privacy)

This is the channel that starts with a correction: “best VPN for scraping” is mostly the wrong question. A consumer VPN terminates in a datacenter, and the targets that matter block datacenter IP ranges — so for scraping, the tools are residential proxies or dedicated datacenter IPs (see the Proxy channel). VPNs are reviewed here for the jobs they are genuinely right for: unblocking geo-restricted pages, keeping browsing private, and avoiding ISP throttling. Every price is read from the vendor’s own pricing page on the date shown, and dated.

What lives here

• Reviews — one VPN per page. URL pattern: /vpn/<brand>-review/ (for example /vpn/protonvpn-review/). Each page covers the published plans, the device limits, the logging claims and what is and is not verifiable from outside.
• The explainer. /vpn/best-vpn-for-scraping/ — why the question is usually wrong, when a VPN is still the right tool, and what to buy instead (pointing at the proxy channel).
• The data table. Every figure on these pages is backed by a dated row in the dataset — see the Data hub.
• The ranked toplist. Once two or more VPN reviews are published, this section ranks them by score, price and date.
• The vendor directory. Every VPN with a page on this site is listed in this section.

The questions every page answers

• What does the cheapest published plan include, and what billing term does the price require?
• What is the no-logs claim, and what can actually be verified from outside?
• How many devices, and how many simultaneous connections, does the plan allow?
• What does the provider publish about its server fleet, protocols and jurisdiction?
• Where a VPN is being considered for scraping, the page says plainly which tool is the right one.

How the numbers are collected

Every figure on this channel was read from the vendor’s own pricing, documentation or legal page on the date shown on the page. If a vendor changes a price, the row is re-read, re-dated and logged in the changelog. The full procedure — including what we test on real machines and what we refuse to guess — is on the How we test page.