Back home

Company

Methodology

The short version: every price is either read directly off the provider's own pricing page, or clearly flagged when it isn't. Nothing is presented as fact that wasn't confirmed.

Two confidence levels, always shown

Every plan on every category page carries a confidence label:

  • Official — read directly off the provider's own current pricing page at the time stated in that page's "last checked" date.
  • Approx — the vendor's pricing page is JS-rendered, bot-protected, or otherwise couldn't be read directly, so the figure is cross-referenced across independent sources instead. Shown with a small "approx" tag and a tooltip explaining why — never silently presented as confirmed.

How ranking works

Rankings are computed, not curated. Each category sorts by a disclosed metric — usually price-per-vCPU or price-per-GB for compute, straight price for VPN and hosting where the spec doesn't translate as cleanly. The default sort and the "best value" stamp are both driven by that same metric, applied identically to every row regardless of whether that provider has an affiliate relationship with us. See how we make money for the commercial side of that promise.

Currency conversion

Prices are shown in the provider's native currency (USD, EUR or GBP) with a USD-equivalent for cross-currency sorting only. That conversion uses a fixed, approximate rate — not a live feed — so treat the USD-equivalent figure as directional, not exact. The native price is always the number that matters.

Automatic and hand checks

Where a provider publishes its prices through a public API — Vultr, Akamai/Linode, Azure and AWS — we ask it every day. For most other providers a script opens the provider's own pricing page every day, the way a visitor from the UK sees it (which is why some prices are in pounds), and reads each plan's price. Every answer is stored, so we keep a price history. A price that jumps more than 60% in one check is held back until a person has looked at it, and a plan the script can no longer find keeps its last checked price and date. The remaining prices are checked by hand against the provider's own pricing page. Each row shows the date its price was last checked: treat that date as the actual freshness of the data, not a cosmetic timestamp.

Benchmarks

Today's rankings compare published specs (vCPU, RAM, storage) against price — not measured real-world performance. Rented, benchmarked instances (sysbench / fio / network throughput) are on the roadmap but not live yet; nothing on the site currently claims to be a benchmark result.

Found something wrong?

Prices change, and hand-checked data can be stale or simply wrong. Tell us and it gets corrected — that's genuinely useful to us, not an inconvenience.