HOW WE SCORE WEB HOSTS
Every host in the directory earns a single number: HRI (HostList Ranking Index), out of 100. It is built from four equal components and an evidence-backed signal layer, with two hard rules underneath it all: no paid placements, and no invented numbers.
The four equal components
The base score is 100 points split evenly: four components, 25 points each. No single component can carry a host on its own, and none can sink it alone. A host is only as strong as its weakest column.
Third-party review signals only: Google Places is primary, with Trustpilot and G2 as corroboration or a discounted fallback. Rating and volume both count, and a bonus couples Trust to verified performance. Poor reviews never subtract, and internal HostList reviews are excluded so every input stays independently auditable.
How fully a profile is documented: description, logo, segment, founding year, plan and feature data. A blank listing cannot outrank a host that has documented what it actually offers. This is the component a host owner can move fastest by claiming and filling in their profile. HRI v3.0 will replace it with verified public disclosures when the delayed cutover ships; HRI v2.2 remains current until a new effective date is published at hostlist.io/hri/changelog.
How recently the profile and its signals were verified. A continuous-time decay curve rewards recent updates and re-verification and quietly discounts stale data, so a score reflects what is true now, not what was true two years ago.
Real speed, not marketing claims. Core Web Vitals are weighted LCP 10, INP 6, CLS 5, TTFB 4, sourced through a strict ladder that always prefers the strongest available evidence over a weaker estimate.
The performance ladder
Performance is the one component where the evidence varies wildly between hosts, so it runs on a strict ladder. Each host gets the strongest signal available, and a stronger rung always wins. We never average a benchmark with a guess.
An identical test site load-tested on the host's real plan. The strongest signal: it isolates server infrastructure from how a marketing site is built.
Google CrUX p75 field data (LCP, INP, CLS, TTFB) on the provider's own website. Real users, but the provider's site, not yours.
An editorial infrastructure-tier read (CDN, caching, platform) used only when no benchmark or field data exists.
A neutral starting score, not a penalty. It rises as data is collected and operational signals are verified.
The full weights and formulas live in the performance methodology and the HRI™ spec.
Signals on top of the base
Two hosts can score identically on the four components and still differ in ways that matter. That is what the signal-augmentation layer captures: verifiable micro-signals like documented free SSL, a public status page, a published uptime commitment or a named data-centre region.
Each signal is evidence-backed. It carries the URL it was captured from and the date it was captured, and it decays over time, so a host has to keep re-verifying to keep the credit. Nothing here is a vendor claim taken on faith. If it moved the score, you can click through and see why.
The promises underneath
A ranking is only worth as much as the rules it refuses to break. These four hold on every host, every month.
Advertising and featured listings never change rank. A sponsor firewall separates revenue from the algorithm, and the ordering you see is the ordering the score produces.
Every signal that moves a score is inspectable on the host profile with an evidence URL and capture date. Missing data is labeled as missing, never guessed or filled with vendor marketing.
Trust relies only on third-party platforms where review provenance can be independently checked. Disputes are adjudicated against recorded evidence, not opinion.
Shared plans are not compared head-to-head with dedicated servers. The label on each score tells you how strong the evidence behind it is, so you read like for like.
Go deeper
Hosting speed vs site speed, the four-rung performance ladder, and how to benchmark your own host.
The complete algorithmic specification: every term, weight, input and audit rule behind the score.
A monthly market snapshot: score, review, price and performance medians across the directory.
TTFB, uptime, Core Web Vitals, CDN and 25+ hosting terms defined for buyers, not engineers.
DNS, SSL, hosting, WHOIS, redirect and blacklist checkers. Run a domain and see the data yourself.
Put any two hosts head to head on score, reviews, performance tier and price.
Frequently asked
What is HRI?
HRI (HostList Ranking Index) is a 100-point algorithmic rating of a web host. It combines four equally-weighted base components, Trust, Completeness, Freshness and Performance, each scored from 0 to 25, with a signal-augmentation layer of verifiable, evidence-backed micro-signals on top. Every host in the directory is scored the same way.
Can a host pay to rank higher?
No. Advertising and featured placements never affect rank. A sponsor firewall keeps revenue separate from the algorithm, and the leaderboard order is produced entirely by the score. This is the single most important promise behind the index.
Why is a host's HRI lower than expected?
The usual reasons, in order of frequency: missing profile fields (description, logo, segment, founding year), no third-party review signal on Google, Trustpilot or G2, a profile that has not been updated in over 30 days, no public status page, and no documented free SSL. Each is independently fixable by the host owner.
Where does the review data come from?
Trust uses third-party platforms only, primarily Google Places, with Trustpilot and G2 as corroboration or a discounted fallback when Google data is absent. Internal HostList reviews are tracked separately and do not feed the score, so review provenance stays independently auditable.
How often is HRI recalculated?
Base components recalculate on every signal write through the enrichment runner. The Freshness curve is continuous-time, so even a page load computes the current decay for each signal. A full directory recompute also runs on the 1st of every month.
How is hosting performance actually measured?
Through a strict ladder that always prefers stronger evidence: a controlled benchmark on the host's real plan beats CrUX field data, which beats an infrastructure-tier estimate, which beats a neutral baseline. We are manually deploying identical test website instances across the top 200 hosts by HRI rank.
Founder, HostList.io · Co-founder, Seahawk Media · Host of the WP Legends podcast.