--- title: "How we review web hosting" url: "https://hostidea.com/methodology" canonical: "https://hostidea.com/methodology" generated: "2026-10-15" note: "Markdown twin generated at build time from the same content as the HTML page. The HTML page is canonical." --- # How we review web hosting What kind of site HostIdea is today, where every number comes from, the monitoring programme we are building, the metrics we record, the weights we apply and how often everything is refreshed. Last updated Oct 13, 2026 ## What kind of site this is HostIdea is, today, a review aggregator and translator with its own view. We read each host’s terms and price pages and turn them into plain tables. We collect the uptime, response-time and load measurements that other people have published, cite them by name, date and method, and put them side by side so you can see where they agree and where they do not. Then we add our own reading: what the numbers mean for the kind of site you run, what a host leaves out of its documentation, and where the published tests are too thin to lean on. What we are not yet is a testing lab. Our own monitoring windows are opening host by host, and until one has run for a host, no uptime or response figure on that review is ours. The review says so in the first line of its results section, and the data file keeps our tests and everyone else’s in separate groups. When our own data exists, it leads; the published measurements stay as context. ## Where the numbers come from today Every figure on a review page belongs to one of three layers, and the page says which: 1. **The host’s own published terms**: prices on every billing term, plan limits, refund rules, uptime guarantees and service credits, infrastructure. We read the official pages and legal documents ourselves, cite the URL, and record the date we checked. Marketing claims are reported as claims. 2. **Independent measurements**: uptime monitoring, response-time and load tests published by named third parties, cited with the reviewer, the date and the method they state. Each review lists them in a table under Test results and stores them in the host’s data file as `thirdParty[]`, apart from our own `tests[]`, so the two never mix. Where sources disagree, the review says so. We also record what a source does not state (plan, location, interval), because a 30-day window on an unstated plan is not the same evidence as a year of 30-second checks. 3. **Our own checks**: price verification on a schedule, real support tickets with timestamps, and monitoring windows as they open. Each is labelled as ours. Where none of the three has a number, the review says “not measured yet” rather than printing one. Scores built without our own monitoring are marked provisional in the review and recomputed when a window completes. ## The monitoring programme The sections below describe the test programme we run as monitoring windows open. A review’s data box states whether a window is open for that host and from when. We buy a plan from each host, install the same WordPress test site on it, and monitor that site every 60 seconds from outside the host’s network for at least six months before a score is based on it. ## Which plan do we buy? The entry shared or WordPress plan, because that is what most readers buy. When a host sells two tiers that differ in a way that matters for performance (for example separate CPU allocations) we buy both and say so in the review. Plans are bought under our own names, paid with our own cards, and renewed at the public renewal price so we see the real cost. ## What is on the test site? - WordPress, current release at the time of install, updated monthly. - The default theme (Twenty Twenty-Five), Akismet, a contact form plugin and WooCommerce with the sample product set. - No page cache plugin. We use whatever server-side caching the host enables by default, because that is what a new customer gets. - The same 1.2 MB home page, the same 24 products and the same image set on every host. ## What do we monitor? | Metric | How | Interval | Where from | | --- | --- | --- | --- | | Uptime | HTTPS check of the home page, 30 s timeout, two failed checks in a row count as down | 60 s | Three regions (US East, Frankfurt, Singapore) | | Response time (TTFB) | Time to first byte of the home page, uncached by a query parameter | 60 s | Same three regions, averaged per month | | Global TTFB | Cached home page from 10 regions | Quarterly | 10 regions | | Load | 0 to 100 virtual users over 60 seconds against the shop page, k6 | Quarterly | US East | | Support | Three real tickets or chats per quarter with timed first response | Quarterly | Our account | | Pricing and features | Manual check of the official pricing page, with the date recorded next to the table | Every 45 days at most | Official site | The monitoring service, the regions and the exact check configuration are listed on each review’s data box, and the raw results are published as JSON at `/data/{host}.json`. ## How do we score? Each host gets six sub-scores from 0 to 10. The overall score is the weighted average below, rounded to one decimal. | Sub-score | Weight | | --- | --- | | Performance | 25% | | Uptime | 20% | | Support | 15% | | Features | 15% | | Value | 15% | | Ease of use | 10% | - **Performance** is the monthly average TTFB from the test site’s own region, mapped linearly: 100 ms or less scores 10, 1000 ms or more scores 0. Load-test response time can adjust this by up to one point. - **Uptime** is the measured uptime over the window: 100 percent scores 10, 99.9 percent scores 7, 99.5 percent scores 3, below 99 percent scores 0. Outage count breaks ties. - **Support** combines first-response time, whether the issue was resolved, and the channels available, from our logged support tests. - **Features** counts what is included at the entry price against a fixed checklist (backups and retention, staging, free domain, SSL, email, SSH, migration, CDN) and penalises paid add-ons for basics. - **Value** compares the renewal price, not the intro price, against the measured performance and features. - **Ease of use** is a hands-on rating of the control panel, onboarding and the number of steps to a working WordPress site. Sub-scores are set by the author, reviewed by a second team member, and every change is dated in the review’s history. ## How often is everything updated? - Monitoring runs continuously. Monthly averages are added to the data files at the start of each month. - Scores are reviewed after each full quarter of data. - Prices and plan features are re-checked at least every 45 days. If a check is older than that, the review shows a warning next to the pricing table. - Each host gets a manual re-review (support test, dashboard changes, new features) every quarter. ## What we do not test - Hosting types we do not buy: dedicated servers, enterprise plans, Windows hosting. - Traffic beyond 100 concurrent users. Our load test is a sanity check, not a stress test. - Sites in regions other than the test server’s region, beyond the quarterly global TTFB sample. - Anything we cannot repeat. If a result cannot be reproduced, it does not go in a score. ## Corrections If a host believes a number is wrong, they can write to us with evidence. We re-run the test, publish the new result with the date and keep the old one in the review history. We do not remove data on request. ## Money HostIdea is operated by DM Research Labs, LLC. Every host, including the ones we feature, is assessed with the same method, on the same schedule and with the same weights, by the same people. Our only income from hosts is affiliate commissions, and commissions never change a score; the weights above are the only inputs. Full details on the [disclosure page](https://hostidea.com/disclosure).