Home › Methodology

Methodology: how providers are compared and ranked

Where the pricing data comes from, exactly how the recommendation is scored, why renewal price is treated as the headline figure, and an honest account of where this data is weakest.

Last reviewed 2026-07-29

Where the data comes from

Every price and specification is transcribed by hand from the provider's own public pricing page and, where the pricing page is silent, their terms of service or knowledge base. Nothing is scraped. No provider supplies data, and no provider sees an entry before it is published.

For each provider the following are recorded:

• Entry price — the lowest advertised monthly rate on the entry-level plan.
• Entry price note — the contract term and prepayment that rate requires. A rate needing 36 months upfront is a materially different offer from the same rate on a monthly term, and quoting the first without the second is the single most common distortion in hosting comparison.
• Renewal price — what the plan costs once the introductory term ends.
• Specifications — storage, bandwidth, sites allowed, email, SSL, free first-year domain.
• Features and drawbacks — a short list each, with drawbacks drawn from documented limitations and widely reported problems rather than individual complaints.

Why renewal price is the headline

Introductory hosting pricing is a loss leader with an expiry date. A plan advertised at $1.99/month commonly renews between $9.99 and $15.99 — a 250–360% increase — and because the introductory rate usually requires paying three years upfront, most people do not encounter the real price until they have been a customer for three years.

This is legal, disclosed somewhere in the terms, and near-universal in shared hosting. It is also the thing you most need to know before buying, which is why it sits in the comparison table rather than a footnote. The renewal price guide works through how to calculate what a plan actually costs across a realistic ownership period.

How the recommendation works

The quiz asks four questions: what you are hosting, expected monthly visits, what matters most (cheapest / balanced / performance), and your technical comfort. It then does two things in sequence.

Step 1: pick a category

A short set of explicit rules, in order:

• Static site → static hosting.
• Backup or bulk file storage → VPS/cloud, with a caveat shown alongside the result. VPS is the closest fit among these four categories because of expandable block and object storage, but dedicated object storage is almost always the better answer — see what hosting is not for.
• Web app or SaaS with a custom backend → VPS/cloud.
• WordPress → managed WordPress, unless you chose "cheapest" and are not expecting 100,000+ visits, in which case shared.
• Online store → VPS if you are a developer; shared if you want cheapest at low or medium traffic; otherwise managed WordPress, since most small stores run WooCommerce.
• Blog or brochure site → VPS if you are a developer who chose performance; managed WordPress if you chose performance at high traffic; otherwise shared.

Step 2: rank providers within that category

Each provider gets a score:

score  = max(0, 20 - entryPrice) × priceWeight
               + featureCount × featureWeight
               - min(renewalPrice / entryPrice, 10) × renewalWeight
               - drawbackCount × 1.5
               (+ storageGB × 2, only if you said "backup / bulk storage")

The renewal-jump penalty is deliberately light when you have said you want the cheapest possible option and heavy otherwise. Someone optimising purely for today's price has effectively told us they will move when it renews; someone choosing "balanced" almost certainly has not thought about renewal at all, and should be steered away from a provider that triples.

The logic is a small, readable rule set rather than a learned model, so any recommendation can be explained. It lives in src/utils/finder.js if you want to read it directly.

Currency conversion

All prices are researched and stored in US dollars. Selecting another currency applies an approximate conversion for display.

This is not a claim about what a provider will bill you. Some localise checkout pricing by region and some charge in USD regardless; local taxes such as VAT are not included; and the conversion rate is periodic, not live. Use converted figures for rough comparison and read the actual price at checkout.

Currency is defaulted from an approximate country lookup: first a small function on Netlify's edge network, which reads the country from the IP address every web request already carries and returns only a country code; and if that is unavailable, your browser's time zone, resolved entirely on your device. Neither uses a third-party geolocation service and no IP address is stored. If neither identifies a supported country, prices stay in USD. Your manual selection always takes precedence. See the privacy policy.

Known limitations

Roughly in order of how much they could mislead you.

1. Renewal prices are the least reliable figure here

They are also the most important, which is an uncomfortable combination. Renewal rates are frequently not published on the pricing page at all — they appear in the terms, at checkout, or only on the invoice a year later. They vary by plan, by region, by promotion, and by what you can negotiate on the phone. Treat renewal figures as a well-researched indication of the order of magnitude, not a quotation.

2. Hosting promotions change constantly

Faster than almost any other software category. Providers run seasonal promotions, regional pricing, and different rates depending on the page you arrive from. A figure correct at the review date can be wrong within weeks.

3. There is no performance testing

No speed or uptime benchmarks are run. Meaningful hosting benchmarks need sustained multi-region testing under realistic load; a single measurement from one location is worse than none, because it looks authoritative while measuring mostly the distance between the tester and the server.

4. Support quality is not first-hand

Where support reputation appears it summarises aggregate public sentiment. Review platforms in this category are noisy — heavily gamed, skewed toward complaints, and covering millions of customers on varying plans. Directional only.

5. The quiz is four questions

Real requirements have more dimensions than that: compliance, data residency, specific software versions, existing team skills, tooling you already use. The quiz gets you to a sensible category and shortlist. It cannot capture a genuinely unusual requirement, and it does not try to.

6. Feature counts are a crude proxy

Scoring counts features, and a feature list is a marketing artefact. Ten trivial features outscore three important ones. This is a real weakness of the ranking, mitigated somewhat by the drawback penalty, and it is why the table shows the actual feature lists rather than only a score.

Corrections

Manually maintained pricing data will eventually contain a mistake. If you spot one, please report it with a link to the provider's page. An actual renewal invoice is the most valuable evidence anyone can send, since that figure is the hardest to research and the easiest to get wrong.

Methodology changes that materially affect recommendations are noted here along with the review date.