Hosting comparison tool
Compare audited hosting companies on prices, staging, backups, migrations and support, two head to head or up to five in a matrix, every cell cited and dated from captured vendor pages, with the gaps the vendors do not publish listed instead of filled in.
Comparisons on Kite do not pick a winner; they state fit from cited, dated facts. Choose two audited hosts for a head to head with fit notes, or up to five for a side-by-side matrix, all from the feature audit with the gaps the vendors do not publish listed alongside.
By Sam Moreton
Researched
Calculator
Noindexed until the audit covers 6 hosts (currently 17); the dataset has reached that bar.
Assumptions
- Entry-price rule: the cheapest published plan of each host, trials excluded, is stated with both prices; the cheaper side is named as a budget fit, and equal prices are stated as equal.
- Comparable-spec rule: when both hosts publish RAM-graded plans, the table prices the cheapest plan of each host that reaches the larger of the two hosts' smallest plan RAM. When either host grades plans without RAM, no comparable-spec row is rendered, because no common class exists; a fallback cell would be a false equivalence.
- Trial rule: a trial that needs no credit card is stated as a test-before-commit fit for that host.
- Custom-pricing rule: a host quoting custom pricing only cannot have a comparable-spec price computed for it, and the tool says so.
- Metering rule: a host publishing unlimited or unmetered visits has that scheme stated; nothing is inferred about billing beyond the captured pages.
- Support rule: paid escalation tiers above the standard channel are stated per host with their prices.
- Benchmarks rule: benchmarks Kite has not personally run appear as not tested, never as modelled numbers.
- Gaps rule: a dimension the captured pages do not state renders as not stated, never filled in.
- No-winner rule: verdicts state fit per host from published facts only; no overall winner is named, because no testing overlap exists between hosts Kite has not benchmarked.
- Matrix rule: the shortlist view repeats the same cells for up to five hosts, one column each, with no cross-host ranking added; a matrix cell says exactly what the head-to-head cell for the same dimension says.
- Price-order rule: sorting a shortlist by entry price works only when every priced host in it publishes in the same currency; a mixed selection disables price order and says why, because ranking across currencies would need an exchange rate this audit does not hold.
- Filter rule: filters narrow the picker from structured audit facts (plan kinds, trial terms, stated features); they never change the data or rank anything.
Formula vv0.2 / last reviewed 2026-11-04
Most hosting comparisons are marketing pages wearing a table. This one is built from captured vendor pages, one audited file per host, with every cell carrying the date its source was checked. Pick two hosts and you get prices, staging, backups, migrations, support and scaling side by side, plus a list of what the vendors do not publish. Down to a longer shortlist? Line up to five hosts in one matrix, narrow the picker with filters built from the same audit, and open any pair as a head to head when you want the fit notes.
Who this is for
Anyone down to a shortlist and tired of tab-switching between feature pages. It suits questions with published answers: does staging come free, what happens at renewal, who migrates the sites for you. It is the wrong tool for performance questions, and it says so instead of pretending.
How to read the table
Prices are snapshots, with the check date on each cell. Introductory rates sit next to their renewals where the pages state them, because the renewal is the price you actually live with. The cells that compare like with like are computed by declared rules, listed under the tool, and where no honest comparison exists the table says so instead of inventing one.
Fit notes sit under the table; there is no winner row. No host has been benchmarked by this site under matching conditions, so an overall winner would be fiction, and the benchmark row reads not tested until that changes.
What the dataset holds, and what it does not
Feature facts are solid ground. As of October 2026, the captures show Cloudways including staging and cloning on every plan and migrating the first site free, with offsite backup storage at $0.033 per GB per server and Cloudflare Enterprise from $4.99 per domain per month. SiteGround ties staging to its GrowBig tier and up, with daily backups on every plan and free site migrations, and no free trial, only a 30-day money-back guarantee. What a capture cannot tell you is what any of this feels like at 3am, or anything the vendor chose not to publish. Those gaps are listed in the table, and the gaps are themselves information: backup retention was not stated on either host's captured pages.
When to trust it, and when to ask around
Trust it for feature and price facts, within the check date on each cell. Hosting pages change often, so confirm the number that matters most on the vendor's own page before you commit. For speed, support quality and behaviour on a bad day, read the hands-on material: the Cloudways review is written from a real monthly bill, and written head-to-heads like Cloudways vs SiteGround carry the reasoning a table cannot.
Where the data goes next
The dataset grows host by host, each new host audited under the same citation rules, so the pair you want may already be in the picker. When it is not, the comparison articles cover the pairs people actually ask about.
FAQ
Why does the tool not declare a winner?
A winner needs testing overlap: the same workload measured on both hosts by the same method. No such runs exist yet, so the tool states fit from published facts and keeps the benchmark row at not tested.
Where does the data come from?
Every cell comes from the per-host feature audit: one cited file per host, each field carrying the date its source page was captured. Vendor pages say what vendors say; the tool dates every cell so you can judge recency, and lists what the captured pages do not state instead of guessing.
Which hosts can I compare?
The picker lists every host audited into the dataset so far, each captured under the same citation rules. The set grows host by host, and the written head-to-heads cover the pairs readers ask about most.
What is the difference between head to head and the shortlist?
Head to head compares two hosts and adds the fit notes the verdict rules derive from cited data. The shortlist matrix lines up to five hosts, one column each, so you can scan a dimension across the set; it adds no ranking, and sorting by entry price only works when every priced host publishes in the same currency.