Marginal notes
Methodology: where every number comes from
Every figure in a report comes from a public source or from a simple formula published on this page. Here is what we check, how and where, and what the limits are.
Turning your input into a domain
You can paste a domain or any link. We keep only the hostname, convert internationalised names to their ASCII “xn--” form (Punycode, using the IDNA 2008/UTS #46 rules), and work out the registrable domain with the ICANN section of the Public Suffix List, which we store on our server and refresh periodically. So https://shop.example.co.uk/basket becomes example.co.uk. A few public suffixes are also registered names with their own websites — gov.uk, for example — and these can be surveyed as names in their parent registry. We don’t accept IP addresses, and we never use any path, port or login details you paste.
Registration
We look up the registry’s RDAP service (the modern replacement for WHOIS) using IANA’s RDAP bootstrap file, which we keep a copy of. For .uk names this is Nominet’s RDAP service. From the answer we keep only non-personal fields: the registration, expiry and last-changed dates, the registrar’s name and IANA ID, the status codes, nameservers and whether the delegation is DNSSEC-signed. Domain age is the time since the registration date in the current record — if a domain lapsed and was registered again, the date restarts. A “not found” answer means the registry has no record, which usually means the name is unregistered. Some country-code registries don’t offer RDAP; for those we say so rather than guessing.
DNS and email
Our server asks its own DNS resolver for the A, AAAA, NS, MX, TXT, CAA and SOA records of the domain, the A/AAAA/CNAME records of www, and the TXT record at _dmarc. An SPF record is a TXT record starting v=spf1; we count the mechanisms that cost a DNS lookup at the top level (SPF allows ten in total, including nested includes, which we don’t expand). DMARC is a TXT record starting v=DMARC1; we show its policy. “IPv6” means the domain or its www name has an AAAA record. “DNSSEC validation” reports whether our validating resolver marked the answers as authenticated. DKIM can’t be checked without knowing the sender’s selector names, so we don’t show it.
Hosting network
We take the first IP address of the website and look up which network announces it using Team Cymru’s IP-to-ASN mapping over DNS. The country shown is where the network is registered according to the regional internet registry, not necessarily where the server is. The CDN or platform hint comes from tell-tale response headers (for example cf-ray for Cloudflare).
Website, robots.txt and sitemap
Our lookup bot (user-agent domainstatistics.co.uk lookup bot (+https://domainstatistics.co.uk/about/)) requests the home page over HTTPS and plain HTTP, following at most five redirects. It only connects to public internet addresses on ports 80 and 443. We read up to 1.5 MB of the HTML, extract the title, meta description, language, canonical link, robots directives, generator tag, Open Graph tags and some markup fingerprints for the CMS hint, and then discard the page — we never store page content. Response time is the time to first byte for the final page, measured once from our server in the UK. Page weight is the size of the HTML alone. We also check whether /robots.txt and /sitemap.xml exist.
SSL certificate and security headers
We connect to port 443 of the final website host and read its certificate: issuer, validity dates, the names it covers (Subject Alternative Names), whether it matches the host name, and whether it verifies against the standard set of trusted root certificates. Security headers (HSTS, Content-Security-Policy, X-Content-Type-Options, frame protection, Referrer-Policy and Permissions-Policy) are read from the HTTPS home page response.
Archive history
The first and latest capture come from the Internet Archive’s Wayback Machine availability API (asking for the capture closest to 1 January 1996 gives the earliest one). The year-by-year chart comes from the Wayback CDX server: we ask for the home page’s captures collapsed to one per month, and plot how many months each year have at least one capture (0–12). That measures how continuously a site has been archived, not how many times. The CDX server is often slow, so on the website we wait at most 8 seconds; if it doesn’t answer, the chart is added on a later re-check.
Search engines
Neither Google nor Bing offers a free way to check whether a site is indexed: Bing’s Search APIs were retired on 11 August 2025, and Google’s Custom Search JSON API is closed to new customers and due to be retired on 1 January 2027. We never scrape search results pages. So, unless the site owner has connected a paid search API in our settings, we don’t check this automatically — instead the report has buttons that open a site: search on Google or Bing for you. Even then, a site: count is only a sample. We also check whether the domain appears in the latest Common Crawl index, an open web archive.
Popularity
We use the Tranco list, a research ranking of the top million registrable domains created by Victor Le Pochat, Tom Van Goethem, Samaneh Tajalizadehkhoob, Maciej Korczyński and Wouter Joosen (“Tranco: A Research-Oriented Top Sites Ranking Hardened Against Manipulation”, NDSS 2019). It averages several sources over 30 days: the Chrome User Experience Report, Cloudflare Radar, Farsight, Majestic and Cisco Umbrella. The copy in use is list Y83KG, downloaded on 4 Oct 2026. A domain that isn’t in the list is not necessarily unvisited — it is simply outside the top million.
Traffic estimate
Rank lists measure relative popularity, not visitor numbers. To give a feel for scale we map the rank to a deliberately wide band of monthly visits. These bands are our own rough assumptions — they are not measured and can be wrong by a large factor. If a domain isn’t in the top million we say it is too small to estimate reliably rather than inventing a number.
| Tranco rank up to | Monthly visits (rough) |
|---|---|
| #100 | 50 million – 5 billion |
| #1,000 | 5 million – 500 million |
| #10,000 | 500,000 – 50 million |
| #100,000 | 50,000 – 5 million |
| #1,000,000 | 2,000 – 500,000 |
Worth estimate
The worth figure is a points score turned into a price band. It is for interest only — not a valuation: real prices depend on buyers, trademarks, backlinks, sales history and timing, none of which a formula can see. Points are added for each factor below (current weights), then the total is matched to a band.
| Factor | Points |
|---|---|
| Name of 1–3 characters | 20 |
| Name of 4 characters | 16 |
| Name of 5 characters | 13 |
| Name of 6 characters | 10 |
| Name of 7–8 characters | 7 |
| Name of 9–12 characters | 4 |
| Name of 13–16 characters | 1 |
| Ending .com | 15 |
| Ending .co.uk | 12 |
| Ending .uk | 9 |
| Ending .net or .org | 8 |
| Ending .io, .ai or .co | 6 |
| Any other ending | 3 |
| Name is one dictionary word | 15 |
| Name is two dictionary words | 8 |
| Pronounceable (letters only, 4–12, balanced vowels) | 4 |
| Each hyphen (max two counted) | -6 |
| Contains digits | -5 |
| Internationalised (non-ASCII) name | -3 |
| Per year since registration | 1 |
| Maximum points for age | 15 |
| Has Wayback Machine captures | 3 |
| First capture 10+ years ago | 5 |
| Tranco rank 1–1,000 | 30 |
| Tranco rank 1,001–10,000 | 22 |
| Tranco rank 10,001–100,000 | 14 |
| Tranco rank 100,001–1,000,000 | 7 |
| Website loads (status 2xx) | 4 |
| Valid HTTPS certificate | 2 |
Only one length row, one ending row and one of “dictionary word”, “two dictionary words” or “pronounceable” applies. Dictionary words are matched against an English word list derived from SCOWL (Kevin Atkinson’s spell-checker word lists).
| Points from | Band |
|---|---|
| 0 | £0 – £50 |
| 20 | £50 – £250 |
| 35 | £250 – £1,000 |
| 50 | £1,000 – £5,000 |
| 65 | £5,000 – £25,000 |
| 80 | £25,000 – £100,000 |
| 92 | £100,000 – £500,000 |
Other endings
For the same name with other common endings we ask each registry’s RDAP service whether it has a record. “Registered” means it does; “apparently available” means it answered “not found”. Registries can reserve, block or charge premium prices for names, so always confirm with a registrar.
Caching, limits and politeness
- Results are stored and reused for 24 hours; after that anyone can press “Re-check”.
- Checks run one at a time, each limited to about 8 seconds, and no more than a few lookups run at once across the whole site.
- To keep the service fair there are limits on how many new domains one visitor can survey per hour and per day (we store a salted hash of your IP address for 48 hours to do this).
- We identify ourselves with our user-agent, cache aggressively, and pause requests to any service that tells us to slow down.
- Data is provided as-is from third parties and may be out of date or wrong. See our terms.