Website data is not one thing. It arrives from three different places, with three different levels of reliability, and no error recurs here more often than quoting one as though it were another. A registration date and a monthly-visits estimate are both website data, but one is a fact recorded by the party responsible for it and the other is a projection from a sample. Treating them alike produces confident nonsense.
The whois layer: what a registry publishes
The first layer is the registry. When a domain is registered, the registry records a small, fixed set of facts about it, and publishes most of them. That publication is the whois record, and its modern structured form is the RDAP response. Between them they carry the creation date, the date the record was last modified, the expiry date, the sponsoring registrar, a set of status codes and the nameservers the domain is delegated to.
This layer is authoritative in the narrowest possible sense: it is correct about what it covers and silent about everything else. A whois record will tell you, beyond argument, that a domain was created on a particular day. It will not tell you whether anything has ever been published on it. The most useful field is also the least discussed: the last-changed date, which moves whenever the registration is materially touched, and which is therefore the field that betrays a domain that has quietly changed hands.
DNS: how a name becomes an address
The second layer is resolution. A registration is an entry in a ledger; it does not, by itself, make anything reachable. That job belongs to DNS, and specifically to the DNS records published in the domain's zone. Seven types account for nearly everything anyone needs to read: A and AAAA for addresses, NS for delegation, MX for mail, TXT for verification and policy, CNAME for aliases, and SOA for the zone's own parameters.
The layer above them is delegation, and it produces the single most misread state in this whole subject. A domain can be registered, paid for and entirely valid while having no nameservers at all. It then resolves to nothing, and most tools report it exactly as they report a name that was never registered. Reading that as "available" is wrong, and it is wrong in a way that costs money.
Which records are authoritative and which are estimates
The third layer is measurement, and nothing in it is published by the site. Traffic figures, authority scores and safety ratings are all produced by third parties who never see the server logs. Traffic estimation works from panels and clickstreams, extrapolated by models that each vendor keeps to itself, which is why two reputable tools routinely disagree about the same site by a factor of five.
Authority metrics, DR, DA, AS, PA, are logarithmic scores computed over link graphs that no two vendors crawl identically. They are useful for comparison and worthless as absolutes. Backlinks and referring domains are the raw material underneath them, and the distinction that matters most there is the one people skip: the most valuable citations on the web, the ones from reference works, pass no ranking signal at all.
Website data that nobody publishes any more
The awkward part of any honest guide to website data is how much of the measurement layer has closed. Alexa Rank ran from 1996 to 2022 and was quoted in press releases, pitch decks and academic papers as though it were a census; it was a toolbar panel. Compete closed in 2016. Yahoo Site Explorer, the last free public backlink index, closed in 2011. SiteAdvisor and Web of Trust, which between them defined site reputation for a decade, are both gone.
That matters for reading old sources. A citation to an Alexa rank is not wrong, it is unverifiable, and the difference is worth stating, which is why how this catalog is written makes it a rule rather than a preference. The dated report on closed measurement tools collects the whole list. And the last page in this section, robots.txt and crawl limits, explains why the kind of open bulk querying that fed those tools is no longer something a catalogue can simply do.
