How popular is this domain?

Alexa is gone, and nothing single replaced it

For twenty years "what is this site's Alexa rank" was the one number everybody quoted. Amazon retired the service in May 2022, and the habit outlived it: rank badges still sit on old pages, and stale copies of the last list are still served by sites that never noticed it stopped moving. A ranking that answers with a number but has not changed in years is the worst kind of source — it looks exactly like a working one.

What replaced it is not one thing but several, each measuring something different and none of them measuring what most people assume. That is the part worth understanding before you trust any number, including the one above.

Three questions that all get called "popularity"

The free public rankings fall into three families, and mixing them up is the single most common mistake:

None of these is "traffic". Actual visit counts — sessions, bounce rate, time on page — come from panels and clickstream data, they are commercial products, and there is no free equivalent that is even close. Anyone showing you free "monthly visits" for an arbitrary domain is modelling, not measuring, and usually will not say so.

Why the numbers disagree, and why that is the useful part

When two rankings put the same domain hundreds of thousands of places apart, the instinct is to decide one of them is broken. Usually neither is. The gap is information:

• High authority, low usage. Many sites link to it, few people go there. Typical of long-established reference pages, standards documents and organisations that publish rarely but get cited forever.
• High usage, low authority. Enormous query volume, almost nobody links. Typical of infrastructure: content delivery hosts, telemetry endpoints, app back-ends. These are not websites in any sense a reader would recognise, but they are certainly popular.
• Agreement across the board. Usually a genuinely mainstream site — read by people, linked by others, and looked up constantly.

This is why a single blended score, shown alone, quietly destroys the most interesting thing in the data. A domain scoring 55 because both signals say 55 and a domain scoring 55 because one says 90 and the other says 20 are not the same domain.

What a rank cannot tell you

Ranks are ordinal, and the distance between places is wildly uneven. The gap between position 1 and position 100 is orders of magnitude of real traffic. The gap between position 800,000 and position 900,000 is close to noise — those two domains may differ by a handful of lookups on the day the list was generated. Any tool that reports movement in the deep tail as if it were meaningful is reporting sampling jitter.

Nor does a rank say anything about whether a site is trustworthy, legitimate or safe. Popularity and reputation are unrelated axes: a widely abused domain can be extremely popular, and a careful, valuable site can be invisible to every list here. If the question is reputation, the answer is a blocklist check and the registry record, not a position in a popularity table.

Absence is not evidence

The lists together cover on the order of a million and a half registered domains. The domain name system holds hundreds of millions. So "not in the list" is the normal state for the overwhelming majority of perfectly real, perfectly working websites — including almost every small business, personal site and internal service in existence.

Read absence as "no public measurement exists", never as "no traffic". Those are different claims, and only the first one is supported by anything.

Registered domains, not hostnames

Popularity is measured per registered domain, which is why www.example.com, blog.example.com and example.com all answer as one entry. Working out where the registered part starts is not a matter of taking the last two labels: bbc.co.uk is a domain while co.uk is a public suffix, and there are thousands of such rules that cannot be guessed from the shape of the name. Getting this wrong produces confident nonsense — a "domain" called co.uk holding the rank of the busiest British site.

One consequence worth knowing: sites hosted on a shared subdomain of a hosting provider are folded into that provider. A blog on a free platform contributes to the platform's rank and has none of its own, because from the outside there is no separate registered domain to measure.

Reading any popularity number honestly

1. Ask what it counts. Queries, links and visits are three different worlds.
2. Ask when it was built. These lists regenerate daily; a copy from last year is not a slightly old answer, it is a different one.
3. Ask how many sources agree. One source averaged with nothing is still one source.
4. Distrust precision in the tail. Position 4 is a fact. Position 743,912 is a rounding artefact with a number printed on it.

Where these numbers come from

Three public lists, each free to download without an account, each measuring something different:

• Tranco — A research ranking that averages several other lists over 30 days, specifically to be hard to manipulate. research list · https://tranco-list.eu/
• Majestic Million — Ranks by how many separate networks link to the domain — authority, not traffic. CC BY 3.0 · https://majestic.com/reports/majestic-million
• Cisco Umbrella — Ranks by DNS queries seen across a large resolver network — closer to raw usage than to human visits. free download · https://s3-us-west-1.amazonaws.com/umbrella-static/index.html

Rankings are ordinal positions published by their operators; the combined position shown above is our own arithmetic over them and is not endorsed by any of them.