If you came here looking for Alexa Rank, the short answer is that it no longer exists. Amazon retired Alexa.com on 1 May 2022, and the announcement was plain about it:
“Twenty-five years ago, we founded Alexa Internet. After two decades of helping you find, reach, and convert your digital audience, we’ve made the difficult decision to retire Alexa.com on May 1, 2022.”
That notice was posted on 22 December 2021. It is worth knowing how this article verified it, because it makes a point of its own: support.alexa.com no longer resolves at all. The announcement of the shutdown has itself been shut down, and the quotation above had to be recovered from an archived copy captured on 19 April 2022. When a company retires a service, it usually retires the explanation too.
So the useful version of this article is not “how to use Alexa.” It is: what Alexa Rank actually measured, why its disappearance matters less than you’d think, and what to use instead.
What Alexa Rank actually was
Alexa Rank was a single number claiming to order every website on the internet by popularity. If your site ranked 1,345, the implication was that 1,344 sites got more traffic.
The number came almost entirely from people who had installed the Alexa toolbar or browser extension and agreed to be tracked. That is worth sitting with for a moment. A ranking presented as a measurement of the entire web was extrapolated from a self-selected sample of people who chose to install a traffic-ranking extension — a group that skews heavily toward webmasters, SEO professionals and marketers, and away from everyone else.
Write-ups at the time routinely cited the extension’s reach — a few hundred thousand browsers — as evidence of scale, then met the obvious objection with some version of “it’s better than nothing.”
That was the wrong answer, and it is the most useful thing to correct here. A confident number from a biased sample is often worse than nothing, because “nothing” invites you to go and find real data, while a precise-looking rank invites you to stop. Alexa Rank looked like a measurement. It was an estimate built on a convenience sample, and it was routinely quoted in pitch decks, “top site” lists and competitive analyses as though it were a census.
First, a name collision worth clearing up
If you search for “Alexa” today you will mostly find the voice assistant in Amazon’s smart speakers. That is a completely different product that happens to share a name.
Alexa Internet was founded in 1996 and acquired by Amazon in 1999. It was a web-traffic and competitive-analysis company, and it is the one that shut down in 2022. Amazon Alexa, the voice assistant, launched in 2014 and is very much alive.
The overlap is confusing enough that it caused real mistakes: articles about the ranking service were illustrated with pictures of speakers, and vice versa. If you are researching website measurement and your results are full of smart-home tips, this is why — add “rank” or “traffic” to your search, or skip straight to the tools below.
The distinction that outlives the tool
Here is the idea worth taking from all of this, and it applies to every analytics product you will ever use:
Measuring your own website and estimating someone else’s are completely different activities. One is counting. The other is modelling.
When you measure your own site, you have access to the ground truth: your server saw the request, your analytics script ran, your search console recorded the impression. The numbers can still be wrong — bots, ad blockers, misconfigured tags — but they are derived from events that actually happened on your property.
When you estimate someone else’s site, you have none of that. You have a panel, or clickstream data bought from third parties, or search-visibility models, and you multiply upward. The output is a plausible range wearing the costume of a precise figure.
Alexa blurred the two deliberately: the same dashboard showed you your own verified metrics and an estimate of your competitor’s, in the same typeface, with the same number of digits. Most tools still do this. Learning to notice which half you’re looking at is the actual skill.
Measuring your own site: use real data
For your own website, three free sources cover almost everything a small site needs.
Google Search Console is the one people underuse. It reports impressions, clicks, average position and the actual queries people typed — data no third party can estimate for you, because only Google has it. If you want to know whether your site is working, this is the primary instrument. It reports on rolling windows, which is why judging a change any faster than about 28 days is mostly noise.
An analytics tool — Google Analytics, or a privacy-respecting alternative like Plausible, Fathom or Matomo — tells you what visitors did once they arrived: which pages, how long, where they came from, where they left.
The Chrome User Experience Report (CrUX) is the one almost nobody knows about. It publishes real performance data collected from actual Chrome users, so you can see how fast your pages genuinely are for real people on real connections, rather than how fast a test harness says they should be. The easiest way in is PageSpeed Insights, which shows CrUX field data alongside its lab score.
Together these answer the questions Alexa’s paid tier was sold for — which pages earn attention, which queries you rank for, what needs fixing — from data about your own site rather than a model of it.
Estimating someone else’s site: know what you’re buying
Sometimes you genuinely need an outside view: sizing a competitor, checking whether a site is worth pitching, judging whether a domain is as big as it claims. A few options, roughly from most transparent to most commercial.
Tranco is the most interesting, and it is free. It is a top-one-million ranking built by academic researchers specifically to fix the problems Alexa had. Its own description of why it exists is worth quoting, because it is effectively a diagnosis of Alexa Rank:
“we showed that these rankings disagree on which domains are most popular, can change significantly on a daily basis and can be manipulated (by malicious actors)”
Their paper is titled Tranco: A Research-Oriented Top Sites Ranking Hardened Against Manipulation, and the list is maintained by researchers at KU Leuven, TU Delft and Université Grenoble Alpes. Rather than relying on one panel, it combines several independent rankings — including Cisco Umbrella, Majestic, the Chrome User Experience Report and Cloudflare Radar — averaged over 30 days to damp out daily noise. Every list is archived with a permanent citation so a study can be reproduced later.
Tranco’s own site notes, matter-of-factly, that “The Alexa ranking has been removed from the default Tranco list, as it is no longer available.” The research community’s replacement for Alexa was not a better panel. It was a reproducible method.
Cloudflare Radar publishes free domain rankings derived from the enormous volume of DNS traffic Cloudflare’s resolver handles. Different vantage point, different biases, also free.
Similarweb, Semrush and Ahrefs are the commercial options, and they are genuinely useful — particularly Semrush and Ahrefs for keyword and backlink work. Just hold their traffic numbers loosely. They are estimates from panels and clickstream data, and when you can compare them against a site’s real analytics, they are often off by a wide margin.
How to read any traffic estimate
Four questions, and they work on every tool in the previous section.
1. Where does the data come from? A browser panel, a DNS resolver, purchased clickstream, or the platform’s own logs? Each has a characteristic bias. Panels over-represent whoever installs panels. DNS data misses traffic that never hits that resolver.
2. Is this counted or modelled? If the tool cannot see the events, it is modelling them. That is not disqualifying — just don’t quote a modelled number to two significant figures.
3. Can it be gamed? Tranco’s authors built their list specifically because earlier rankings “can be manipulated.” Any ranking that people benefit from appearing in will attract attempts to manipulate it.
4. Would two tools agree? Check the same site in two. When they disagree — and they will, sometimes by multiples — that disagreement is the real finding. Report the range, don’t average it into false precision.
A worked example: sizing up a competitor in 2026
Say you want to know whether a competitor is genuinely bigger than you. Here is the sequence that replaces “check their Alexa Rank,” and it takes about ten minutes.
Start with whether they rank at all. Look them up in Tranco’s top million. Presence or absence is a coarse but honest signal: a site outside the top million is not a mass-traffic destination, whatever its marketing implies. Cross-check in Cloudflare Radar, which uses a different data source entirely.
Then look at what they rank for, not how much. This is the step people skip, and it is the one that actually informs decisions. Search a handful of the queries you care about and see whether they appear. Ten pages ranking for terms your buyers search is worth more than a hundred thousand visits from somewhere irrelevant — and you can observe it directly, for free, with no tool at all.
Then check their real performance. If their domain has enough Chrome traffic to appear in the Chrome User Experience Report, you can see genuine speed data for their pages — real users, not a lab test. Run their URL through PageSpeed Insights and it will show the CrUX field data alongside the lab score.
Only then reach for an estimate, from Similarweb, Semrush or Ahrefs, and treat the traffic figure as an order of magnitude rather than a number. If two of them disagree by 3×, you have learned that nobody outside that company knows, which is itself useful.
Finally, ask what you would do differently if the estimate were double or half. If the answer is “nothing,” you did not need the number. That question retires most competitive-analysis busywork, and it is the discipline Alexa’s dashboard actively discouraged by making a precise-looking figure the first thing you saw.
What to actually track
If you want a short answer to “how should I measure my website,” it has not changed much, and it never involved a single vanity number:
- Impressions and clicks from Search Console, on a 28-day window, by query and by page. This tells you what the world is asking for and whether you are being chosen.
- Which pages earn engaged time, from your analytics. A page nobody finishes is a page that needs work regardless of its rank.
- Real-user performance, from CrUX or your analytics. Slow pages lose readers before they read anything.
- Direction over absolute value. A rank of 1,345 told you nothing actionable. “These four pages gained impressions and lost clicks” tells you to rewrite four titles.
The lesson Alexa Rank leaves behind is not that measurement is futile. It is that a number’s authority comes from its method, not its precision — and the method is usually one click away, in documentation nobody reads. Alexa’s method was a browser panel. Amazon eventually decided that was not worth maintaining, and the research community replaced it with something that shows its work.
Frequently asked questions
Is Alexa Rank gone for good? Yes. Amazon retired Alexa.com on 1 May 2022 and the service was not sold on or revived. Any site still displaying an “Alexa Rank” is showing you a number that stopped being updated years ago.
What is the closest free replacement? For a ranking, Tranco or Cloudflare Radar. Neither is a drop-in substitute, because neither pretends to be a single authoritative popularity score — which is the point.
What replaced Alexa’s paid features? The keyword research, content research and competitive analysis that Alexa’s subscription covered are now served by Semrush, Ahrefs and Similarweb. For your own site, Google Search Console does the most important part of it for free.
Did my site’s Alexa Rank mean anything? It meant something narrow: roughly where your traffic sat relative to other sites, as seen by people who had installed Alexa’s extension. It was never a measure of revenue, engagement quality, or how well you served the people who did arrive.
What should I check instead, if I only have ten minutes a week? Search Console, on the queries and pages tab, comparing the last 28 days with the previous 28. That single habit will tell you more than Alexa Rank ever did.
Is a lower number always better? With Alexa Rank, lower meant more traffic — rank 2 was Facebook. But direction mattered more than position, and neither told you whether the traffic was worth having. Ranking is not a goal; it is a symptom.
More on measurement is collected under measuring what matters. You may also find measuring network performance and bandwidth, explained useful.