The most visited websites in the US, and what AI does to the count

“The most visited websites in the US” sounds like one number per site that someone counts. It is not. Four companies publish a list, and each one counts something else: panel visits, visitors and pageviews, search clicks, or DNS lookups. Past Google and YouTube, they split.

The rulers also share a blind spot that gets bigger every month. Each one counts a person loading a web page. More and more reading now happens inside an AI assistant, which fetches the page, writes the answer, and sends nobody to the site.

The 26 sites the three visit counts agree on

These 26 domains are in the top 50 of Similarweb, Semrush and Ahrefs. That makes them the safest answer to the question. They are ordered by their middle rank of the three. Cloudflare is left out because it does not count visits.

#	Domain	Similarweb	Semrush	Ahrefs
1	google.com	1	1	5
2	youtube.com	2	2	2
3	reddit.com	5	3	1
4	facebook.com	3	4	3
5	amazon.com	4	5	7
6	yahoo.com	6	6	12
7	instagram.com	8	7	6
8	wikipedia.org	11	8	4
9	chatgpt.com	9	9	16
10	x.com	10	10	10
11	tiktok.com	16	14	14
12	espn.com	14	25	11
13	walmart.com	15	16	23
14	linkedin.com	13	19	18
15	nytimes.com	19	24	19
16	pinterest.com	21	23	9
17	ebay.com	17	22	24
18	microsoft.com	30	28	22
19	netflix.com	26	29	36
20	apple.com	34	30	8
21	zillow.com	25	34	33
22	target.com	33	41	29
23	etsy.com	32	40	35
24	homedepot.com	43	39	32
25	roblox.com	50	37	41
26	cnn.com	49	47	49

Checked on 12 October 2026. The periods differ: Similarweb is September 2026, Semrush is August 2026, Ahrefs is its October 2026 crawl. Every source is listed under the article.

The consensus list has a hole in it. Semrush ranks pornhub.com 12th and xvideos.com 13th, above TikTok, LinkedIn and the New York Times. Similarweb has four adult sites between 40 and 44. They are not in the table above because two of the four rulers leave them out. That says more about the rulers than about the sites.

The four top 50s, side by side

Rank	Similarweb	Semrush	Ahrefs	Cloudflare Radar
1	google.com	google.com	reddit.com	google.com
2	youtube.com	youtube.com	youtube.com	microsoft.com
3	facebook.com	reddit.com	facebook.com	googleapis.com
4	amazon.com	facebook.com	wikipedia.org	amazonaws.com
5	reddit.com	amazon.com	google.com	apple.com
6	yahoo.com	yahoo.com	instagram.com	gstatic.com
7	bing.com	instagram.com	amazon.com	facebook.com
8	instagram.com	wikipedia.org	apple.com	akamai.net
9	chatgpt.com	chatgpt.com	pinterest.com	cloudfront.net
10	x.com	x.com	x.com	akadns.net
11	wikipedia.org	duckduckgo.com	espn.com	akamaiedge.net
12	cloud.microsoft	pornhub.com	yahoo.com	aaplimg.com
13	linkedin.com	xvideos.com	yelp.com	azure.com
14	espn.com	tiktok.com	tiktok.com	ui.com
15	walmart.com	bing.com	imdb.com	googlevideo.com
16	tiktok.com	walmart.com	chatgpt.com	fbcdn.net
17	ebay.com	temu.com	spotify.com	windowsupdate.com
18	instructure.com	weather.com	linkedin.com	icloud.com
19	nytimes.com	linkedin.com	nytimes.com	cloudflare.com
20	weather.com	usps.com	fandom.com	lencr.org
21	pinterest.com	paypal.com	nih.gov	instagram.com
22	temu.com	ebay.com	microsoft.com	apple-dns.net
23	usps.com	pinterest.com	walmart.com	office.com
24	gemini.google.com	nytimes.com	ebay.com	cloudflare-dns.com
25	zillow.com	espn.com	clevelandclinic.org	amazon.com
26	netflix.com	twitch.tv	indeed.com	sharepoint.com
27	fandom.com	imdb.com	mayoclinic.org	googlesyndication.com
28	duckduckgo.com	microsoft.com	quora.com	doubleclick.net
29	live.com	netflix.com	target.com	cloud.microsoft
30	microsoft.com	apple.com	tripadvisor.com	ntp.org
31	twitch.tv	xhamster.com	canva.com	youtube.com
32	etsy.com	discord.com	homedepot.com	trafficmanager.net
33	target.com	live.com	zillow.com	tiktokcdn-us.com
34	apple.com	zillow.com	merriam-webster.com	ubuntu.com
35	claude.ai	spotify.com	etsy.com	adobe.com
36	zoom.us	msn.com	netflix.com	fastly.net
37	shop.app	roblox.com	foxnews.com	live.com
38	duosecurity.com	indeed.com	nfl.com	windows.net
39	paypal.com	homedepot.com	mlb.com	grammarly.io
40	pornhub.com	etsy.com	justanswer.com	office.net
41	xvideos.com	target.com	roblox.com	launchdarkly.com
42	xhamster.com	instructure.com	lowes.com	grammarly.com
43	homedepot.com	claude.ai	adobe.com	appsflyersdk.com
44	onlyfans.com	github.com	allrecipes.com	netflix.com
45	chase.com	whatsapp.com	primevideo.com	adobe.io
46	nextdoor.com	onlyfans.com	clever.com	edgekey.net
47	discord.com	cnn.com	fifa.com	amazonvideo.com
48	foxnews.com	canva.com	kahoot.com	akamaized.net
49	cnn.com	quora.com	cnn.com	dropbox.com
50	roblox.com	fedex.com	cvs.com	bing.com

What each list counts

	Data period	What it counts	Where the data comes from	Adult sites in its top 50
Similarweb	September 2026	Estimated monthly unique visitors plus pageviews, desktop and mobile web	Direct measurement, a contributory network of consumer products, partnerships including ISPs, public data	4 (ranks 40–44)
Semrush	August 2026	Visits, which Semrush defines as sessions	A panel of 200M+ users and hundreds of clickstream providers	4 (ranks 12, 13, 31, 46)
Ahrefs	October 2026 crawl	Estimated organic search visits only	Keyword rankings, search volumes and a click-through rate model	None in its top 100
Cloudflare Radar	12 October 2026	Estimated number of users whose devices look the domain up	Queries to the 1.1.1.1 DNS resolver	None in its top 100

Similarweb and Cloudflare print only a rank. A rank with no number cannot show that site 2 is half of site 1, or a fifth.

Cloudflare counts DNS lookups, not visits

Cloudflare’s list is the odd one out, and it is worth being blunt about why. It ranks domains by how many users’ devices ask the 1.1.1.1 resolver for them. Cloudflare says itself that it does not tell a website from an infrastructure system, and that servers, cloud infrastructure, IoT devices, home routers and bots all use the resolver.

So its top 50 is mostly names no person types. googleapis.com is 3, amazonaws.com is 4, akamai.net is 8. 38 of its top 50 are on none of the three visit lists. They are CDNs, cloud APIs, ad servers, and update and certificate services that apps and phones call in the background, many times a day, whether anyone is looking at a screen or not. reddit.com is number 1 on Ahrefs and is not in Cloudflare’s top 100.

Read Cloudflare’s list as a map of what the internet’s plumbing talks to. It is not a list of where people spend their attention.

The adult sites: one ruler that hides them

AhrefsTop’s list has no adult site in its top 100. The page says only that it ranks sites “by Ahrefs organic search traffic estimates”. So we asked Ahrefs’ own Site Explorer the same question for the US on 14 October 2026:

Domain	Ahrefs Site Explorer, US organic visits a month	Place on AhrefsTop
reddit.com	487.0M	1
pornhub.com	136.4M	not listed
xvideos.com	80.4M	not listed
nytimes.com	77.2M	19
chatgpt.com	41.4M	16
cvs.com	12.4M	50

The two sources do not line up exactly. AhrefsTop shows 858.2M for reddit.com, where Site Explorer shows 487.0M for the US, so the list is built on a different cut of the data. But the gap is far too big to be rounding. On Site Explorer, pornhub.com has nearly twice the search traffic of nytimes.com and eleven times that of cvs.com, AhrefsTop’s number 50. The adult sites are not missing for lack of search traffic. The list leaves them out, and does not say so.

Cloudflare’s absence is less clear. Its method post describes no category filter. On a DNS ruler, a site people visit competes with domains that every device calls constantly, which may be enough to push it out. Cloudflare does not say, and we could not test it from outside.

The lesson is general. A ranking is a product with an editor, and an editor makes choices. Check what a list leaves out before you quote what it puts in.

What AI does to the count

All four rulers were built for a web where reading meant a person loading a page. That web is shrinking in three ways, and none of the lists shows it.

1. Fewer clicks reach the page. Pew tracked the browsing of 900 US adults in March 2025. When Google showed an AI summary, people clicked a normal result in 8% of visits, against 15% without one. They clicked a link inside the summary in 1% of visits. They ended their session on 26% of pages with a summary, against 16% without. Ahrefs and Semrush estimate search traffic from rankings and a click-through rate model. A model that predates the summary will credit a page with clicks the summary now keeps.

2. The answer is read somewhere else. chatgpt.com is number 9 on Similarweb and on Semrush, with 1.01B visits in August. Ahrefs puts it at only 41.4M organic visits, because people go there directly, not from a search. The visit counts once, to chatgpt.com. The sites whose pages built the answer get nothing in any of these lists. And Similarweb builds its rank from web traffic, desktop and mobile, so time spent in the ChatGPT, Claude or Gemini apps does not feed it.

3. Machines do the reading. Cloudflare measured AI crawlers in 2025. Training drove nearly 80% of AI bot activity. In July 2025, Anthropic’s crawlers fetched about 38,000 pages for every visit they referred, OpenAI’s 1,091 and Perplexity’s 194.8. Google’s fetched 5.4. A crawler’s fetch is not a visit in any panel. It still costs bandwidth, and it is where the reading now starts.

Put together, the most visited list is turning into a list of where people start: Google, YouTube, Reddit, ChatGPT. The sites they used to land on are read more than ever, by machines, and visited less.

One publisher, six years apart

Semrush published a US list for September 2020. Put it next to August 2026 and it looks like a collapse:

	September 2020	August 2026
google.com	30.9B visits	24.54B visits
facebook.com	8.4B visits	2.37B visits
craigslist.org	number 13	number 61

The list cannot show how much of that change is real. Part of it can be the ruler: Semrush says its updates to how it processes traffic data “may cause some shifts in metrics”. A trend line that spans a method change measures two things at once, and the chart cannot tell you which.

What to do with these numbers

1. Name the ruler every time. “reddit.com: 2.57B visits (Semrush, August 2026)” is a number you can check. “Reddit gets 2.57B visits” is not.
2. Compare sites on one list, never across lists. A competitor’s Similarweb rank and your Ahrefs traffic are two different measurements. Put them in one chart and the chart is wrong.
3. Check what the list leaves out. AhrefsTop drops adult sites, Cloudflare counts machines, and no list sees an app. A missing site is a fact about the list until proven otherwise.
4. Discount organic estimates where AI answers show. On queries with an AI summary, a click model built on the old results page is too high. Your own search console data is the check.

Even for your own site, no outside count will match your analytics. Semrush calls a session a visit, and your analytics tool counts sessions too. So why do the two counts never match? A panel scales a sample up, and a session is not something a tool sees. It is a rule each tool applies to draw the boundaries of a visit.