Did 90% of the Top 50 US News Sites Lose Traffic in September, and What Does It Mean for Where Coverage Gets Read?

Update, October 18, 2026: CNN is launching a website paywall in the UK and Canada on October 20, 2026. See what it changes for coverage.

What Do the September Figures Show?

Press Gazette ranks the 50 biggest US news websites each month using Similarweb data. For September 2026, 45 of them were down on a year earlier, which Press Gazette notes is “a higher proportion than 82% in August” (Press Gazette, US list). The largest falls were at Newsweek, the Daily Mail, Axios and CNN. Newsweek also had the biggest month-on-month rise, up 32% on August, which shows how far the two comparisons can diverge. Month on month, the Sun fell 17% to 17.8 million visits and NBC News 16% to 51.8 million.

The chart shows that large and small outlets fell by similar proportions, with CNN’s 48% drop matching Axios’s despite a site more than ten times the size, while the few growers include a newsletter platform and an international broadcaster; it does not show why.

Press Gazette says the Daily Mail’s figures “may be impacted by its April domain switch from .co.uk to .com,” and that Newsweek, the Daily Mail and Axios have all made job cuts this year. Newsweek’s chief executive told Adweek in July that the company was on track to pass $100 million in revenue despite the readership decline, Press Gazette reports, which is a reminder that site visits and business health are different measures. For the ways publishers are reframing audience, see Reach’s shift to engaged time.

Is the Fall the Same Outside the US?

Broadly, yes. On Press Gazette’s list of the 50 biggest English-language news sites in the world, half saw traffic fall by more than a fifth in September compared with a year earlier, and seven grew (Press Gazette, world list). Newsweek again led the declines, down 62% to 35.8 million visits, followed by the Hindustan Times (down 56% to 76.9 million) and News 18 (down 53% to 66.4 million). Euronews nearly doubled, up 99% to 44.8 million, and Substack rose 37% to 172.2 million. The two lists are built differently, which is why the same outlet shows different numbers: Newsweek has 30.7 million visits on the US list and 35.8 million on the world list.

How Reliable Is Similarweb Data?

It is modeled. Press Gazette describes Similarweb’s method as applying “machine learning and modelling to the statistically representative datasets that the company collects,” from sites that share first-party analytics, contributory device-data networks, partnerships and public data extraction, and says it refines Similarweb’s news category “to exclude some sites with a less journalistic focus” (Press Gazette, US list). Similarweb’s own data page lists the same four kinds of input and says its data science converts “noisy online data into a clear signal” (Similarweb). That makes the rankings useful for direction and relative size, and weaker for any single site’s exact count. A publisher’s audited or first-party figures, where they exist, are better evidence; see what audited circulation checks.

What Does the Data Say About the Cause?

Very little. The tables report what happened, not why. Search is the most-cited explanation in the trade press. Future’s chief executive told staff in a memo, reported on October 5, that cuts were “primarily driven by the decline in search” (Press Gazette, Oct. 5); we covered that in what happens to a pitch when a site loses its editors. Pew’s 2025 browsing study found that users who saw a Google AI summary clicked a traditional result in 8% of visits, against 15% for those who did not, and clicked a link inside the summary in 1% (Pew Research Center).

Google disputes the broad reading. In August 2025 it said “total organic click volume from Google Search to websites has been relatively stable year-over-year,” that third-party reports of “dramatic declines in aggregate traffic” were often based on “flawed methodologies,” and that traffic is shifting between sites, with forums, videos and first-person content gaining (Google). Both can be true at once: aggregate clicks stable, and large news sites losing share. For the evidence on AI answers, see whether AI Overviews reduce clicks and how much traffic AI assistants refer.

Does Less Site Traffic Mean Less Valuable Coverage?

Not automatically. A placement’s value depends on who reads it, what it says, whether it is linked, syndicated or cited, and what it is for, such as credibility with a buyer or an investor. Site visits are one input, and Similarweb’s site tables do not capture readers reached through newsletters, apps, social video or syndication. A smaller site audience also does not change an outlet’s editorial standards, which are much of what makes coverage worth having; see what makes an outlet tier one and whether tier-one media features are still worth it. What it does change is the arithmetic of reach: a 48% drop in visits at a large site is a large drop in the people who might see any one article there; see what happens to a placement’s traffic after the first week.

What This Means for Anyone Buying a Placement

These are our recommendations, not Press Gazette’s findings. First, ask for an outlet’s current audience figures and their date; a number from two years ago may overstate what a piece will reach now. Second, a paid or sponsored article is advertising and must be labeled, and an advertorial on a well-known site does not become editorial coverage because the site is famous; see how paid content must be labelled. Third, do not judge a placement by one outlet’s site visits; weigh newsletter, social and syndication reach, and measure what the coverage does for your own site; see how to measure whether a placement worked. Fourth, no agency or outlet can promise a number of readers.

Methodology and Limitations

Four limits apply. First, all traffic numbers are Similarweb estimates as presented by Press Gazette, not audited or first-party figures, and we did not query Similarweb. Second, Press Gazette chooses which sites to include and removes some, so the “top 50” is its list. Third, the Daily Mail result may reflect its domain change. Fourth, these are September figures published October 8; October’s ranking will replace them, and year-on-year changes say nothing about the quality of any one article.

Conclusion

If Similarweb’s estimates are right, nine in ten of the biggest US news websites reached fewer visitors in September than a year earlier. A buyer should ask what a placement will reach now, in writing, rather than assume a masthead’s past audience.

Frequently Asked Questions

Sources

This report is built from two Press Gazette rankings and one Press Gazette news report, Similarweb's data page, a Pew Research Center study and a Google statement, all read on October 18, 2026. Advice is our recommendation. We did not query Similarweb or any publisher's analytics.

1. trade press Press Gazette, Top 50 US news websites: 90% see traffic drop in September, October 8, 2026. Supports the US figures, the chart, the methodology description and the Daily Mail note. Read October 18, 2026 (archived copy of October 16).
2. trade press Press Gazette, Top 50 English-language news sites in the world: Half see declines of 20% or more in September, October 8, 2026. Supports the world-list figures. Read October 18, 2026.
3. trade press Press Gazette (Charlotte Tobitt), Fresh redundancies at Future as tech brands hurt by AI Overviews, October 5, 2026. Supports the Future memo quote. Read October 18, 2026.
4. vendor Similarweb, Similarweb's Data: How We Measure Digital Traffic. Supports the description of Similarweb's data sources and modeling. Read October 18, 2026.
5. survey research Pew Research Center, Google users are less likely to click on links when an AI summary appears in the results, July 22, 2025. Supports the 8%, 15% and 1% click figures. Read October 18, 2026.
6. platform documentation Google, AI in Search is driving more queries and higher quality clicks, August 6, 2025. Supports Google's statement on organic click volume. Read October 18, 2026.