SEO
Most of the fifty biggest news sites are down year on year. The causes are not specific to publishing, which is why the trend is worth reading closely.
Press Gazette publishes a monthly ranking of the most-visited news websites, one for the world and one for the United States. It is a useful thing to read every month, and through 2026 it has been reading the same way: down.
That is interesting for publishers. It should be interesting to everyone else too, because none of the mechanisms behind it are specific to news. If your site has a resources section, a guides library or a blog that brings in search traffic, the same forces are pointed at you - you are just further down the queue.
The picture in the figures published through August 2026, which draw on Similarweb estimates:
The distribution matters more than any single number. This is not a sector-wide tide going out at an even rate - it is a small number of sites growing strongly while the majority lose a quarter, a third or half of their audience.
Before drawing conclusions from any of this, three honest caveats.
These are estimates, not measurements. Similarweb models traffic from panel data, device data and public sources. It is directionally reliable at this scale and it is the best public comparison available, but it is not the publisher's own analytics. Treat a 45% decline as "a large fall" rather than as a precise figure.
Technical changes distort individual lines. The Daily Mail moved from a .co.uk domain to .com during the period, which splits and reshapes its measured traffic independently of whether anyone stopped reading it. Any big year-on-year move for a single site deserves a check for a migration, a paywall change or a redirect before it is treated as audience loss.
Comparisons run into event cycles. US political sites surged through the November 2024 election and fell back afterwards. A chunk of the 2026 year-on-year decline for those titles is a comparison against an unusually high base, not a new problem.
The useful reading is the aggregate, not the leaderboard. Individual rankings move for individual reasons. Thirty-nine out of fifty moving the same direction is a pattern, and that is the part worth acting on.
Publishers quoted in the coverage point at the same short list, and it matches what we see in client search data.
AI answers absorbing informational clicks. AI Overviews and AI Mode resolve a growing share of "what is", "how does" and "X vs Y" queries inside the results page. The query still happens. The impression still registers. The click does not. News sites are unusually exposed here because explainer content is a large share of what they publish.
Core algorithm updates. Google's May 2026 core update was named repeatedly, as earlier updates were before it. Core updates redistribute rather than destroy traffic, but if you are on the wrong side of one the effect on your own numbers is indistinguishable from the sector declining.
Social referral has largely gone. The decline in referral traffic from Facebook and X to news sites has been running for years and is now close to complete for most publishers. Traffic that used to arrive from a feed does not reappear somewhere else.
These stack. A site that built its audience on explainer content, ranked through a core update it later lost, and used to be topped up by social referral has had three separate channels contract at once.
News is the canary, not the special case. It sits at the extreme end of a distribution every content-driven site is somewhere on: a very high proportion of informational, top-of-funnel, zero-intent queries.
The thing to understand is that the mechanism is intent-based, not sector-based. Query types are affected roughly in this order:
So the question is not whether your industry is affected. It is what share of your current traffic comes from queries in the first group - and whether that traffic was ever producing anything.
This is an afternoon of work in Search Console and analytics, and it is worth doing before changing anything.
The common outcome: a third to a half of sessions come from pages that have never produced an enquiry, and those are exactly the pages losing traffic. That is not a crisis. It is a vanity metric correcting itself.
Five things, in the order we would actually do them.
1. Change what you report on. If sessions are the headline number in your monthly report, that number is going to fall for reasons outside your control, and it will drive bad decisions. Report qualified sessions, enquiries and pipeline. A site with 40% fewer sessions and 20% more enquiries has had a good year.
2. Move effort to content that cannot be answered in a panel. Original pricing data, comparisons drawn on your own project history, case evidence with real numbers, calculators and tools, anything requiring a judgement a model cannot make on your behalf. Our own pricing page exists for this reason: it answers a question most competitors will not answer publicly, so the answer has to come from us.
3. Earn the citation. When an AI answer names a source, that is visibility even without the click. Clear structure, question-shaped headings, self-contained sections and specific, quantified claims all make a passage easier to retrieve and cite. We wrote about the specifics in getting cited in AI search.
4. Build channels you own. Email lists, a returning direct audience, client relationships. Substack growing while the rest of the list falls is not a coincidence - it is what a direct relationship with a reader looks like when intermediated discovery stops working.
5. Do not panic-rewrite the site. The instinct after reading a chart like this is a wholesale content overhaul. Most of the traffic being lost was not converting. Work out which of it mattered, protect that, and let the rest go.
The uncomfortable version: a decade of SEO advice optimised for a metric - sessions - that is now decoupling from revenue. The sites that come through this are the ones that were already measuring the right thing, and they will barely notice.
Press Gazette's monthly rankings are worth a standing look, whatever industry you are in. The aggregate direction is the most honest available preview of where organic discovery is heading, and it arrives about a year before it shows up in your own numbers.
Need a hand with this? Orange Orbits builds and markets websites for businesses across the United States. Book a free 45-minute scoping call and we will give you a straight answer about your specific situation.
Keep reading
An update stops new attacks. It does nothing about files an attacker already wrote to your server. If your site was patched late, here is how to find out whether someone got in first.
Read articleA critical flaw in the best-selling Avada theme lets an attacker take over a WordPress site with no login and no click. The fix has been out for weeks. Many sites still have not applied it.
Read articleThe third WordPress security release in a month fixes seven vulnerabilities, from stored XSS to SQL injection. Here is what each one means, and which sites should care most.
Read articleBook a 45-minute call. We will ask about your goals, your constraints and your timeline, then send a written recommendation with real numbers. If we are not the right fit, we will tell you who is.
