Nobody Clicks Anymore

The exception was Contrastly, the photography site I launched in 2012. Contrastly had real SEO traction through its life and watching it ride through algorithm changes over the years taught me more about search than any course could have.

That experience made me pay attention to the moves, but I didn't apply the same discipline to my other properties until recently.

For DailyPhotoTips specifically, the newsletter was the whole thing for years. I launched it in 2019 and ran it as an email-first property, without much of a blog attached.

I only started writing content on the DailyPhotoTips site last year, which meant walking into SEO with a fresh domain in a mature vertical, which is already hard before you factor in what's happened to search since.

Then I started paying real attention, digging into Search Console data, studying what was ranking and why, reading everything from the SEO industry about how to do this properly, and thinking about content through that lens for the first time as an active discipline rather than a passive byproduct.

That was when I noticed the ground shifting under me in ways that made the entire practice feel stranger than anything I'd seen in twenty years of publishing online.

The Zero-Click Era

The numbers from 2025 are hard to process even when you've been watching the trend for a while.

Zero-click searches rose from 56% to 69% between May 2024 and May 2025, according to Similarweb data. Chartbeat tracked organic Google traffic across more than 2,500 news sites and found it down 33% globally in 2025 and 38% in the U.S.

Ahrefs ran an analysis across 300,000 keywords comparing December 2023 to December 2025 and found that position-one click-through rates for informational queries dropped from 7.6% to 3.9% on pages without AI Overviews, and from 7.3% to 1.6% on pages with them.

That last number is the one that matters most for independent publishers.

A 58% reduction in CTR on pages that trigger AI Overviews, controlling for general trends.

You can be number one in Google for a query that matters to your business and watch most of the traffic never arrive at your site, because Google answered the question for the user before they had a reason to click.

Chegg reported a 49% decline in non-subscriber traffic in 2025 and cited AI Overviews in an antitrust lawsuit arguing that Google trained the feature on educational content that it now competes with directly.

Daily Mail publisher DMG Media reported click-through rate drops of up to 89% on certain queries. The Reuters Institute surveyed 280 media executives across 51 countries in late 2025 and found that publishers expect search referrals to fall by 43% over the next three years.

The exact stuff that lives in every indie publisher's archive.

The HCU Scar Tissue

Before AI Overviews made the situation visibly worse, there was the Helpful Content Update.

Google rolled out the HCU in September 2023 with stated goals about rewarding content written for people rather than search engines. What actually happened is that a lot of independent publishers got destroyed.

HouseFresh, Charleston Crafted, and hundreds of other expert-driven niche sites reported traffic losses between 50 and 90 percent. Reddit, meanwhile, surged 1,300% in search visibility between July 2023 and April 2024 according to SISTRIX.

The pattern was consistent across verticals.

Google stopped rewarding niche expert content and started rewarding user forums, major brands, and aggregators.

Some sites have partially recovered through subsequent core updates. Most haven't, because the algorithmic evaluations became sticky in ways that made recovery unusually hard.

The takeaway is that the trust Google was willing to extend to independent publishers got reduced structurally well before AI Overviews arrived.

The current AI-driven traffic collapse is sitting on top of a publisher ecosystem that was already wounded.

The Playbook I Walked Into

When I finally started paying attention to SEO as a real practice, the playbook I encountered was the one the industry had been refining for years.

Write comprehensive guides, structure them with clear headings, earn backlinks from relevant sites in your niche, keep content updated, focus on search intent, build topical authority over time.

Slow but reliable, the advice went. You could plan a year ahead because you knew what would happen if you published consistently.

For DailyPhotoTips that meant leaning into detailed tutorials on camera settings, composition techniques, post-processing workflows, and gear recommendations.

I started seeing rankings move, traffic picking up on specific posts, and the feedback loop of SEO starting to feel tangible in a way it hadn't before.

That feedback loop is already breaking.

New content still ranks sometimes, but the rank matters less because the click rate has collapsed on queries where AI Overviews intercept the answer before the user has any reason to visit the page.

What SEO Became

SEO splintered into five overlapping disciplines with different optimization targets, and anyone still running the 2020 playbook in 2026 is optimizing for a game that's only part of the board.

• Traditional SEO still matters but generates much less traffic per rank than it did in the past.
• GEO is the new discipline for getting your content cited by AI systems.
• Community signals are about being visible in the user-generated spaces that Google and AI engines increasingly trust.
• Brand search is the safe harbour because nobody's AI Overview eats the query "DailyPhotoTips newsletter." When someone types your brand name, they want to land on your site rather than read a synthesized summary of what your site is about.
• And distribution diversification is the recognition that being dependent on Google for your traffic is now a strategic vulnerability rather than a minor risk.

Any independent publisher who only does the first one is going to watch their traffic curve flatten through 2026.

Writing for Machines Again

The ironic part of GEO is that it's a reversal of the advice we were all given for the last decade.

From about 2015 onward the SEO industry repeated a single piece of guidance, which was to prioritize readers over algorithms by ditching the robotic keyword stuffing, writing in a natural voice, using conversational language, and structuring content around how real readers think rather than how bots parse.

That advice aligned neatly with Google's direction and with the instincts of good writers, and it produced a generation of content that was measurably better to read than what came before.

Now we're back to writing for machines, though these machines are LLMs that synthesize answers by pulling from multiple sources, which means the goal is being one of the sources they cite rather than being the page the user clicks.

The structural implications of this are similar to the old machine-writing era.

Research on what gets cited by AI engines keeps pointing at the same structural elements, including definitional sentences in the first paragraph that follow patterns like "[Entity] is a [category] that [differentiator]," clear heading hierarchy that makes extractable sections, tables for comparisons, numbered lists for processes, factual density per paragraph rather than long discursive setup, and schema markup for machine readability.

A CMU study from 2024 found structured markup was among the top-five features correlated with higher LLM citation rates.

All the structural discipline that felt robotic a few years ago is now the stuff that gets you surfaced by Perplexity or cited in an AI Overview.

The conversational essay style that won the last decade is worse for GEO, even though it's still better for brand and reader relationship building. Which means the new writing practice involves producing content that works on both fronts.

Quotable structure for the machines, real voice for the humans.

The Dashboard Problem

The other part of this that's quietly breaking is measurement.

I've been building SEO Dash to help track what actually matters across my properties, and the building process has been a lesson in how much the old toolkit assumes.

Ahrefs, Moz, Semrush, and Google Search Console were designed for a world where ranking and traffic were tightly coupled.

I mean, rank went up, traffic went up. Rank went down, traffic went down. The metric was the proxy.

That proxy has broken.

You can watch a page climb from position five to position one in Search Console and see the clicks go down because an AI Overview appeared above your result.

You can lose a query in Google entirely but pick up Perplexity citations that drive traffic you're not tracking anywhere. You can get mentioned in an AI Overview and never appear in the citations list because the overview summarized your page without linking to it.

The metrics that matter now are a different set, including AI citation rates across the major LLMs, zero-click impression counts, share of voice in AI answers for your category queries, Reddit and Quora mention frequency, brand search volume relative to non-branded queries, and return visitor percentage as a proxy for actual audience building.

Most of the existing tools are starting to track some of these, but slowly and unevenly.

Building something in-house to see what's happening across your specific properties is part of the adaptation now, not because the existing tools are bad but because they were built for a different game.

What Still Works

If you're running an independent site in 2026, the practices that still produce results are less glamorous but more honest than the 2020 playbook.

Go narrow on a subject where your actual expertise is obvious, because generic content is what AI Overviews handle best while specialized content with a clear point of view is harder to summarize and more likely to get cited as a source.

Build an audience you control through email lists, direct visitors, podcast subscribers, and community members who aren't subject to Google's traffic decisions.

Then create content that's quotable by using clear structure, concrete facts, definitional openers, and extractable sections.

You also have to participate in communities where your audience actually gathers, because Reddit matters now for reasons beyond SEO, and your brand gets mentioned and your expertise gets recognized in ways that feed back into how LLMs characterize you.

Own the brand search so that people know to type your name directly into Google or ChatGPT, which builds something that survives the zero-click era.

Diversify distribution aggressively because any single channel can get cut in half in eighteen months, as Google has just demonstrated.

Measure what matters instead of what's easy, since rankings are comfortable to track because the tools are mature, while traffic, conversions, and audience growth are what actually matter to your business.

The Long View

I've been publishing online since 2005, been writing about design, dev, and UI/UX since, and also running photography content since 2012. In that time the core skill has never changed. Make stuff worth coming back to.

Every SEO era has rewarded slightly different tactics, but the underlying question has stayed the same.

Can you produce work that real people want to read, trust enough to share, and remember well enough to return to? The tactics on top of that foundation shift every few years, sometimes dramatically, but the foundation itself is the same as it was in 2012 or 2008 or 1998.

The AI-driven traffic collapse of 2025 and 2026 is real, and it's hitting publishers harder than most of the previous shifts.

But treating it as the end of SEO misses what's actually happening.

SEO became SEO plus GEO plus community plus brand plus distribution, and any publisher who treats all five as parts of the same job will be fine. The ones who keep optimizing only for Google rankings while the click rate disappears will not.

I avoided the "SEO is dead" framing for this piece because that take misses what's actually happening.

SEO is different now, and the reality is weirder than a death.

Google is still the largest traffic source for most independent publishers, and ranking still matters.

Rank without clicks is becoming the default state for a growing share of queries, and the work of being found has spread across more surfaces than any single dashboard can capture right now.

Anyone who's been publishing independently for a decade has seen harder shifts than this one. The work changes, the tools catch up eventually, and the people who keep making good things keep getting found.

That part has never actually changed.