The Great Decoupling: Why the "Click" is a Legacy Metric in 2026

Our next guest post, from Lou Peck, CEO The International Bunch, who debates why we should no longer be obsessed with clicks. An audio version is provided and can be accessed below.

The Answer Economy is shattering the logic in which we have optimised for the ‘click’ being the final destination. AI summaries and ‘Zero-Click’ search are decoupling research impact from website traffic, feeding a lethal ‘Crocodile Effect’ for traditional metrics. We need to move beyond being a destination and become the verified foundation of global discovery.

Alternative Text: Unhappy face on a yellow sticky note with a performance graph in the background, going down as the measured performance drops.

Licensed by The International Bunch from AdobeStock

Discover the audio version read by guest author Lou

For decades, we have lived by a simple, comfortable playbook: publish research, make it discoverable, measure usage and evidence impact. We focused on economies of scale, optimising for ‘the click,’ and treating our websites and content platforms as the final destination. Did you know that AI searches do not result in a measured “click,” are not counted, and are called “zero-click” or “zero-data” events? This means massive amounts of usage and impact are simply not being reported.

Today, the technological landscape is akin to the “Wild Wild West”. We have shifted from making people want things (push) to making things people want AND making them easy for machines to find (pull). Quantity over quality has been a core strategic focus for submissions and traffic on sites, coupled with optimising the human experience, and measuring how many people visit landing pages, whether our intended audiences, the accidental visitor or average user. Over time, we have tied our entire value proposition, including subscription renewals, to metrics that can no longer be used in isolation, like HTML views and full text downloads.

We are seeing structural discontinuity, moving from traditional searching methods to synthesis. With the launch of Google Scholar Labs in November 2025, Google Scholar is set for a major shift in 2026. Pivoting from a library index of links to a synthesis engine, it will prioritise a paper’s ability to answer a question over its decades-old citation count. What does this even mean? It no longer just finds research; it reads it, summarises it into a single-line AI takeaway, and offers researchers an instant answer that bypasses the need to ever click through.

These changes confirm a new reality, that we are currently being swallowed by what has been termed “The Crocodile Effect” or “Crocodile Gap”, popularised in tech and SEO by Rand Fishkin. This is the lethal, widening gap between our content’s visibility and our actual site traffic, where the crocodile is literally eating our value. The top jaw is impressions, and the bottom jaw is the click-through rate. Charlie Rapple explores this concept in depth in her Scholarly Kitchen article Responding to the Threat of Zero-Click Search and AI Summaries: How Do We Tame The Crocodile? As AI-generated summaries satisfy user intent directly on the search results page, the jaws of the crocodile are snapping shut on our traditional usage data. The first click now is often a zero-click.

So the logic of discovery has been fundamentally rewired, and if you are still measuring success by “sessions”, you are not just behind, you’re invisible in a landscape full to the brim of noise. The Wild Wild West of AI traffic is now; there’s no getting away from it.

The Data of Disruption

So what happens if you resist the Wild West? In a recent Silverchair panel discussion, Understanding AI Traffic - Bots, Crawlers, and What They Mean for Your Platform, Paul Gee gave an interesting example of what happens when you turn off bots from accessing your platform, including the impact and lessons learned.

Let’s put into context the evidence from the last 12 months of a “Great Decoupling” between research impact and website traffic:

• The search collapse - the predicted 25% decline in traditional search volume has begun as users migrate from search and scroll to answer and explore with answer engines

• The death of the fold - for high-intent research queries, 83% now trigger an AI Overview and these AI summaries average 1,200 pixels in height. That means that organic links are generally pushed off the first screen you land on in the search page - this part of the page is known as “above the fold”

• The zero-click baseline - 60% to 80% of searches end without a single click. If an AI tool is providing the answer with your data, the reader may have achieved their goal without knowing it was your brand, causing brand dilution

• The machine surge - global internet traffic rose 19% this year, but it wasn’t us humans driving the spike. AI training bots and high-speed crawlers now represent a massive share of activity, capable of requesting one million pages in under 60 seconds during peak bursts

The Shift From Destination to Foundation

Websites are no longer destinations; they are seen more as places to verify information and, hopefully, be recognised as trusted sources and achieve inference impact. Researchers are using tools to do their research and writing, like Agentic Browsers, e.g., Atlas, MultiOn, Skyglass, etc. or tools like ChatGPT, Gemini, Claude, You.com, so many to mention, we all have our favourites. The researcher only comes to your platform when they need to validate a reference or citation, read further in depth, and verify the truth, ensuring integrity in their work - that is, if they don’t use their tool to do that for them. If you are chasing average user clicks, you need to stop and start ensuring you have the market share of trust. If AI slop ranks higher than peer-reviewed content, we have failed in our mission.

Update your marketing strategy to build trust around your technology and content:

1. Abandon traffic as a primary KPI. If 80% of searches are zero-click, your traffic will go down. That doesn’t mean your impact has. Change how you measure your impact and engagement in online environments. Inference citation frequency is about how often your content is the source of truth when an AI tool is looking for an authoritative answer; this should become your new focus.

2. Defend your trustworthiness. If the AI tools don’t trust you, you aren’t a source of truth, and they will discard you. Defend your credibility and ensure your measured trust score is nurtured and improved. Be careful when using tools like Google Disavow Tool to ensure you are cutting off only those suspect and spam sites (SEMRush usually lists these for you via a traffic light system, red, orange and green). These sites link to your online presence and proactively manage your experience, expertise, authoritativeness, and trustworthiness (EEAT) signals.

3. Building a bilingual platform doesn’t just mean providing text language translations. This means ensuring better accessibility and inclusivity for all. Avoid putting off AI tools with limited discoverability, but what about people…humans…with disabilities and differing abilities, like being neurodiverse, or simply supporting more established learning styles everyone has through visual, auditory and kinesthetic ways. Use technology to add in ways to improve how people digest information through podcasts, videos of performances and experiments, additional resources, audio, author-led audio and video content, 4D, etc. The list goes on, but it’s more about improving the type of access to content that people can have and not focusing on HTML and full text PDF downloads. Align more with an institution’s goals for improving accessibility for their student, staff and researcher communities. A content platform or website can be beautiful to those who see it and its functionality, but are you presenting your reader, whether human or AI, with great metadata and easy navigation? If not, then you may find you won’t exist in the answer economy we find ourselves in now

4. Define usage standards fit for purpose. Many business models are tied to legacy usage data, which no longer measure what we need or tell the whole story. Many stakeholders, such as libraries, use this data to make informed decisions, about content usage and renewals. COUNTER and NISO’s collaboration for COUNTER 5.1 Access_Method: Agent standard goes some way toward demonstrating value, based on how often data has informed an AI search and response, rather than on how many PDFs were clicked. Adopt better ways of reporting on data and informing stakeholders about how things are changing and what can help them make more informed decisions

5. Optimise for AI tools and think more about generative engine optimisation (GEO) rather than just search engine optimisation (SEO). As Stephanie Lovegrove Hansen and Kristina Henrikson argue in their Scholarly Kitchen article SEO still matters, it has evolved into a multi-channel effort. You must be as discoverable in a ChatGPT prompt, for example, as you are in a library catalogue. Ensure your content is trusted, review and improve your trusted score with small wins and bigger long-term goals. Identify what is actually having a negative impact that you can easily change. As an example, Google Scholar Labs synthesises information to rank papers based on their semantic relevance to answering a question, rather than just keyword matching. There will be big pushes and big budgets behind these tools as core drivers in supporting current behaviour and engagement.

Move beyond the click

Your current trajectory may mean you are diluting your brand, becoming invisible in a competitive landscape, and inadvertently devaluing your content before it is even published. But there is hope, just make a start, no matter how small. And if you have made a start, what can you do next? One thing we commonly see is trying to force people to be where you want them to be, instead of being where they are (and they want you to be, there are places they don’t want you, especially on social media).

Be authentic, trusted and credible. Are the metrics you are using to measure, and more importantly, be measured by, really still relevant? What if you could tell a more transparent, compelling story by not chasing the clicks, but being the most authoritative source in a person’s or an AI tool’s mind? It’s quality over quantity, and that is a real mindshift change.

Wouldn’t you rather have one AI bot correctly identifying a research study and citing it to 1,000 real researchers than have 1,000 average user visits to a landing page that does nothing for your goals and objectives? Stop chasing the phantom of the click and start anchoring trust. Legacy depends on being the verified foundation on which the future is built. Start with a small adjustment that could have a profound long-term impact.

As with all of our articles, this post reflects the views of the author and we hope will stimulate some discussion and conversation around big questions in scholarly communications.