.png)
We know AI can generate stuff.
Words, images, video, and code. We have all been impressed with how seemingly human much of its output is. But while the industry wrings its hands over the "drowning" of genuine insight in a sea of synthetic content, a far more structural shift is taking place beneath the surface.
There is a not-so-quiet problem drawing the attention of clients and agencies alike. It isn't just about how to swim through more content; it is about how to be found at all.
The tide of non-human “users”
To understand the scale of what is happening, look at the infrastructure. Cloudflare CEO Matthew Prince recently shared a statistic at Cannes Lions that should stop every digital strategist in their tracks.
As of June, Google sends roughly 18 bots to crawl a site for every human visitor it refers. That was the old deal: you allow them to crawl your site, they send you traffic. But the new deal looks different. Anthropic is now sending over 6,000 crawlers to your site before returning a single human visitor!
We have always designed for humans. Digital experiences that drive consistent returns have held fast to "human-centered" design principles. But there is a new user on the block and they aren’t human. The volume of their visits is exponential, their hunger for data is insatiable, and they don't care about your UX design. They care about your semantics.
The visibility gap
This bot inflation coincides with a downward trend for traditional metrics. In Australia, we have seen organic traffic erode by between 5% and 35% in the last six months alone. For agencies, this shift is creating immediate pressure. As organic results decline and AI-driven behavior changes how users find information, leaders are facing increased ROI scrutiny and confusion over how to adjust strategies that worked just a year ago
Crucially, this drop isn't being replaced 1:1. Our research across 10 industries shows that referral traffic from LLMs is statistically tiny, often hovering between 0.05% and 1.5%.
This is the Visibility Gap. The human canvas is shrinking, while the AI canvas, the machine-readable layer that feeds the agents, is expanding rapidly.
The quality paradox
However, the data reveals a counterintuitive truth. While traffic volume is dropping, the value of the traffic is skyrocketing.
Our analysis of recent referral traffic reveals that while ChatGPT sends only a handful of visits, the quality of those visits is exceptionally strong. Of the sessions tracked, 80% were deeply engaged, with an average engagement time of two minutes 46 seconds versus the site-wide average of just 10 seconds. That is nearly 17x higher.
These visitors triggered 8.6 events per session compared to the average of 4.46.
This signals a massive shift in intent. When an AI agent does pass a user to your site, that user is pre-qualified, informed, and ready to act. The AI is effectively filtering out the noise.
The dual strategy: sanctuary and source
This reality demands we abandon the "one-size-fits-all" approach to digital experience. There is a growing urgency to adopt a dual-track strategy now: Defensive design for high-trust human interaction, and Offensive design for machine visibility. Waiting to address the machine-readable layer risks allowing AI models to define your brand for you.
1. The human sanctuary (high-trust actions) We must scientifically identify and preserve the human-first interactions that matter. There are high-trust contexts where we don't want AI involved, and neither does the user. A bank doesn't want a customer completing a home loan application via an agent; they need accuracy and identity verification nor can a user feel an emotional connection to a brand positioning through a ChatGPT window.
These moments require a "human sanctuary." This means rich, immersive, and connective experiences where we need to feel as much as see. Because the AI is sending us fewer, but higher-intent humans, our websites need to be better at converting that high-value attention into emotional buy-in.
2. The machine ambassador (orchestrated semantics) Simultaneously, we must control how we appear to the 60,000 bots knocking at the door. This is the shift toward optimizing content not just for keywords, but for the answers AI models provide to users.
The risk isn't just being invisible; it is being misrepresented. We recently worked with a major entertainment precinct that was losing conversions because ChatGPT was scraping an unstructured PDF and presenting factually incorrect seating capacities in a comparative table. The AI was hallucinating their business value, and customers were going elsewhere.
By restructuring data for "readability," feeding the agents with specific content rules and semantic depth, we have seen readability scores lift by up to 90%. We can create a space where we show and share our brand positioning to AI, ensuring that when it answers a question about us, it does so with authority and accuracy.
Immediate actions for agencies
To navigate this shift, agencies can take quick steps for themselves and their clients:
- Audit for machine-readability: Transition critical business data out of "locked" formats like PDFs and into structured, semantic HTML to ensure AI agents cite the brand accurately today. Review and understand how LLMs see your clients site and favor them.
- Identify “sanctuary” pages: Audit client sites to find high-conversion, high-emotion pages. Double down on the UX of these "Sanctuaries" to capture the value of the high-intent traffic AI sends. A starting point is looking at the current content that gets citations and visibility references from LLM’s as this is an indication of where the human needs sits.
- Evolve Reporting: Move beyond tracking raw organic traffic. Start measuring "LLM Referral Quality" and "Brand Citation Accuracy" to prove the ROI of AI optimization strategies.
It’s not human or machine; It's human/machine
The challenge isn't choosing between the human and the machine. It is to recognize that they now require distinct strategies.
We need to quickly reassess which parts of our brand experience are "sanctuaries" for humans, places of emotion, trust, and decision, and how we orchestrate our content to serve the LLMs that act as the gatekeepers.
The differences between the two and what matters will shift, keeping an eye on the signals to ensure you optimize for both constantly, will just be a new reality if you want to win at being a brand that's found and valued by both.
The Visibility Gap is real. The brands that can master this dual focus of designing for AI and Humans will close the gap and ensure their brand is represented the way it should be whilst being found, presented, and cited. This will happen as people use AI more and more to get fully formed and trusted answers.
