Chartbeat data shows year over year declines in Google referral traffic across the board, and while big publishers are feeling a dent, medium sized publishers are feeling it more, and small publishers are looking at a genuine problem. AI is increasingly answering queries directly, without sending a click anywhere, and the publishers producing the content that trains and grounds those answers are the ones losing the visits. 

That was the starting point Max Willens, Principal Analyst at eMarketer, gave the room at Affiliate Summit East this year, and it set up the real question of his session. Not whether AI search is reshaping publisher revenue, since that is already happening, but what publishers should actually do about it. He walked through three distinct response strategies, the commercial logic behind each one, and the trade offs that do not get discussed loudly enough. 

His message was clear from the outset. This is not a single choice decision that publishers make once and move on from. It is closer to a toolbox, and every publisher in the room needs a different combination of tools depending on their size, their content, and their appetite for risk. 

Why This Is Happening Now 

Before getting into strategy, Willens grounded the room in data that put the scale of the shift into perspective. eMarketer forecasts that over 152 million people will use generative AI monthly in the US by the end of 2026, which works out to about 45% of internet users, climbing to 55% by 2030. Ipsos tracking data backs this up, showing that regular AI users have become a majority in just two years, with the share of Americans who say they never use AI dropping from a third in 2024 to 17% today. 

Penetration is not the same thing as disruption, though, and Willens was careful to make that distinction. Most people are still using AI the way they might use a souped up search engine, looking for facts, explanations, and comparisons rather than completing purchases outright. Shopping shows up further down the list of use cases, and the number of people buying something immediately after an AI assisted research session remains small, with roughly one in ten doing so on a weekly basis. 

What is not small is how much AI now relies on the internet's crawl infrastructure to get those answers. AI bot scraping now accounts for more than half of all crawls Cloudflare observes across its network, and that figure climbs closer to eight in ten once grounding and retrieval activity is included. Publishers are being scraped constantly to power responses that then keep the reader inside the AI platform instead of sending them onward, which is exactly why referral traffic keeps sliding even as AI usage keeps climbing. 

There is a genuinely encouraging thread in the data too. Willens shared AI Visibility Index figures showing that publisher domains cited in personal care and beauty queries doubled from one in ten in March to one in five by May, which suggests real trust is being placed in publisher content. So the content is valued. It is simply not being compensated in the way clicks and ad impressions once were, and that gap between value and payment is exactly what the next three strategies are built to close. 

What eMarketer Says To Actually Do About It 

Block Access and Demand Compensation 

Tools like Tollbit, alongside hosting companies and cloud providers, now give publishers genuine options for restricting or monetizing AI crawler access rather than simply absorbing it for free. The case for blocking is strong. It signals that content is not free, and for most publishers, current referral traffic from LLMs is small enough that losing it entirely would not be catastrophic to the business. 

The trade off is that this becomes an ongoing cat and mouse game. BuzzStream data shows that many of the very publishers most frequently cited by ChatGPT and Google's AI mode are also the ones actively working to block crawlers, so blocking does not mean disappearing from AI answers altogether. It means opening a negotiation. 

Pursue Direct Licensing Deals 

The biggest and most recognizable publishers have already signed direct licensing deals with OpenAI, Perplexity, and Amazon, many of them struck defensively to keep powerful newsrooms from turning hostile. Smaller publishers without a partnerships team on speed dial now have options too, including set your own price tools like Tollbit and Snowflake Cortex, marketplace style models, and collective licensing frameworks such as RSL. 

The benefit is not only revenue but also insight, since understanding how frequently your content is being scraped starts to inform a broader content and audience strategy over time. The risk is a race to the bottom, because relatively little content across the web is truly differentiated, and downward price pressure remains a real possibility as this market continues to mature. 

Explore Branded AI Visibility Packages 

The most nascent, and arguably most promising, strategy pairs publishers directly with brands, embedding products into content specifically to boost AI citation and visibility. eMarketer and Partnerize research found that nearly all brands surveyed, 98% of them, now believe AI visibility has changed their approach to marketing. Yet fewer than three in ten have a clear, unified strategy for managing it, and under 10% believe their current attribution model fairly compensates the publishers who help drive that visibility. 

That gap represents the opportunity. With almost no brands currently equipped to formally compensate partners for AI visibility, publishers have a rare chance to help define what fair actually looks like before the market hardens around someone else's terms. 

The Bottom Line 

Blocking buys leverage and sends a message, though it requires an ongoing operational commitment to stay ahead of the crawlers. Direct deals convert scraping into revenue and insight, but come with real pricing risk as the market matures around them. Brand partnerships remain wide open right now, with brands actively looking for publishers willing to help them figure AI visibility out together. 

None of these strategies is mutually exclusive, and Willens was clear that most publishers will end up blending all three depending on their content and their goals. But the traffic decline that started this conversation is not going to reverse itself. The publishers who treat this as a negotiation, rather than a loss to absorb, are the ones who will be paid for the value they are already creating.