The open web is shrinking, and product discovery is moving with it

Digiday recently reported that publisher ad supply fell by up to 40% in Q2 as AI search and zero-click behavior reduced traffic to open-web sites. The story is about publishers, but the signal is much bigger than media supply.

For publishers, that means fewer pageviews, fewer ad calls, and more pressure on the economics of the open web. For brands, it is an early warning sign that discovery is moving into environments where the click is no longer guaranteed.

How consumers find information is changing rapidly. Search is answering more questions directly. Social platforms are keeping discovery inside their own feeds. Retailers are building AI shopping assistants that guide consumers through product decisions without sending them to a long list of links.

For brands, that creates a visibility challenge. If discovery happens inside AI-generated answers, conversational shopping tools, retailer ecosystems, or social feeds, the old playbook for being found starts to lose coverage.

Publishers are seeing the first economic impact

Digiday’s reporting shows how quickly this shift is affecting the open web. According to Ozone benchmark data cited in the article, publisher ad request volumes fell by roughly 32% to 37% year over year in the U.S. and 39% to 41% in the U.K. between April and June 2026. In plain terms, fewer consumers are arriving on publisher pages, which means fewer pageviews, fewer ad calls, and less inventory to sell.

At the same time, the article notes that eCPMs are rising, which suggests demand for quality inventory has not disappeared. The issue is that supply is becoming scarcer as discovery moves into environments controlled by search engines, social platforms, and apps. Publishers are responding by leaning harder into logged-in experiences, newsletters, subscriptions, apps, events, and first-party data.

That is the bigger lesson for brands. When platforms answer questions in place, the downstream journey changes. Publishers are seeing it in lost referral traffic and fewer ad calls. Brands may feel it when consumers get product recommendations before they ever reach a traditional search results page.

It also raises the bar for paid media. If premium inventory is getting scarcer, brands need to be more precise about where their dollars go and which impressions actually change consumer behavior. Within Amazon DSP and Amazon STV, closed-loop tools like Amazon Marketing Cloud can help brands understand which sites, publishers, audiences, and frequency levels are driving product detail page views, purchases, and new-to-brand growth. Flywheel’s Amazon DSP Site Filtering and Frequency Optimization approach builds on that idea, using retail outcome data to reduce waste and reinvest in the inventory that is more likely to move consumers closer to conversion.

Discovery is becoming more compressed

Traditional search gave brands significant surface area. A consumer could scan a results page, compare several links, open multiple tabs, and move through a more visible consideration journey.

AI search changes that behavior. The consumer asks a specific question and gets a short answer, a summarized recommendation, or a narrow set of products to consider. There is less scrolling, less clicking, and less room for brands to be discovered by accident.

That is the same pattern showing up in the open-web data. When platforms answer questions in place, fewer consumers move downstream to publisher pages. In commerce, the same dynamic can happen when AI shopping tools answer product questions before a consumer ever reaches a full search results page.

The consideration set gets smaller, the path gets shorter, and the cost of missing from the answer gets higher.

Product content becomes the new discovery layer

Brands have spent years optimizing for keywords, rankings, paid search, and retail media placements. Those all still matter, but AI shopping tools operate differently.

When a consumer asks, “What is the best protein powder for a sensitive stomach?” or “What are natural brands of sunscreen for kids?” an AI assistant is looking for product information that helps answer the question. Ingredients, use cases, age ranges, certifications, benefits, reviews, comparison points, and contextual details all become signals.

That means product content is no longer just there to support conversion after the consumer lands on a PDP. It increasingly determines whether the product is included in the journey at all.

If the right signals are missing, unclear, or buried, the AI has less reason to recommend the product, even if the brand performs well in traditional search.

This is bigger than AI search alone

Discovery is moving into environments where platforms control the journey more tightly. Search engines, social platforms, retail media networks, apps, newsletters, and AI assistants are all creating more closed or curated discovery experiences.

For brands, this raises a practical question: can your product information travel across those environments?

A PDP written only for a human scanning a product page is not enough. Product content now has to serve multiple jobs at once. It has to support traditional search. It has to help consumers understand the product quickly. It has to give retailer algorithms the right signals. And increasingly, it has to give AI systems enough structured information to answer consumer questions confidently.

This is where Generative Engine Optimization, or GEO, becomes an important part of the commerce toolkit.

GEO helps brands prepare for AI-mediated discovery

GEO is the practice of making product information structured, complete, and usable by AI systems when they generate recommendations or answer shopping questions.

The goal is to understand the questions consumers are asking, identify the product signals AI systems need to answer those questions, and close the content gaps that could keep a product out of the recommendation set.

For example, a brand may already rank well for “face sunscreen” in traditional search. But if a consumer asks an AI assistant for “a sunscreen that works under makeup and won’t irritate sensitive skin,” the model may look for SPF level, skin type, texture, ingredient details, fragrance information, dermatologist testing, and consumer reviews about wearability. If that information is incomplete or unclear, the product may not be surfaced.

That is the visibility gap that GEO is designed to address.

What brands should do now

The first step is to stop treating AI search as a distant channel and start treating it as an emerging discovery behavior.

Brands should pressure-test their product content against the questions consumers are likely to ask before buying. Which attributes matter? Which concerns need to be answered? Which use cases are missing? Which claims are clear to a person but not structured enough for a machine to interpret?

They should also evaluate where GEO, SEO, and media efficiency intersect. Traditional search is still critical, and brands should not sacrifice keyword visibility in the name of AI readiness. At the same time, media dollars need to be pointed toward the inventory, audiences, and frequency levels that actually drive outcomes. The opportunity is to build a visibility strategy where product content and paid media work harder together.

Flywheel’s GEO Optimization service, a new capability within Omnicom's global portfolio of GEO solutions spanning commerce, media and public relations, was built to help brands do that in a structured way across Amazon, Walmart, and Target. It identifies where product pages are ready for AI search, where signals are missing, and what content needs to be improved so products are more likely to be understood and recommended by AI shopping tools.

As discovery gets more compressed, product content has to carry more weight.

The brands that adapt early will have the advantage

Discovery shifts tend to move faster than operating models. Publishers are already adapting to this reset by building more controlled, high-attention environments. Commerce brands should take the same signal seriously. This requires product content that is strong enough to travel into the places where discovery is moving.

Consumers are not going to stop searching, comparing, or asking questions. But the places where those questions are answered are changing.

Brands that build for this now will be better positioned as AI shopping becomes a larger part of the consumer journey. Brands that wait may find themselves optimizing for a discovery path consumers have already moved beyond.

GEO is still early, but the window to build advantage is open now. Read more about GEO Optimization in our whitepaper, “From powerless to proactive: How brands can own the AI search revolution.”

If you are starting to evaluate how your products show up in AI shopping results, Flywheel’s GEO Optimization service can help identify where your content is ready and where it needs work.