Should Affiliates Trust Google’s AI Ad Tools?

Google had a strong 2024. Revenue exceeded $400 billion for the first time, ad revenue in the fourth quarter hit $82.28 billion, up 13.5% year on year, and YouTube brought in $11.38 billion in the same period. By any standard measurement, the company most people predicted would be disrupted by generative AI has not been disrupted. That context matters for what follows.

At Shoptalk Spring, Modern Retail spoke with Courtney Rose, Google's VP of Retail at Google Ads, about how AI is reshaping the company's advertising products. The interview is worth reading carefully, not only for what Rose says, but for the specific framing she uses and the questions the conversation does not reach.

The results Google is citing

Rose describes Aritzia as a case study. The brand enabled AI Max and reportedly saw an 80% increase in revenue. She also names E.l.f. Beauty, Chewy, and L'Oreal as early testers of the “direct offers” pilot, which uses Gemini to identify purchase intent within AI Mode and serve personalised promotions at that moment. On the Universal Commerce Protocol, developed with Shopify, she describes a future where brands can complete purchases directly inside AI conversations.

These are the results Google wants affiliates and advertisers to hear. They are also, to be precise about it, case studies selected and narrated by Google. There is no independent verification of the Aritzia figure, no methodology for how revenue uplift was isolated from other variables, and no data on how many brands ran AI Max and saw nothing move. “We're seeing great performance” is the framing. The qualifier doing the work there is “we.”

The Modern Retail piece itself notes that results across the broader AI search advertising market are underwhelming. Perplexity introduced ads a year and a half ago and has since begun phasing them out. Amazon's sponsored prompts inside its Rufus shopping assistant are reportedly generating limited traffic compared to traditional Amazon ads, according to The Information. These are not minor footnotes. They are the market context that the Aritzia example sits inside.

The traffic picture Google's ad team does not address

Rose speaks about Google's advertising products for brands running paid campaigns. Her job is not to discuss what is happening to affiliate publishers who depend on organic search traffic. But that distinction is exactly what affiliates need to hold in mind when reading her comments.

AI Mode's expansion to over 180 countries is happening at the same time Google is making the case for investing more in its paid AI tools. Those two things are not unrelated. As AI Overviews answer queries directly inside Google's interface, fewer users click through to external sites. Data from Chartbeat tracking over 2,500 publisher websites shows Google search referrals fell 33% year on year in 2025. Similarweb data puts zero-click searches at 69% of all Google queries as of May 2025, up from 56% twelve months earlier. When an AI Overview appears, research from Pew found only 8% of users clicked an organic result, compared to 15% when no AI summary was present.

As we covered in our analysis of query fan-out in Google's AI search, position-one click-through rates dropped 34.5% when AI Overviews are present, and Google's top organic CTR fell from 28% to 19% following AI Overview expansion. For affiliate publishers whose revenue depends on organic traffic converting, these numbers describe a structural shift in how Google distributes attention, not a temporary fluctuation.

The company's AI tools for advertisers are designed to capture the users who remain inside Google's ecosystem. That is a coherent strategy for Google. For affiliates operating as publishers, it is a different equation: Google is simultaneously reducing the organic traffic that reaches affiliate content and selling paid solutions to reach the users it has retained.

The attribution problem that gets no mention at Shoptalk

Rose discusses the Universal Commerce Protocol as a way for brands to “enable purchases directly within AI conversations.” What she does not discuss is what that means for affiliate attribution.

Traditional affiliate tracking depends on a click, a cookie, and a redirect. A consumer reads a review, follows a link, lands on a merchant site, and converts. Every step generates a trackable signal that an affiliate network can record and credit. UCP-enabled agentic commerce removes that sequence entirely. When a transaction completes inside Google's AI Mode through Google Wallet, the affiliate publisher whose content may have shaped that recommendation receives no credit. The click does not happen. The cookie is never set.

We reported on this when UCP launched with Shopify in January. The protocol has since secured endorsements from Target, Walmart, Etsy, and Wayfair. Its reach is expanding faster than any solution to the attribution problem it creates for affiliates. Google's documentation makes no attempt to address affiliate commission structures within the UCP framework. The protocol is designed for machine-to-machine transaction efficiency. Referral economics are not part of the specification.

This is the gap between Google's framing and the affiliate reality. Rose speaks about a world where AI understands purchase intent more richly than ever and matches brands to users at the right moment. That world may well deliver better conversion rates for brands running paid campaigns. It is also a world where the content publishers who influenced those conversions at an earlier stage of the journey have no mechanism to claim credit.

What the AI Max black box means for campaign measurement

Even for affiliates running paid campaigns rather than relying on organic traffic, the transparency problem with AI Max is documented and unresolved. The core issue is attribution cannibalism: AI Max claims credit for conversions that existing exact-match and phrase-match keywords already delivered. An advertiser reviewing AI Max performance metrics sees conversion counts and return on ad spend that appear to reflect new traffic. A substantial portion of that data is existing traffic Google has reassigned to the AI Max attribution bucket.

Isolating incremental AI Max gains from redistributed existing traffic requires manual spreadsheet analysis. Google released improved search term reporting in early 2025 and added campaign-level negative keyword lists, which advertisers had been requesting for years. These are genuine improvements. They do not resolve the underlying measurement problem. As we have reported in the context of SEO volatility and how publishers are adapting, the affiliates who have managed Google's algorithm changes most successfully are those with diversified traffic and first-party data, not those who handed measurement control to Google's systems entirely.

The question Rose's framing leaves open

There is one exchange in the Modern Retail interview worth sitting with. Rose is asked how companies can use ads in a way that safeguards user trust. Her answer is about Google acting as a matchmaker, about merchants maintaining control over their creative, about relevance building trust. It is a coherent answer about the advertiser-user relationship.

The question of what Google's AI products mean for the publisher-user relationship, the content creator whose review led the consumer to form a preference before they ever entered a search query, does not appear in the conversation. That is not a criticism of Rose. It is outside her brief. But it is exactly the question affiliate program managers should be asking.

Google has introduced features framed as publisher-friendly, including expanded inline linking in AI Mode and Preferred Sources. The company is not indifferent to publisher concerns. But those features exist within a product architecture designed to keep users inside Google's ecosystem as long as possible. The traffic that publishers receive is what Google chooses to pass through, not what the open web once delivered by default.

For affiliates evaluating Google's AI ad tools, the Aritzia case study is a starting point, not a verdict. The more relevant question is what the full picture looks like for brands and publishers who are not the chosen example at a Shoptalk panel, and whether the infrastructure Google is building for agentic commerce leaves affiliate tracking with a role at all.