AI Agents and the Future of Ad-Revenue Businesses

Source: Emil Rustamli; The Terminal Value (Medium) · 2025-05-03

In late 2024, Perplexity launched a shopping feature to let users find and buy products via AI chat. Following suit, OpenAI just rolled out an shopping mode in ChatGPT, where the chatbot will recommend products with images, reviews, and direct purchase links. And in perhaps the biggest sign of this trend, Mastercard introduced “Agent Pay,” an integration that lets conversational AI agents go from personalized recommendations straight through checkout, and Visa announced a partnership with tech players like OpenAI, Perplexity, Stripe, Microsoft, and others to enable “Intelligent Commerce”; in other words, letting AI agents seamlessly find and pay for things using Visa’s network.

Klarna’s CEO Sebastian Siemiatkowski is even predicting shoppers might stop visiting retailer websites altogether, instead just “sending” their AI to make purchases. And this makes me wonder, if we’re not actively browsing marketplaces or doom-scrolling search results anymore, what happens to the hundreds of billions in ad revenue tied to product discovery and shopping funnels?

AI assistants excel at giving you one concise answer or recommendation. That means users rely less on sprawling search result pages or endless product listings, and those are the very places where digital ads live. A recent Bain & Company brief calls these new AI-guided journeys “zero-click” searches because users get what they need without additional clicks. As a result, the discovery-to-decision process is compressed, reducing opportunities for brands to influence consumers or even appear during the journey.

Early data is eye-opening. When Google started testing its own AI-generated answers in search (the Search Generative Experience), advertisers immediately felt the pinch. One analysis found that when an AI answer was present at the top of Google’s results, click-through rates on paid search ads plummeted from ~21% to under 10%. Organic results fared even worse, seeing a ~70% drop in clicks. And unlike a human user who might browse multiple sites, an AI agent isn’t “curious”; it won’t wander through page 2 of Google results or click an eye-catching banner ad. It will go straight to the relevant source or skip to the transaction. Fewer searches and site visits mean fewer ad impressions, period.

And this matters because search advertising is Google’s entire vibe. In 2023, ads on Google search brought in $175 billion, more than half of Alphabet’s total revenue and more than they make from YouTube, Android, or literally anything else.

On the other hand, some might think Amazon wins regardless. If AI agents skip Google and go straight to Amazon, that’s fine for Amazon, right? Not exactly.

Even Amazon could feel the pinch if user behavior changes. Consider how an AI shopping assistant might work: it could query Amazon’s database via an API, find the top-rated item matching your criteria, and place the order. All without you ever “shopping” on Amazon’s site or seeing sponsored products. In that scenario, Amazon becomes a transaction back-end rather than a digital mall with aisles of ads.

Obviously, Amazon will still make money selling products. But the lucrative ad fees it charges sellers for premium placement could seriously diminish if fewer people actually browse the Amazon website.

Amazon sees this coming, and they’re already hedging by enabling their own AI voice assistant to handle purchases and expanding “Buy with Prime” so external sites can use Amazon for checkout. The arms race is on to integrate with AI agents rather than be circumvented by them.

Google, for its part, has reportedly begun experimenting with ads in its chatbot conversations. But integrating ads into an AI answer is a delicate task. OpenAI’s CEO Sam Altman has openly said he finds ads in AI “uniquely unsettling” and would only consider it as a last resort. Still, OpenAI’s leadership has hinted that some form of paid placement might be on the table in the future. The challenge will be maintaining user trust — if people feel the assistants’ recommendations are biased by advertisers, the whole appeal of an “objective” AI shopper could vanish. It’s a high-stakes balancing act: tech giants must reimagine their ad models without alienating users who prefer these AI tools specifically because they cut out the ad noise.

Here’s where things get interesting (and why you shouldn’t panic if you’re a content creator). AI agents excel at fulfilling needs, but they’re trash at creating wants.

When you need a new vacuum cleaner because yours just died, that’s perfect for an AI assistant. “Find me the best bagless vacuum under $300 with pet hair features”. Boom, done. The AI bypasses all the ads and sponsored listings to give you exactly what you need.

But what about those random purchases you never planned to make? The lip gloss your favorite TikToker swears by? The home decor piece that popped up in your Instagram feed that you didn’t know you needed until you saw it? These wants are created through discovery, vibes, and social influence, areas where AI simply can’t compete with the dopamine hit of scrolling.

This is why social media and influencer marketing might actually become more valuable in an AI-dominated world. Scrolling TikTok or Instagram isn’t just about information; it’s entertainment, inspiration, and community. You discover new skincare from that beauty influencer’s genuine review, or get travel “inspo” from a friend’s aesthetic vacation post.

The numbers back this up. Influencer marketing budgets are projected to hit $9.3 billion in 2025 (up 14% from 2024). And more marketers are planning to boost spend on creators across TikTok, YouTube, and emerging platforms in 2025 as consumers explore new social channels. In other words, while AI might help you find what you already know you need, humans will still be discovering what they want through other humans.

Shopping isn’t the only arena being transformed.