AI Search Advertising Is Here. Nobody's Figured It Out Yet.

AI Search Advertising Is Here. Nobody's Figured It Out Yet.

2 weeks ago, I sat across from teams at AdMedia and adMarketplace. Both have live AI advertising products. Both are figuring it out in real time. That's not a criticism. That's the honest state of the industry right now.

AI search advertising is not coming. It is here. But the model, the rules, the measurement, the creative formats: all of it is still being written in real time.

Here is what I am seeing on the ground.

ChatGPT Finally Turned On the Ad Machine

In early 2026, OpenAI launched its advertising pilot. The program started with a $150,000 minimum commitment and a $60 CPM. Enterprise-only, by design. Six months later the minimum dropped to $50,000, cost-per-click became available at $3 to $5, and self-serve opened to all US businesses. What was interesting was not the launch. It was how fast the terms changed.

They did not build everything from scratch. They plugged directly into the existing ad tech stack and let it carry them.

Criteo brought its Universal Catalog, processing $1 trillion in annual transactions and tracking 720 million daily shoppers, directly into ChatGPT via API. Kargo handles specialized creative execution inside the chat interface. StackAdapt joined on May 5, 2026, giving programmatic buyers a familiar on-ramp into conversational AI placements without learning a new platform from scratch.

The early numbers are not projections. They are actual performance data. Within months of launch, over 1,000 brands were live. AI-referred conversion rates approached 2x traditional search benchmarks in categories like consumer electronics, lifestyle, and home goods. Click-through rates for native conversational formats sat at roughly 3x the industry average.

Source: Criteo / PPC Land, May 5, 2026

Why? Because someone using an AI chat to ask "what is the best air fryer for a family of four" is not casually browsing. They are in deep consideration mode. The intent signal is extraordinarily high. When the right product surfaces at that moment, the conversion math changes entirely.

This is not a new keyword auction. It is something more contextual, more embedded, and much more dependent on product data quality than anything that came before it.

The Infrastructure Layer Nobody Is Talking About

While everyone watches the ChatGPT and Google headlines, a company called ZeroClick is building something that will matter more in three years than it does today.

ZeroClick is an ad network designed specifically for AI chatbots. Their model connects user questions to relevant brand recommendations at the moment of conversational intent, on a CPC basis. They raised $55 million at launch, backed by the same investor group that took Honey to a $4 billion sale to PayPal, and acquired Sleek, a Y Combinator-backed startup, to extend their reach beyond chatbots into shopping experiences. The founder is Ryan Hudson, who co-founded Honey. That lineage matters.

Here is why this matters: Criteo, Kargo, and StackAdapt are doing direct deals with ChatGPT. That is the premium inventory. But there are thousands of AI apps, companions, agents, and niche chatbots that will never have a direct partnership with a major ad tech player. ZeroClick is building the infrastructure for that long-tail universe, the same way ad networks democratized display advertising in the early 2000s.

If you want to understand where AI advertising goes in the next five years, watch what ZeroClick does next.

The Model That Is Not Advertising (But Kind Of Is)

Mindtrip is worth a separate section because it represents a category I have been calling partnership-native AI.

Mindtrip is an AI travel platform backed by Amex Ventures, Capital One Ventures, and United Airlines Ventures. They recently launched the first all-in-one agentic AI flight booking experience in partnership with Sabre and PayPal. You chat with the AI, it plans your trip, and it books your flight through Sabre's infrastructure with PayPal checkout. All inside the conversation.

There is no "Sponsored" label. No keyword bid. The influence is embedded in the business model itself. The investors are the partners. The partners are the inventory.

This is fundamentally different from buying an impression on ChatGPT. It is a commercial relationship that shapes the recommendation layer before any ad auction takes place. For marketers in travel, hospitality, and commerce more broadly, this model is a preview of what brand influence will look like when advertising and recommendation collapse into a single moment.

Two Companies I Met With That You Should Know

AdMedia / Search.com

AdMedia runs one of the largest search marketplaces outside of Google and Bing, with 11 billion+ monthly impressions and deep first-party intent data. Search.com is their AI-native layer: AI-generated answers, SERP placements, search suggestions, and commerce units built for performance advertising.

The positioning is compelling. Take the intent signal from an AI search interface and convert it directly to a performance outcome. The infrastructure behind it is proven, the scale is real, and the AI layer adds a contextual targeting dimension that traditional search lacks. If you are running search campaigns and have not looked at what they are building, you should.

adMarketplace / Opera Aria

adMarketplace took a different path. Their product, AMP Discover, powers advertising inside Opera's Aria AI chat application. The mechanics are clean: ads pull from product feeds, you pay CPC, and the ad surfaces when Aria determines commercial intent within the conversation.

Opera does not have Chrome's market share. But Aria users are engaged, and the signal-to-noise ratio of a chat-based commercial intent moment is significantly higher than a passive scan down a SERP. This is exactly the kind of niche, high-intent inventory that sophisticated performance marketers should be testing right now.

What This Means for You

There is no finished playbook. The brands with early data built it by running imperfect tests in environments that do not yet have perfect measurement. That is not a knock on them. That is how every new channel starts.

A few things I know for certain:

Product data is now a competitive advantage. GEO (Generative Engine Optimization) is not a marketing tactic. It is an operational requirement. If your product catalog has incomplete attributes, inconsistent descriptions, or empty backend fields, AI systems will skip you entirely and recommend a competitor with better structured data. Every blank field is a missed match.

Transparency will not be optional. 91% of consumers expect brands to disclose when they are using AI in marketing. The platforms building durable advertising models understand this. The ones that do not are building on sand.

Source: Emplifi

The long tail is where the real opportunity is early. The ChatGPT premium inventory is competitive and expensive. ZeroClick, AMP Discover, and Search.com represent adjacent surfaces with lower competition and high intent. Test there first.

Luxury brands should rethink their ChatGPT allocation. ChatGPT ads surface exclusively for Free and Go tier users. That is not the luxury buyer. High-consideration, high-income shoppers index toward Claude, Perplexity, and Brave. Putting luxury budget into ChatGPT's free tier is a demographic mismatch. The reach is real. The audience fit is not.

Watch the partnership model, not just the ad model. Companies like Mindtrip are showing that the deepest form of AI advertising influence will not look like advertising at all. It will look like integration.

The Creative Question Nobody Is Answering Yet

In AI search, context is the targeting. The same person asking "best running shoes" after a morning run and after reading a marathon recap is not the same buyer in the same moment. AI systems understand that distinction. They are built around it.

Which means creative has to move with context too.

Product data feeds handle this naturally. The catalog has the attributes. The AI matches them to the moment in conversation. That is why product feeds became the default first ad unit inside AI search environments. They are already structured for it.

Brand creative is a different problem.

Can a brand trust an AI to assemble copy, messaging, and tone in real time to match a conversation it never saw coming? Eventually, probably yes. But the infrastructure, the guardrails, and the brand standards needed to make that work do not exist at most companies yet.

This is not a new tension. Google ran into it with Performance Max. Responsive Search Ads asked advertisers to hand over a set of headlines and descriptions and let the algorithm decide what combination to show. Many brands pushed back. Not because the performance was bad. Because they lost visibility into what the ad actually said when it ran.

AI search advertising takes that same friction and multiplies it. In a Performance Max placement, the worst case is a mismatched headline. In a conversational AI placement, brand voice, offer, and context all have to cohere inside a live dialogue that the brand never scripted. That is a fundamentally different creative problem.

Dynamic creative is not optional here. It is a structural requirement for the channel to work at all. But building it with real guardrails takes more than a brief and a creative team. Most brands are not there yet. And the industry is not having that conversation seriously yet.

ChatGPT ads launched six months ago at prices that made it enterprise-only territory. Those prices have already dropped. The measurement is still catching up to the spend. Nobody has the full picture yet. That is the honest state of things.

AIVO tracks AI visibility and citation performance across ChatGPT, Perplexity, Google AI Overviews, and other AI search surfaces. If you want to understand where your brand stands in the AI search landscape, start here.

Author: Sebastian Pinzon is Co-Founder of AIVO, the AI Visibility Intelligence Platform. After 15+ years in digital marketing at Publicis, WPP, and Omnicom, he helps mid-market brands measure and improve their presence across ChatGPT, Perplexity, Google AI Overviews, and Claude.

Connect on LinkedIn | tryaivo.com