ChatGPT Ads Are Live. 100 Million Weekly Queries. Almost Zero Competition.

Here are the numbers that matter right now.

100 million weekly AI search queries across ChatGPT and Perplexity. 67% of people under 50 using AI to evaluate organizations before they act. 60% of adults over 25 using AI search weekly. 37% of consumers starting their research in AI instead of Google.

And almost no one is advertising there yet.

A keyword gives you a topic. A prompt gives you a person.

The difference between a traditional search query and an AI search prompt is the difference between guessing and knowing.

“Animal charity” tells you almost nothing. You know the subject, but not the motivation, the concern, the comparison the user is making, or how close they are to taking action. Every advertiser bidding on that keyword is guessing.

“Which organization should I support for animal rescue, and how do I know they are trustworthy?” tells you everything. Need. Motivation. Comparison. Trust concern. Next step. All in a single prompt.

AI search platforms like ChatGPT and Perplexity are creating a new demand-capture layer where intent is revealed before it ever becomes a traditional search query. The user explains the full situation, not just a keyword. And that richness of signal changes what’s possible at every stage of the media strategy — more audience segments, more creative angles, more landing page paths, and more retargeting pools, all built from real intent rather than inferred behavior.

The performance data reflects this. AI search sessions run roughly 30% longer than traditional search. AI-referred organic traffic converts at 4 to 5 times the rate of standard organic. Engagement with AI search ads is 2.8 times higher than traditional ad formats. And the addressable audience is 2 to 3 times larger than what most organizations are currently reaching through conventional channels. Brand search lift from AI search exposure is running at 15% or higher.

For organizations that have been trying to personalize at scale but never had the signal quality to do it well, this changes the math entirely. Every prompt is a window into what the user actually cares about — not what a keyword planner guesses they might care about.

The cost is low because nobody else is there yet.

There is no minimum spend to run ads directly through OpenAI right now. Zero. The audience skews educated, senior-level, and high-income — 80% hold an undergraduate degree, 30% are in senior leadership roles, and 65% work in high-income white-collar professions. And the auction is nearly empty.

Low cost and high performance don’t usually show up at the same time. When they do, it’s because a channel is new and the market hasn’t caught up. That’s exactly where AI search advertising is right now.

The economics will change. As more advertisers discover the channel, competition increases, CPCs rise, and the cost advantage narrows. The organizations testing now will have locked in learning, built audience pools, and established creative benchmarks at a fraction of what those same assets will cost in 12 months.

First movers are building the benchmarks everyone else will chase.

There are no credible AI search advertising benchmarks for most industries yet. No established CPCs. No proven creative formats. No standard audience segments. No playbook.

That’s not a risk. That’s the opportunity.

The organizations that test now will set the CPCs that latecomers will have to outbid. They’ll discover the creative norms that latecomers will have to match. They’ll build the audience segments that latecomers will have to buy at a premium once the auction is crowded.

A 90-day pilot is enough to generate directional signals across conversation themes, engagement quality, and early conversion intent. By day 45 you have a signal report — CTR, CPC, and prospect-to-engager rate via AI search intent. By day 90 you have a conversion report — cost per acquisition compared to your existing search baseline.

That means an organization that starts testing today will have real performance data before peak season in Q4. An organization that waits until September will be starting from zero at the exact moment when costs are highest and competition is fiercest.

The first mover advantage here is not about being early for the sake of being early. It’s about having 90 days of compounding learning that nobody who starts later can shortcut. Every week in the channel is a week of data that informs what to start, what to stop, and what to scale. That learning curve is the moat.

How it fits into the paid media stack.

AI search is not a replacement for Google or Meta. It’s an upstream layer that strengthens everything else.

A user searches in ChatGPT or Perplexity. GA4 captures the visit. If they don’t convert, Google Ads and DV360 can retarget them across search and the open web. The hypothesis — supported by the early data — is that AI-referred users carry stronger intent than cold prospecting audiences, which means better CPA, better ROAS, and better portfolio efficiency across the full media stack.

The longer AI queries also inform media strategy beyond the AI channel itself. The conversation themes reveal what users actually care about, what they’re comparing, what objections they have, and what would move them to act. That intelligence feeds audience segmentation, creative development, and budget allocation across every channel — not just AI search.

The window is open. It won’t stay open.

The gap between a new channel opening and the auction getting expensive is always short. It was short with Google Ads. It was short with Facebook. It was short with programmatic. AI search is in that window right now.

The question is not whether your audience is searching in AI. They are. The question is whether your organization will be there when they do — and whether you’ll have 90 days of data behind you when everyone else shows up.