We’re Thinking About ChatGPT Ads Wrong

Thanks for reading. Feedback?
Reply to this email or message [email protected]

Let’s get into it

We’re Thinking About ChatGPT Ads Wrong

ChatGPT ads don’t sell clicks. They sell a seat inside
the model’s reasoning.

We’re thinking about ChatGPT ads the wrong way.

Not a little wrong….Fundamentally wrong, in a way that makes smart teams spend real money and still feel confused by the results.

“This isn’t a channel change. It’s a transformation of how the funnel is evolving.”

Distance is disappearing

In ChatGPT’s search, the distance between awareness and action is collapsing.

In the traditional Google search, there was a space between “I’m curious” and “I’m buying.”

That was space where ads could intercept, retargeting could chase, content could warm, and attribution could pretend it understood what happened.

They don’t want ten tabs and a spreadsheet. They want one clean answer that lets them move, and the model’s job is to produce that answer without getting anyone hurt, embarrassed, or misled.

When the distance collapses, the systems built on distance start to malfunction.

Last-click attribution starts telling stories instead of telling the truth. Funnel stages blur so much that you can’t even point to where “awareness” ends anymore.

Now, it’s all in one moment…The question, the reasoning, the decision.

So the budgeting question changes, whether you like it or not.

The new budgeting question

The old question marketing teams would ask was:

“Which channel deserves more budget?”

The new question is more focused:

“Is my brand selectable in the answer at all?”

And then, immediately after:

“Once I’m considered, can I actually convert the answer to a purchase?”

If the model doesn’t trust you, it won’t recommend you. 

If it won’t recommend you, ads don’t magically create that trust; they just rent attention for a brand the system is still hesitant to stand behind.

You have to be selectable. That means, can the model say your name with confidence, or does it hesitate because it can’t defend the recommendation?

The framework that replaces brand vs performance

Stop splitting budgets into “awareness vs conversion.” With the advent of ChatGPT ads, the split that matters is different. It’s:

Eligibility Spend vs Capture Spend.

Two buckets, two jobs, two completely different failure modes.

• Eligibility Spend: the investment required for ChatGPT to include you credibly

• Capture Spend: the investment required to win the recommendation once you’re included

This is how you avoid dumping money into the wrong lever while thinking you’re “testing ads.”

A lot of brands are about to run capture spend before they’ve earned eligibility, and the results will look like… nothing.

Eligibility Spend is about being safe to recommend

Eligibility answers one question:

Can the model mention you without embarrassing itself?

That sounds dramatic, but it’s literal. ChatGPT is trying to reduce risk.

It is constantly deciding what it can safely say, defend, and justify if the user asks “why them” or “how do you know.”

What builds eligibility comes down to two things:

• Decision-support content: pages that make choosing you feel safe, specific, and defensible

• Trust signals: third-party proof and consistency across the web, so the model believes the pattern

Decision-support content (what most teams underbuild)

A lot of marketing content is written like it’s afraid to commit to its brand.

It’s so general, even vague, that nobody can argue with it. In ChatGPT’s search, that kind of content is functionally useless because the model can’t anchor to it.

Eligibility content has edges.

It tells ChatGPT where you belong and where you don’t…It does it clearly enough that the model can confidently place you into an answer without hesitating.

Instead of “All-in-one platform for modern teams,” it’s:

• Who it’s for

• What outcomes it drives and what it replaces

• Can it be integrated? What does it integrate with?

• What are the constraints? The tradeoffs?

Not the “if you’re X, don’t buy this” lines that most marketers refuse to write because they think honesty reduces conversions, even though it increases trust and the right kind of conversions.

It wants language that feels like an expert on the situation wrote, not like someone who heard about it on the news and decided to write a blog post.

Trust signals: What makes ChatGPT willing to stick its neck out

Even if ChatGPT understands what you are, it still needs to believe you’re real and reputable.

That means it wants external gravity: reviews, credible mentions, consistent positioning across sites that aren’t yours.

ChatGPT also wants the story to match wherever it looks.

If one page says you’re “AI-powered workflow automation,” but another says you’re “a data platform,” ChatGPT doesn’t see versatility; it sees inconsistency.

Inconsistency reads like risk.

Here’s the metaphor that won’t leave my head:

Running capture ads with low eligibility is like pouring lighter fluid on a dead campfire; you might get a flash, you won’t get heat, and you’ll still be cold.

Capture Spend is where ads actually belong

Capture is a different job.

Capture answers this:

Now that ChatGPT considers you legitimate, can you tip the decision in your favor?

That’s where ChatGPT ads make sense. Not as a trust factory, but as a decision nudge in your direction.

It works when you’re already one of the “reasonable options” in ChatGPT’s internal shortlist…The ad reinforces why you’re the best choice in this context.

It’s amplification, not invention.

It’s like the difference between trying to get invited to the meeting versus being in the room and having a seat at the table.

The split: 70–90% eligibility, 10–30% capture

If you want a rule you can actually use, start here:

• 70–90% on eligibility

• 10–30% on capture

Eligibility compounds…It makes every future impression cheaper because ChatGPT gets more confident over time.

It reduces friction across every surface, not just ads, because you’re building a defensible brand footprint that ChatGPT can lean on again and again.

It underperforms in a way that’s easy to misread, because the ads will still “run” and the numbers will still populate, but ChatGPT won’t lean toward you consistently, not in the moments that matter.

Early LLM ad environments will almost certainly protect UX aggressively. They have to.

If sponsored placement degrades answers, users stop trusting the product. So the system will naturally reward brands that already look safe, already look consistent, and like they belong.

Skip eligibility and jump straight to capture, and you get what I’ll call quiet cash burn… Spending goes out, reports come back, and nobody can diagnose the real bottleneck because everything is technically “working.”

Google Ads vs ChatGPT ads

Google Ads sell clicks. ChatGPT ads sell placement inside reasoning.

“Extends” is an important word because the sponsored recommendation appears where it logically fits within the answer. If it’s done well, it feels additive, like the model is giving you an extra layer of help.

If it’s done poorly, it feels like the model got bought, and users hate that feeling more than they hate banner ads.

This isn’t about creating demand. It’s decision-layer amplification.

You’re not buying attention at the top of a funnel.

You’re buying influence inside the moment the user is already trying to decide. That’s why eligibility matters so much, because if ChatGPT doesn’t believe you’re safe, it won’t let you sit inside that moment.

The line to keep on your wall

If you want to buy outcomes in ChatGPT, you don’t start with ads.

You start by becoming selectable.

Quick debate starter

Two questions, because I want to know if this matches what you’re seeing:

• If you had to cut budget tomorrow, what gets cut first: eligibility or capture?

• What’s your brand doing right now that would make an LLM hesitate to recommend you?

Jason Patel
Co-founder & CEO, Open Forge AI
AI agents that do the AI SEO work for you.

AI SEO Strategies

If you liked this newsletter, feel free to forward to your friends in marketing and AI.

If you got this from a friend, you can sign up to get the newsletter: https://aen.beehiiv.com/subscribe