OpenAI Ads Manager Beta dropped the $50K minimum, CPC bidding shipped, and Criteo reported LLM-referred traffic converting at 1.5x other channels. The catch: this isn't another performance channel. It's the first ad surface built around conversation intent, and it signals where the entire ad market is going.
OpenAI Ads Manager Beta dropped the $50K minimum spend, CPC bidding shipped, and Criteo reported LLM-referred traffic converting at 1.5x other channels. The catch: this isn't another performance channel. It's the first ad surface built around conversation intent, and it signals where the entire ad market is going.
Performance marketing is plateauing because the buyer's decision has moved upstream of the click. It now happens in Reddit threads, X replies, LinkedIn posts and AI-assistant answers. ChatGPT Ads is the first major ad surface that targets the buyer's situation, not the buyer's keyword. Budget will follow attention, and attention is moving to wherever intent gets formed. To compete in the next market, brands need to operate across three layers: SEO answers intent, social conversations create intent, and AI-native ads capture intent at the moment of decision. The teams that win won't buy more clicks. They'll show up earlier in the chain that produces the click.
The rest of this piece is the case.
You'll see this shift called intent marketing, intent based marketing or search intent marketing in different posts. They all describe the same move: paid channels stop being the entry point of the buying decision and become a closing layer on intent formed elsewhere. For the technical breakdown of each search layer in this shift (SEO, AEO, GEO and the conversation intent layer underneath), see SEO vs AEO vs GEO: how brands get ranked, quoted and recommended in AI search.
The performance marketing playbook of the last decade was simple. Pick a platform that controls distribution. Give it data, budget and creative. Let it find the buyer.
That trade has weakened on every axis.
CAC keeps rising across paid social and paid search. Attribution is noisier post-IDFA and post-cookie. The big platforms are saturated, which means the marginal click costs more and converts less. And on top of all of it, buyer trust in ads is at a multi-year low. Buyers trust peers, creators, communities and AI-generated recommendations more than they trust the brand running the campaign.
So the buyer's decision moves elsewhere. Half-made before the website visit. Sometimes wholly made before the website even appears in the consideration set.
That's the actual problem with performance marketing in 2026. The funnel didn't break. The decision moved out of it.
OpenAI's ad product is interesting not because it adds a placement. It's interesting because it changes the targeting unit.
On Google Ads, the unit of intent is a keyword string. On Meta, it's an audience definition. On ChatGPT Ads, it's a conversation: the buyer's task, their constraints, the alternatives they've already named, the moment they're in. OpenAI calls the targeting input "context hints", and the official ad groups doc is explicit that hints describe situations rather than matching exact terms.
The early numbers back the model. Criteo reported LLM-referred traffic converting at roughly 1.5x other referral channels across a 500-retailer sample. Observed CPCs cluster around $3–5 for B2B intents, with default CPMs that started near $60 and drifted toward $25 as more advertisers came online. Not cheap. But cost-per-qualified-lead is what matters here, and the intent density is the highest of any ad surface that currently exists. We covered the tactical setup in our 2026 ChatGPT Ads Manager guide; this piece is about why the surface exists at all.
The shift in one line:
Old advertising targets the person. New advertising targets the decision context.
That's the whole frame.
Every major paid channel of the last fifteen years monetised attention. Meta monetised the feed. Google monetised the query. TikTok monetised the scroll. LinkedIn monetised professional identity. The display ecosystem monetised web inventory.
The mechanic was the same across all of them:
find the user
interrupt the user
send the user to a landing page
retarget the user
attribute the conversion
scale the budget
It worked because the platforms controlled where buyers spent their attention. They don't anymore. At least not exclusively. Today the same buyer who clicks a Google ad has already asked ChatGPT for alternatives, read a Reddit comparison thread, scanned X to see what insiders think, and checked LinkedIn for whether the founder of the company is credible. In many verticals Reddit threads now dominate Google's commercial SERPs, which means even traditional search is increasingly mediated by conversational sources the brand never paid for.
The platforms still own distribution. They no longer own the moment of decision.
That's why performance feels harder this year than last year. Demand hasn't dropped. It just moved into channels where traditional ads have weaker permission to interrupt.
Traffic is a visit. Intent is the reason behind the visit. They are not the same metric, and most performance dashboards conflate them.
A user searching best CRM for small B2B team gives you a keyword. A user asking ChatGPT something like:
We're a 7-person B2B team running outbound, investor updates and founder-led sales.
We need a CRM that doesn't turn into heavy enterprise software.
What should we use?
That gives you the entire qualification stack. Company size, workflow, constraint, alternatives, posture. That's what a conversation surface monetises. The ad isn't competing for the buyer's attention. It's appearing inside the buyer's reasoning.
The harder question is whether the model will name your brand at all when the answer gets generated. That's a different operation. We unpacked it in how to get cited by Perplexity, ChatGPT and AI Overviews. The short version is that paid placement doesn't fix it. Sponsored slot and cited source are two different surfaces. You want both. They require different work.
Intent does not appear at the moment of the click. By the time the buyer types the query, intent has already been shaped by what they've read, who they've trusted, what context they've absorbed.
The sources that shape it look like this:
Reddit threads
X replies and conversations
Threads posts and replies
LinkedIn posts and comments
YouTube comparisons
AI assistant answers
Comparison and review articles
Founder content
Community discussion
Product reviews
This is why "SEO" as a discipline is broadening. Google's own search starter guide defines SEO as helping users find and decide whether to visit a site. That's intent work, not keyword work. But Google is now one channel in a three-surface intent map:
Search intent Google, Bing, comparison pages
Social intent Reddit, X, Threads, LinkedIn, niche communities
AI intent ChatGPT, Perplexity, AI Overviews
Most marketing teams are still organised around the first surface. Some have a handle on the second. Very few are operating across all three coherently. That's the gap the next ad market is going to monetise.
A lot of teams still treat social as a brand channel. Post. Grow followers. Drive some clicks. That framing is two cycles out of date.
For most B2B and consumer categories in 2026, social conversations are the first layer of market research. Buyers read Reddit before they trust a landing page. They scan X to see what people who actually use the product say. They check LinkedIn to decide whether a company is credible. They use social proof to determine whether a brand even makes the shortlist.
We've spent the last 18 months operating in this layer directly through our Reddit Resident Network, coordinated X narrative engineering, and LinkedIn operator-voice presence. The pattern we see consistently is that brands cited in conversational sources outperform brands of comparable size that aren't. The lift shows up in assisted conversions, AI citation rate, and pipeline velocity from organic channels. Done badly, this layer leaves obvious fingerprints, which is why we wrote the bot-detection checklist read inverted as a residents playbook. Done well, it becomes infrastructure.
The clean way to think about the three surfaces:
ChatGPT Ads capture intent. SEO indexes intent. SWARM creates intent.
Different jobs, same chain.
| Dimension | Google Ads | ChatGPT Ads |
|---|---|---|
| Unit of intent | Keyword string | Conversation |
| Targeting input | Match types, audiences | Context hints describing situations |
| Buyer context | Sparse (the query) | Rich (the prior turns) |
| Optimal ad voice | Hook + offer | Continuation + relevance |
| Failure mode | Wasted match-type spend | Wasted intent mismatch |
| Reported CVR (Criteo, 500 retailers) | Baseline | ~1.5x baseline on LLM-referred traffic |
Same auction shape. Different lever. The keyword research muscle becomes intent definition muscle, and most performance teams do not yet have it.
Reach is becoming less valuable as trust drops. A banner can build awareness. A feed ad can earn a click. A search ad can capture demand. But a Reddit thread that ranks in Google, an X conversation that insiders quote, or an AI answer that includes your brand in the shortlist: all of that influences the buyer before the paid click ever happens.
You can see the substitution starting to bite in the data. When Reddit's own AI ads cut CPA 15% but did nothing for AEO visibility, the brands that won were not the ones that bought more inventory. They were the ones that were already inside the conversation the ad pointed at. Same audience, two different layers, only one of which compounds.
Ads aren't disappearing. The mix is changing. Companies will still buy paid search, paid social and retargeting. But the marginal budget dollar is starting to move into operations that build the conditions for conversion:
community presence
AI-search visibility
conversation monitoring
social proof at scale
comparison content
category narrative
founder-led distribution
LLM citation strategy
This is not "ads vs organic". It's the substrate that makes ads work better, or fail more expensively when it's missing.
Stop briefing in word counts and keyword density. Start briefing in decision contexts.
A bad brief reads like this:
Keyword: AI social media tool
Word count: 2,000
Include keyword 12 times
A better one reads like this:
Audience: B2B founder trying to grow on X without hiring a full social team.
Comparing: agencies, freelancers, AI tools.
Worries: sounding like a bot, brand safety, whether social produces pipeline.
Explain: options, tradeoffs, when AI agents make sense, when they don't.
The second brief produces a page that ranks in Google, gets cited by AI assistants, and works as a landing page for intent-based ads. The first produces an article that almost no one reads. We pulled apart the AI ranking mechanics in the CoinMarketCap AI ranking signals breakdown. The principle generalises.
The planning question is no longer "which audience and which keyword". It's: what decision is this buyer making, what context makes our product relevant, what objections have to be neutralised before the click, and where was this intent formed in the first place.
The ad creative inherits the same constraint. In ChatGPT Ads the ad runs inside a conversation, which means it has to read as a continuation of that conversation rather than an interruption to it. Banner voice is the fastest way to burn $5 CPCs. The full mechanics (bidding, ad groups, measurement, landing structure) are in our ChatGPT Ads setup guide.
A content calendar is not an intent strategy.
An intent strategy maps the public conversations that shape demand: which Reddit threads rank for the category, which X accounts define the narrative, which LinkedIn posts influence the buying committee, which comparison queries are showing up in AI answers, which objections repeat across communities, which competitors are being recommended by default.
The work then becomes building credible presence inside those environments. Not spamming links. Not running ads at communities. Becoming part of the conversation before the buyer reaches the website. On LinkedIn specifically, the gap is enormous. The reply-rate math on AI SDRs vs operator-voice outreach and the credibility delta between cold DMs at 1% and operator profiles at 15% are doing more of the work than any of the messaging tactics layered on top.
ChatGPT Ads is not just a new media placement. It's the first scaled signal that the ad market is moving from buying attention to engineering intent.
For marketers, the operating question is no longer "how do we get more clicks." It's "how do we become part of the decision before the click exists." That answer requires a stack that wasn't standard a year ago: SEO that answers intent, social that creates intent, AI-native ads that capture intent, analytics that connect intent to revenue, and a sales motion that closes high-intent buyers with the context already loaded.
The performance marketing era was about buying attention. The next era is about shaping intent. The companies that internalise this early won't just compete for clicks. They'll compete for the buyer's mind before the auction even starts.
Is performance marketing dead? No. Performance marketing still works for direct-response acquisition with measurable creative and clear offers. What's breaking is the assumption that paid acquisition alone can compensate for weak intent infrastructure. Brands without presence in the conversations that shape buying decisions are paying a hidden tax on every paid click.
What is intent marketing? Intent marketing is the discipline of identifying and influencing the moments where buyer decisions are formed (Reddit threads, X conversations, LinkedIn posts, AI assistant answers), rather than only buying placements that intercept buyers after the decision is already half-made. It treats conversation surfaces as the substrate that makes paid channels work.
How is ChatGPT Ads different from Google Ads? Google Ads targets keyword strings. ChatGPT Ads targets conversation contexts via "context hints" that describe the situation the buyer is in. Early data from Criteo shows LLM-referred traffic converting at roughly 1.5x other channels, consistent with higher intent density per click. The mechanics are similar; the lever moves from keyword research to intent definition.
What is conversation intent? Conversation intent is the full reasoning the buyer brings to a question: company stage, constraints, alternatives, budget posture, the alternatives they've already ruled out. A search query exposes a fragment of intent. A conversation with an AI assistant exposes most of it, which is why ChatGPT Ads has a structurally higher intent ceiling than keyword-based platforms.
Will AI replace performance marketing teams? No, but it changes what those teams optimise for. The performance team of 2026 spends less time on bid management and creative iteration and more time on intent mapping, landing-page-to-intent fit, and integration with the organic conversational layer. The teams that don't make this transition will keep hitting the same CAC ceiling.
Is intent marketing the same as LLM marketing? No, but they overlap. Intent marketing is the whole stack of identifying and influencing buyer decisions across SEO, social conversations and AI ads. LLM marketing is the subset focused on getting cited by large language models like ChatGPT, Perplexity and Claude. We cover the LLM-side playbook separately in how to get cited by Perplexity, ChatGPT and AI Overviews.
What's the difference between paid ChatGPT placement and being cited in a ChatGPT answer? Ads put you in the sponsored slot. Citation puts you in the answer itself. They use different signals and require different work. The paid side is straightforward (Ads Manager). The citation side is conversational-source presence. We covered the citation mechanics separately in how to get cited by Perplexity, ChatGPT and AI Overviews.
How do I start building intent infrastructure? Start with a citation audit on your top 20 commercial queries. See which sources Perplexity, ChatGPT search and AI Overviews currently cite, and whether your brand appears anywhere in them. From there, the operation is conversational presence in the sources that already rank, plus comparison content for the gaps no source has filled yet. Why monitoring dashboards alone won't get you cited explains the trap most teams fall into here.