OpenAI Ads Manager Beta is live, CPC bidding shipped, and the JavaScript pixel is in production. The catch: ChatGPT Ads doesn't reward Google-style keyword logic. It rewards conversation intent. The setup, from access to first campaign.
ChatGPT is no longer just a place where people ask questions. It is where they shortlist vendors, compare alternatives, and stress-test a purchase before they ever click a link. That shifts the unit of intent from a keyword to a conversation, and OpenAI's new ads program is built around that shift.
In May 2026, OpenAI confirmed that advertisers can buy through partners or the new self-serve Ads Manager Beta, added CPC bidding, and shipped expanded measurement tooling. This is the playbook for getting in cleanly.
To launch ChatGPT Ads in 2026:
The rest of this guide is the version of that list with the pitfalls each step hides. For the bigger picture of where ChatGPT Ads sit between SEO and GEO, see SEO vs AEO vs GEO: the full search visibility stack.
ChatGPT Ads are sponsored placements that appear inside the ChatGPT interface during relevant conversations. Unlike search ads, where targeting is a bag of keywords, the targeting surface here is the conversation itself: the buyer's task, their constraints, the alternatives they have already named, and the moment they are in.
OpenAI describes the channel on the official Ads page as a way to appear when users are exploring, comparing, and deciding. The company has stated explicitly that ads are separated from organic answers, marked as sponsored, and that user conversations are not shared with advertisers. See OpenAI's Testing ads in ChatGPT post.
The honest read: this is a high-context, high-intent surface where the buyer is mid-decision. Done well, that means lower volume than Google but materially higher quality per click. Done badly, it burns budget faster than display, because the click costs you a strong moment with a serious buyer.
Ads Manager is in beta. OpenAI is opening access in waves and expanding regions gradually. The Ads Manager Beta overview describes it as a self-serve platform for creating, launching, and managing ChatGPT Ads campaigns, plus performance reporting.
If your account is not live yet, two paths work: apply through the official site, or run through an OpenAI Ads partner. Either way, this is not a queue you join silently. The teams who get prioritised are the ones with a real product, a clear use case, and a credible landing experience already up.
The account creation flow:
OpenAI's account setup guide is the authoritative walkthrough. Before you start, have your favicon, brand logo, billing profile, and team access list ready. The quickstart guide covers the same ground from the campaign launch side.
The hierarchy is familiar:
Campaign
Ad Group
Ads
The trap is treating it the same as Google Ads.
Campaign. Top level. Goal, budget, bidding strategy, overall logic.
Ad Group. A theme, intent cluster, or buyer situation. This is the most important layer in ChatGPT Ads, because the ad group is where you tell OpenAI's system what conversations you want to show up in.
Ad. The creative itself: copy, image, link, and offer.
OpenAI's Create Ad Groups for ChatGPT doc spells this out: ad groups include context hints that describe relevant conversations, topics, and keywords. These hints are not exact-match targeting rules. They are guidance the system uses to decide where you are a fit.
The implication: stop thinking in keywords. Start thinking in situations.
This is the single biggest difference between ChatGPT Ads and Google Ads, and it is where most teams waste their first month of budget.
A Google query:
best crm software
The same buyer in ChatGPT:
We are a small B2B sales team running outbound to founders and operators.
We need a CRM that handles follow-ups, threading, and the kind of
investor-style relationship tracking a normal sales CRM doesn't do well.
Budget is tight and we don't want to glue 3 tools together. What should
we use?
The second query carries the entire qualification stack: company stage, use case, integration pain, budget posture. Your ad has to meet that.
We group conversation intents into four buckets.
Buyer knows the pain, has not picked a category.
How do I get more B2B leads from social media?
Why are my LinkedIn posts not converting?
How do I scale founder-led content without burning out?
Buyer is shopping the category.
Best AI tools for social media marketing
Social media automation software
AI agents for marketing
Buyer is choosing between named alternatives.
AI social media agents vs an agency
Buffer vs AI automation
Should I hire an SMM agency or use AI?
Buyer is ready to book or buy.
Best tool to generate leads from X for B2B
AI marketing tool for an early-stage startup
Book demo social media automation
Each cluster deserves its own ad group, its own context hints, its own ad, and its own landing page. Stacking all four into one ad group is the cleanest way to never see what is actually working.
Most B2B teams will instinctively point ad clicks at their homepage. This is the most expensive mistake you can make in ChatGPT Ads.
The buyer arrived mid-conversation. Your landing page needs to continue that conversation, not restart it.
The pattern that works:
/use-cases/ai-social-media-agents
/use-cases/twitter-growth
/use-cases/b2b-lead-generation
/compare/ai-agents-vs-agency
/solutions/social-media-automation
A use-case page that mirrors a single intent cluster. Five sections, opinionated, written to one buyer:
H1 The specific problem or use case
Sub The outcome the buyer gets
1 Why the old approach breaks
2 How the new approach works
3 Proof: cases, screenshots, metrics
4 Who this is for (and who it is not)
5 How to start
CTA Book demo / start trial / get audit
This page is also an SEO asset. It can rank organically, your blog content can link into it, and a well-built use-case page is the surface AI search engines tend to cite when they answer category-level questions. We covered the citation mechanics in how to get cited by Perplexity, ChatGPT and AI Overviews.
A ChatGPT ad is not a banner. The user is mid-thought. The ad either reads as a useful continuation or as an interruption. Interruptions get scrolled past.
The classic banner voice:
The #1 AI Marketing Tool. Try now!
The continuation voice:
Automate social-media engagement with AI agents that find relevant
conversations, reply in your brand voice, and turn organic discussions
into pipeline.
A formula that consistently performs:
buyer's problem → specific outcome → why now → one CTA
A full example for our category:
Trying to turn X and LinkedIn into pipeline without hiring three people
to run them? NotPeople deploys AI residents that find relevant threads,
reply in your founder's voice, and route qualified replies to your sales
inbox. See a sample week of output.
That ad reads like a sentence the buyer would have written themselves. That is the bar.
Skipping tracking is how ChatGPT Ads quietly becomes "we spent money but cannot tell you what happened." Do not ship a campaign without it.
OpenAI supports two complementary methods:
| Method | What it does | When to use it |
|---|---|---|
| JavaScript Pixel | Browser-side events fired from your landing page after the click | Standard web events: PageView, Lead, Signup, DemoBooked |
| Conversions API | Server-to-server conversion events | Deeper funnel events: qualified lead, closed-won, revenue |
The events worth firing at minimum:
PageView
Lead
Signup
BookDemo
StartTrial
Purchase
QualifiedLead
For B2B, the surface-level events lie. Signups do not tell you if the pipeline is real. Send deeper events as your CRM enriches them:
demo_booked
form_submitted
sales_qualified_lead
paid_customer
The Ads Manager Beta reporting surface shows impressions, clicks, spend, CTR, average CPC, average CPM, and conversions. Treat that dashboard as the input to your real reporting, not the output.
The minimum analytics chain to set up before launch:
ChatGPT Ad Click
Landing Page Visit
Signup / Demo Request
Qualified Lead
Customer
Revenue
If you can run pixel and Conversions API in parallel, do that. Server-side is more reliable, browser-side gives you faster optimisation signal. They complement each other.
The first wave of ChatGPT Ads ran on CPM. OpenAI later added CPC bidding so advertisers can pay closer to action.
For performance teams, default to CPC. It chains cleanly into the metrics you actually care about:
CPC → landing CVR → CPL → SQL rate → CAC
CPM has a job for brand-led campaigns where the goal is visibility in a new category rather than direct lead flow. For most B2B teams shipping their first ChatGPT campaign, CPC is the answer.
The metrics to watch in week one:
| Metric | What "good" looks like | What to do if bad |
|---|---|---|
| CTR | Above category baseline | Test context hints + ad copy |
| CPC | Predictable, not erratic | Tighten ad group focus |
| Landing CVR | At or above your paid baseline | Rebuild the landing for the intent |
| CPL | Below your blended target | Filter audience earlier in funnel |
| Qualified lead rate | Within 20% of organic | Tighten messaging, raise commit gate |
| Pipeline / revenue | Trending up week-on-week | Promote ad group, kill the rest |
A minimum viable launch.
Step 1. Pick one use case. Not three. Not the whole product. One.
AI residents for B2B social-media lead generation
Step 2. Build a dedicated landing page. It answers this intent only.
Step 3. Build 3 to 5 ad groups. Each maps to one intent cluster:
AI social-media automation
X / Twitter lead generation
Social-media agency alternative
Founder-led content scaling
B2B organic acquisition
Step 4. Write context hints per ad group. Describe the situation:
People asking how to generate B2B leads from X/Twitter, automate
social-media replies, scale founder-led content, replace or augment
an SMM agency, or use AI agents for organic acquisition.
Step 5. Write 2 to 3 ads per ad group. Test the angles:
Save time
Generate leads
Replace manual SMM work
Scale founder voice
Automate relevant replies
Step 6. Wire conversion tracking. PageView + Lead + DemoBooked + Signup at minimum.
Step 7. Launch small. The goal of week one is learning, not scaling.
Which intents produce CTR?
Which ads earn clicks?
Which landing pages convert?
Which leads are actually real?
Do not optimise to CTR alone. A high-CTR ad can be a curiosity trap that does not convert. Watch the whole funnel:
Impressions → Clicks → CTR → CPC →
Landing CVR → CPL → Qualified lead rate → CAC → Revenue
What to test, ranked by impact:
1. Context hints (biggest lever)
2. Offer
3. Landing headline + first section
4. Ad creative angle
5. CTA
6. Conversion event definition
7. Ad group structure
Diagnostic rules of thumb:
| Dimension | Google Ads | ChatGPT Ads |
|---|---|---|
| Unit of intent | Keyword | Conversation |
| Targeting input | Keyword lists, match types | Context hints describing situations |
| Buyer context | Sparse (the query string) | Rich (the prior turns of conversation) |
| Optimal ad voice | Hook + offer | Continuation + relevance |
| Conversion path | Click → landing → action | Click → landing → action (same) |
| Failure mode | Wasted match-type spend | Wasted intent mismatch |
The shape is the same. The lever shifts from keyword research to intent definition.
1. Sending all traffic to the homepage. A homepage rarely answers a specific intent. Dedicated landings convert 2 to 4 times better in practice.
2. Porting Google Ads structure one-for-one. Keywords still matter, but they are an input to context hints, not the unit of organisation.
3. Skipping conversion tracking. Without pixel or Conversions API there is no ROI conversation, just spend.
4. Writing banner-voice ads. ChatGPT is a high-trust surface. Pushy ads underperform even when they get clicks.
5. Launching too broad. One use case plus 3 to 5 intent clusters beats five use cases with 1 to 2 clusters each.
6. Optimising to CTR. CTR rewards curiosity. You want pipeline.
If Ads Manager is not live for your account yet, the prep work compounds:
1. List your commercial and informational conversation intents
2. Build use-case landing pages for the top 3 to 5 intents
3. Ship blog content that reinforces topical authority
4. Wire analytics + server-side conversion events now
5. Prepare segment-specific offers
6. Write FAQ and objection-handling pages
7. Stress-test messaging on Google Ads, LinkedIn, or organic first
The strongest setup we see is one where SEO, organic citation, and ChatGPT Ads work as one funnel rather than three:
SEO article → use-case landing → demo / signup → retargeting / sales
Concretely:
/blog/how-to-automate-twitter-replies → /use-cases/twitter-reply-automation
/blog/best-ai-social-media-tools → /solutions/ai-social-media-agents
/blog/social-media-agency-alternative → /compare/ai-agents-vs-agency
The blog captures informational demand. The landing closes commercial intent. The ad amplifies the same pages inside ChatGPT.
ChatGPT Ads is not another keyword-buying surface. It is a performance channel where the targeting unit is the buyer's situation, not the buyer's query.
The teams that win here are the ones who already think in intents, build use-case landings, instrument conversions properly, and treat ChatGPT Ads as a continuation of work they are already doing on SEO and AI-search citation. For organic citation strategy, see Reddit owns Google for crypto and GEO dashboards vs acquisition.
The teams that lose port their Google Ads playbook word-for-word and wonder why CTR is high but pipeline is empty.
Minimum viable launch, again:
one use case
one dedicated landing page
3 to 5 intent-based ad groups
well-written context hints
2 to 3 ads per group
conversion tracking before launch
optimise on qualified leads, not CTR
How much do ChatGPT Ads cost? OpenAI hasn't published a fixed rate card. Pricing runs on CPC and CPM auctions, so cost is set by competition for your conversation contexts. Early-beta reports cluster around mid-single-digit CPCs for B2B intents, but expect this to move as more advertisers come online. Budget for learning, not for scale, in week one.
What is a context hint in ChatGPT Ads? A context hint is the description you write inside an ad group telling OpenAI what conversations and situations your product fits. It's not exact-match keyword targeting. It's closer to a one-paragraph brief on when your offer is relevant. See OpenAI's ad groups doc for the official frame.
Can small businesses and startups run ChatGPT Ads? Yes. There is no minimum spend tier published for Ads Manager Beta. The bigger constraint is whether you have a clear use case, a dedicated landing page, and conversion tracking in place. Without those three, ChatGPT Ads burns budget faster than Google Ads. With them, it works at startup scale.
Can I actually run ChatGPT Ads right now? Yes. OpenAI has launched Ads Manager Beta and continues to open access in waves. You can also work through OpenAI Ads partners.
Where is the ChatGPT ads dashboard? ads.openai.com. OpenAI's quickstart guide walks the campaign launch flow.
Is CPC bidding available? Yes. OpenAI announced CPC bidding so advertisers can pay per click rather than only per impression.
Can I track conversions? Yes. OpenAI supports a JavaScript pixel and a Conversions API. Run both if you can.
Do ads influence ChatGPT's organic answers? According to OpenAI, ads are kept separate from organic answers, labelled as sponsored, and user conversations are not shared with advertisers.
What is the difference between paid ChatGPT placement and being cited organically? Ads put you in the sponsored slot. Citation puts you in the answer the model gives. You want both, and they require different work. The organic side is covered in how to get cited by Perplexity, ChatGPT and AI Overviews.
Related reading: How to get cited by Perplexity, ChatGPT and AI Overviews · GEO dashboards vs acquisition · Reddit owns Google for crypto
Want help building the intent map and landings before you launch? Run a 30-minute audit. We'll map your commercial intents, score your existing landings against them, and tell you which 2 to 3 ad groups will move pipeline first. While you ship paid, the organic citation side is covered by our Reddit Resident Network and LinkedIn Resident Network.