ChatGPT Ads for US Businesses: Costs, Formats, and Who Actually Wins
What ChatGPT ads cost in 2026 ($25/day minimum, $3–5 starting bid), how conversation-intent targeting works, which US verticals win, and a 30-day test plan.
In short
ChatGPT Ads launched on February 9, 2026, opened a self-serve platform at ads.openai.com on May 5, 2026, and reached a $1 billion annual run-rate by August 31, 2026. You buy sponsored answers and cards inside the conversation, targeted by what the user is trying to accomplish rather than by keyword. Entry is cheap at $25 a day, and OpenAI recommends starting bids of $3–5. Clicks are also rare: one advertiser reported a 0.91% click-through rate against a 6.4% benchmark for Google Search in the same sector. That trade is worth making when your lead is expensive and your sales cycle is long, because a person who clicks after a conversation is far closer to buying than one who clicks a headline. The US verticals that win first are B2B SaaS, fintech, legal, healthcare, and high-ticket ecommerce.
What ChatGPT Ads are (and are not)
ChatGPT Ads are paid placements inside the ChatGPT conversation itself. They appear as sponsored answers or cards, clearly labeled as advertising, in the flow of a user’s chat. They are not banners in a sidebar and not a bidding war for a keyword slot above ten blue links.
Three dates matter for a US advertiser:
- February 9, 2026: advertising launches in ChatGPT (TechCrunch).
- May 5, 2026: the self-serve platform at ads.openai.com opens (Axios), so any verified business can set up campaigns without a sales rep.
- August 31, 2026: OpenAI’s ad business hits a $1 billion annual run-rate (PYMNTS).
That last number tells you the channel is past the experiment stage. Budgets that big do not come from curiosity; they come from advertisers who saw the numbers and came back.
One more thing ChatGPT Ads are not: a checkout. OpenAI retired the initial version of ChatGPT Instant Checkout on March 24, 2026, and shifted ChatGPT back toward product discovery, with the transaction handled on the merchant’s own site (CNBC). The operating model is “discover in chat, transact on site.” If you sell products, the feeds that matter are ACP, built by Stripe, OpenAI, and Meta, and UCP, built by Google, but the transaction still happens on your page. Your landing page is still doing the closing.
Targeting by conversation intent, not keywords
This is the part that trips up teams coming from Google Ads. In search, you bid on words. In ChatGPT, you target the topic of the conversation and the intent behind it.
Consider what that means in practice. Someone typing “best payroll software” into Google is one of a thousand people who typed the same three words. Someone in ChatGPT has written four paragraphs about a 40-person company in two states, contractors in a third, a bookkeeper who quits every spring, and a budget ceiling. ChatGPT knows the context. Your ad can be matched to that context, not to a string of characters.
The practical consequences:
- Your creative has to answer the situation, not the query. A headline that repeats the keyword is wasted. A headline that names the problem the user just described gets read.
- Negative matching works differently. You are excluding topics and intents, not misspellings. You will spend the first two weeks learning which conversations your ads should never appear in.
- Volume forecasting is harder. There is no keyword planner telling you how many people will describe your exact problem this month. You learn volume by running.
If you want the step-by-step mechanics of the platform, read how to launch ChatGPT Ads step by step. This article is about whether you should, and what it will cost you.
What ChatGPT Ads cost in 2026
The numbers below are the working benchmarks as of September 2026.
| Metric | ChatGPT Ads | Notes |
|---|---|---|
| Minimum budget | $25 per day | Low enough for a serious test in any vertical (OpenAI help center) |
| Recommended starting bid | $3–5 | OpenAI’s own guidance for new advertisers |
| CTR (one advertiser, one sector) | 0.91% | Versus a 6.4% Google Search benchmark in the same sector (Campaign, citing Adthena) |
| Manager accounts | None | Advertiser verifies its own entity |
A $25 daily minimum means a 30-day test starts at $750 in media. Nobody serious runs at the floor, but the floor is not the barrier. The barrier is the click price: clicks are not cheap, and reported click-through rates run well under 1%, so you will pay for a lot of impressions per visit. That is exactly why the channel only makes sense when each visit is worth a lot.
Why a 0.91% CTR is fine
A Google Search ad got about 6.4% CTR in one sector. A single advertiser’s ChatGPT ad got 0.91% in that same sector, according to Adthena data reported by Campaign in March 2026, not a platform-wide average. On paper that is a seven-fold gap. In practice, the two clicks are different products.
A search click comes from someone who scanned headlines for a second and picked one. A ChatGPT click comes from someone who has described their problem, read an answer, and decided your offer fits the answer. The industry shorthand is that AI traffic runs 4–5 times “warmer” than organic, because the user arrives after a dialogue, not after a headline.
The published conversion data backs that up:
- Opollo: 14.2% conversion from AI traffic versus 2.8% from organic, from a 2026 AI Search Benchmark Report on 312 B2B technology firms.
- Semrush: AI-referred visitors convert 4.4 times better, per Semrush’s July 2025 study.
- Ahrefs: 23 times more sign-ups per visit from AI sources, based on Ahrefs’ own traffic.
- Shopify: AI visits convert nearly 50% higher on product detail pages, per Shopify’s own merchant data.
These are different companies measuring different things, so do not average them into a single promise. What they share is direction: the click is rarer and it is worth more. If your funnel math is built on volume of cheap clicks, ChatGPT Ads will look terrible in the first report. If it is built on cost per qualified lead, the channel starts to look like the best inventory you can buy.
The “no agency manager account” rule
OpenAI made one structural decision that changes how US businesses work with agencies: there is no agency manager account.
Here is how it works instead:
- Your company creates the account at ads.openai.com and verifies its own legal entity.
- Country, currency, and time zone are set at verification and locked permanently (OpenAI’s help center). A US entity bills in US dollars, forever.
- The agency is invited into your account as a user. It builds, runs, and reports inside your property. It does not own the account, the data, or the billing relationship.
Why this matters for you:
- You own everything. Campaign history, audience learnings, conversion data: all of it stays with you if you change agencies. This is better than the Google and Meta norm, where agencies sometimes hold accounts hostage.
- Onboarding takes longer. Someone with legal authority at your company has to complete verification. Budget a week, not an afternoon.
- Verify under the right entity. If you have a holding company and an operating company, or a US entity and a Canadian one, pick carefully. You cannot change country or currency later.
- New accounts start in one country. A fresh account can advertise only in its home country at first; other countries open automatically once verification clears and the account reaches a required spend there, an amount OpenAI does not publish (OpenAI help center).
- Billing is direct. Your card or invoice terms, your receipts. The agency bills its fee separately.
If an agency tells you it can run your ChatGPT campaigns from its own master account, it either has not opened the platform or is describing something else.
Which US verticals should test ChatGPT Ads first
The channel rewards businesses where one good lead pays for a lot of impressions. Here is how the main candidates stack up.
| Vertical | Fit | Why | Watch out for |
|---|---|---|---|
| B2B SaaS | Strong | Long research phase, users describe their stack and pain in detail | Free-trial funnels need a nurture sequence, not just a click |
| Fintech | Strong | High lifetime value, comparison-heavy buying | Compliance review on every creative |
| Legal | Strong | People explain their situation to ChatGPT before they call a lawyer | Ethics rules on advertising vary by state bar |
| Healthcare and clinics | Strong | Symptom and treatment conversations are high-intent | Sensitive-topic restrictions; landing page must not overpromise |
| High-ticket ecommerce | Good | Considered purchases where the user asks for recommendations | No in-chat checkout; product page has to carry the sale |
| Real estate | Good | Relocation and financing conversations reveal intent early | Geographic targeting is coarser than local search |
| Local services (plumbing, HVAC) | Weak for now | Emergency intent still goes to search and maps | Test only if ticket size is high (remodels, solar, roofing) |
If you sell something under $100 with a same-day decision, stay on Google and Meta. If your average deal is four figures or higher and the buyer researches for weeks, ChatGPT Ads deserve a test budget.
A 30-day test plan that produces a real answer
Most ChatGPT Ads tests fail before launch: too little budget to learn anything, or a landing page built for keyword traffic. Here is a structure that avoids both.
Before day 1: fix the landing page. The visitor arrives mid-conversation. Your page should continue that conversation: state the problem in the user’s language, show the answer, make the next step obvious. Also make sure ChatGPT can see your site at all: check whether your robots.txt blocks OAI-SearchBot and GPTBot. About 73% of sites have technical barriers, including robots.txt blocks, CDN rules, and JavaScript rendering, that keep AI crawlers out, according to Otterly.ai’s 2026 AI Citations Report. Details in how to get your site indexed by ChatGPT.
Week 1: breadth. Run against a broad set of conversation topics with two or three creative angles. Spend enough per day that you see impressions across every topic you targeted. The goal is not conversions yet; it is a map of where your ads show and which conversations are irrelevant.
Week 2: cut. Exclude the topics that produced clicks but no engagement on site. Double down on the two or three intents with the longest time on page or the most form starts. Rewrite the weakest creative.
Week 3: depth. Concentrate spend on what survived. Add a second landing page variant for the top intent. Start tracking cost per qualified lead, not cost per click.
Week 4: decide. Compare cost per qualified lead against your Google Search and LinkedIn numbers. Not cost per click, not CTR: cost per lead your sales team accepted.
Budget logic. Set the daily budget high enough that a week produces a readable number of clicks in each topic group, and hold it steady rather than changing it daily. The channel is priced on CPC and CPM; a daily budget that only buys a handful of clicks will not tell you anything by day 30. Treat the first month as tuition, and size it so the lesson is worth learning.
How MOVE runs ChatGPT Ads for US clients
We run AI advertising as an add-on to the search and paid social work most clients already have with us, because that is where the comparison data lives. You verify your entity and invite us in as a user; we build the topic map, write the creatives, wire up conversion tracking, and report cost per qualified lead against your other channels every week. Management is priced from 15% of ad spend. If you are not ready to spend, we start with an AI visibility audit from $1,500 that shows where you already appear in ChatGPT answers, where you do not, and what is blocking you. Google Partner and Meta Partner, with a track record in B2B, fintech, clinics, and ecommerce.
6 mistakes that burn a ChatGPT Ads budget
- Copying your Google keyword list into the topic targeting. Keywords are not intents. You will show up in conversations that mention your term but have nothing to do with buying.
- Judging the channel on CTR. You will lose against search every time on that metric and miss the point. Judge it on cost per qualified lead.
- Sending traffic to your homepage. The user arrives with context. A generic page throws it away.
- Verifying under the wrong entity. Country and currency are locked forever. Get legal and finance in the room before verification.
- Testing at the $25 floor. You can, but at that spend and a sub-1% CTR, you will finish the month with too few visits to make a decision.
- Blocking AI crawlers while paying for AI ads. If GPTBot and OAI-SearchBot cannot read your site, you are paying to send users to a page ChatGPT cannot verify or describe.
Where to start
If your average deal is worth four figures and your buyers research before they call, ChatGPT Ads are worth a structured 30-day test. Start with the entity decision, fix the landing page and crawler access, then run breadth-cut-depth-decide. If you want a second pair of hands, book an AI visibility audit or see how we structure AI advertising engagements. More guides, including the launch tutorial and indexing checklist, are in the blog. Questions before you commit: contact us.