ChatGPT Ads vs Earning the Answer: What a Founder Should Buy First

ChatGPT has ads. OpenAI began testing them in the US on February 9, 2026, per its own help centre, and eligible businesses can now buy placement below ChatGPT's responses through a self-serve Ads Manager. What no budget can buy is the answer itself: when ChatGPT recommends a brand inside its response, that position is earned through generative engine optimization (GEO), the discipline that also travels as answer engine optimization (AEO). A founder weighing the two is choosing between renting attention next to the answer and holding a place within it, and the sequencing matters more than most of the launch coverage admits.

I run GEO at Lore, where we sell the earned side for a living, so you should weigh my incentives while you read. My position is that organic AI visibility is the better first purchase for most founders, that ChatGPT ads deserve a place in the mix in specific situations, and that the deciding evidence is your own data rather than my preference or yours. The rest of this piece sets out what the ads can and cannot do, what the earned alternative involves, and how to choose.

Does ChatGPT have ads, and what do they cost?

OpenAI's test started in the United States for logged-in adults on the Free and Go plans; Plus, Pro, Business, Enterprise, and Edu accounts never see ads, which shapes who you can reach and comes back later in this piece. Per OpenAI's help centre, ads appear below the end of a response, labelled as sponsored and visually separated from the answer, and they are kept away from sensitive topics such as personal health and politics. OpenAI has since expanded the programme well beyond the US, and in its August 2026 announcement reported that the platform passed $1 billion in annualised revenue run rate in under 200 days, with availability in over 40 countries. The inventory and the competition for it are both growing.

Buying runs through the Ads Manager at ads.openai.com, structured as campaign, ad group, and ad, which will feel familiar to anyone who has run Google or Meta campaigns. The sharp difference is targeting: there are no keywords. You write context hints at the ad group level describing the conversations where your product is relevant, and OpenAI's system matches your ad to chats by expected relevance. The hints guide matching without guaranteeing delivery in any specific conversation.

On cost, OpenAI's advertiser documentation supports impression and click billing, recommends a starting maximum bid of $3 to $5 per click, and resolves placement through a relevance-weighted second-price auction. Search Engine Land's step-by-step guide from early campaigns, written by John Horn in August 2026, reports average CPCs in the $2 to $5 range across industries and daily budgets from $25. A meaningful test costs less than most founders assume, and that low entry price is itself a reason the auction will get more crowded.

What ChatGPT ads cannot buy

The sponsored slot sits below the answer. The recommendation inside the answer runs on a separate system, and OpenAI is explicit on this: ads do not influence ChatGPT's responses, and advertisers have no ability to shape, rank, or alter them. The Search Engine Land guide confirms the same point from the buying side, noting that running ads has no effect on whether a brand gets mentioned organically.

For a founder, this is the detail that defines the whole decision. When a buyer asks ChatGPT which product, platform, or provider to choose, the names inside that response were earned, and no bid lifts yours into it. The ad underneath reaches the same reader a moment later, with a sponsored label attached, after the answer has already framed the market for them. Both positions have value. They are different products, priced and won in different ways.

Earning the answer is the other purchase

A place inside the response is won through GEO work: content structured so retrieval systems can lift and attribute it, technical access so AI crawlers can read your site at all, third-party mentions that teach models what you do and who you serve, and monitoring that tracks how often the engines name you. I have mapped the discipline against its neighbours in GEO vs SEO and GEO vs AEO if the vocabulary is new.

The receipts for the earned side come from client work. BizScout, a business marketplace we work with, built over 450 AI mentions across twelve months of this programme, and ChatGPT now cites them as a go-to source for business buyers. Tides Mental Health reached 4.7x organic traffic growth in five months of an integrated programme, with AI engines recommending them for anxiety treatment. And the market numbers we plan against at Lore explain why the slot is worth contesting: organic search traffic is down roughly 30% across the board, while the traffic AI engines refer converts at 3 to 6x the rate of traditional organic. Fewer visits carrying several times the buying intent, and no middle ground: a brand is either in the response or invisible to that buyer.

Pay to play, or compound

Paid has a genuine place, and a startup that needs leads this quarter should take it seriously. The trade is structural. You are paying to play consistently, and to compound the lead funnel you also have to compound the spend; switch the budget off and the impressions stop the same day. Earned work runs in the opposite direction, since a page that wins citations today keeps paying at one month, six months, and next year, while the cost of producing it stays fixed.

There is a second difference founders underweight. Acquirers pay more for steady non-paid inbound than for an ad account the next owner must keep funding, which reframes AI visibility as an exit lever rather than a lead-gen line item. Running a business with no exit thinking at all strikes me as naive even if you intend to keep the company forever, and the channel mix you build is part of what a buyer eventually prices.

When ChatGPT ads are the right first buy

Four situations where I would put paid placement ahead of, or alongside, the earned programme:

• You need pipeline before earned work can mature. Citations take months to build, and I say so to clients before they sign. If the runway question is measured in weeks, rent the presence while the compounding asset gets built.
• You want message testing against live buying conversations. A headline and description served into relevant chats, at a few dollars a click, is a cheap laboratory for the positioning your earned content will later commit to.
• The answer in your category cannot be won yet. Where entrenched incumbents dominate the citations for every question that matters, paid placement is the realistic near-term route to appearing anywhere near the response while the mention-building does its slower work.
• Your buyers sit on the Free and Go plans. The inverse also disqualifies: if your audience lives on Plus, Pro, or business tiers, the ads cannot reach them at all, and the earned answer becomes the only surface available.

One performance caveat from the early data: in Search Engine Land's GA4 comparison of around 1,500 users, traffic from ChatGPT ads engaged less and converted at a far lower rate than traffic from Google Ads in their test. One advertiser's result settles nothing on its own, and the platform is changing monthly, so judge the channel by what a customer costs you through it. If you want the setup end to end on screen, Blake Bauer's beginner walkthrough covers account creation through first campaign:

<YouTube id='9f-oE1pLDRQ' />

How to decide: data over anyone's opinion

My opinion does not settle the sequencing question, and neither does yours. Three checks produce the answer for your specific business:

• Ask the engines your buyers' questions. Put your category's real purchase questions to ChatGPT and see which brands the responses name. If competitors are cited and you are absent, the earned gap is measurable and urgent. If nobody owns the answer yet, you have found whitespace worth taking before the auction below it gets expensive.
• Check where your buyers sit. Audience research tells you whether they use ChatGPT free tiers, paid tiers, or classic search, and search behaviour splits by age, education, and geography. The channel your specific audience uses decides where the budget starts.
• Give every pound a metric it answers to. OpenAI's reporting covers impressions, clicks, spend, and conversions, and UTM parameters carry through to your analytics. Hold the earned programme to the same standard through citation monitoring and referral tracking, because unattributed budget on either side is budget spent without knowing its value.

Sequenced properly, the two spends cooperate: ads deliver leads while the earned programme builds, and the earned asset later lets you dial paid down instead of ratcheting it up forever.

Frequently asked questions

Why am I getting ads on ChatGPT?

Ads appear for logged-in adults on ChatGPT's Free and Go plans in regions where OpenAI's test is live. Paid tiers never include them. Per OpenAI's help centre, you can turn off ad personalisation in Settings, switch to an Ads-Free version of the Free plan with lower usage limits, or upgrade to a paid plan to remove ads entirely.

How much do ChatGPT ads cost?

OpenAI recommends a starting maximum bid of $3 to $5 per click, with placement decided by a relevance-weighted second-price auction, and supports impression-based buying as well. Search Engine Land reports average CPCs of $2 to $5 across industries in early campaigns and daily budgets from $25, so a pilot sits within almost any marketing budget.

Can you pay ChatGPT to recommend your brand?

No. OpenAI states that ads run on systems separate from the chat model and that advertisers cannot shape, rank, or alter ChatGPT's responses. The brands named inside an answer get there through the signals GEO works on: citation-ready content, accessible sites, and third-party mentions that teach the models who you are.

Should a founder run ChatGPT ads before investing in GEO?

Run ads first when you need leads before earned visibility can mature, when your buyers use the free tiers, or when incumbents currently own your category's answers. Fund the earned side first when your buyers already ask AI engines purchase questions and the responses name competitors, because that gap compounds against you every month it stays open.

If you are deciding where the first pound goes, book a free 30-minute strategy call and we will put your category's questions to the engines together and look at what your data says before you spend on either side.