ChatGPT Ads in 2026: Costs, Eligibility, and How to Run Them

What are ChatGPT Ads?

ChatGPT Ads are paid placements that surface inside ChatGPT conversations, labeled as sponsored and shown below the answer rather than inside it. OpenAI has said ads appear for logged-in Free and ChatGPT Go users; paid Plus, Pro, Business, Enterprise, and Edu tiers stay ad-free. The ad is matched to what the conversation is about, not to a keyword you bid on.

The reason the channel matters is scale plus timing. ChatGPT reports on the order of 900 million weekly users and billions of prompts a day (OpenAI), and chatgpt.com now sits among the most-visited sites on the web (Similarweb). Its users skew younger and college-educated and lean on it for practical guidance and research (Pew Research; NBER working paper #34255) — including a meaningful share of pre-purchase product research. An ad inside that answer is in front of a decision as it forms.

How is it different from Google and Amazon ads?

Ads inside AI answers are no longer OpenAI alone — Google runs ads in its AI Overviews and AI Mode, and Amazon's shopping assistant works the checkout end. The three do not fight for the same spot; each owns a different stretch of the buyer's journey.

The money is following. eMarketer estimates US AI ad spend growing from roughly $32B in 2026 to $68B by 2030, most of it beside AI content. Named brands already testing ChatGPT ads in public reporting include Best Buy, AT&T, Expedia, Qualcomm, Pottery Barn, and Canva — no longer a fringe experiment, but a channel major brands are seriously trialing.

Who can advertise on ChatGPT Ads?

Eligibility is set by market and by how you buy. As of 2026 the ad inventory is open in the US, Canada, Australia, New Zealand, and the UK, with Japan and South Korea rolling out and other markets in progress. The EU is not yet live, held up mainly by GDPR's prior-consent rules; once it opens, personalization is expected to be limited to users who explicitly consent.

The practical takeaway for a founder going global: if your customers would Google your category and discuss it in ChatGPT, you are likely a fit — but if you do not have a US advertiser entity, your real choice is an agency or a managed service, not the self-serve dashboard.

How much do ChatGPT Ads cost?

ChatGPT Ads are bought on a cost-per-click (CPC) or cost-per-thousand-impressions (CPM) basis, and there is no fixed minimum spend to start. Early figures widely reported through 2026 put CPCs around $3–5 and CPMs near $25 at launch, with pricing drifting toward CPC as more advertisers enter and competition normalizes. Treat these as directional, launch-era numbers — they vary by category and by how much buying intent overlaps with your product.

For context on click-through: Google Search's all-industry CTR runs about 3–6.6% (WordStream), and social feeds sit well below that. ChatGPT ads land in the same neighborhood as search — meaningfully above a feed — because the click comes from someone already reading a synthesized answer, not scrolling. The honest read on cheap clicks is that a low CPC usually means you matched into the wrong conversation; judge the channel on cost per qualified outcome, not cost per click alone.

How does ChatGPT Ads targeting work?

Targeting on ChatGPT is not keyword bidding and not social personas. There are no interest tags, lookalikes, or customer lists to upload. Instead you give the system context hints — the topics and scenarios you want to appear in — and delivery is decided by how well your ad fits the live conversation, your bid, and predicted click-through. In already-open markets, finer personalized targeting is rolling out on explicit user consent.

Many buyers read "no interest tags" as weaker targeting. It is the opposite: legacy targeting guesses who you resemble from the past, while ChatGPT reads what a person is asking right now. Someone who just typed "compare these two flagship phones" is more certainly in-market than any modeled segment. Your three levers are the product description, the audience-and-scenario context, and your brand positioning — write the ad as a useful addition to the question, not a generic slogan.

• Context hints Free-form topics/scenarios that tell the system "show me in this kind of conversation." This is your primary steering wheel.
• Product feeds For catalog advertisers, product-feed ads let you upload item_id, price, availability, and GTIN so your product enters the comparison set inside the answer.
• Relevance as targeting The closer the creative sits to the question, the more it serves and earns clicks. Precision comes from fit, not from a segment you pre-declared.

What creative and review rules should you know?

Most downtime on this channel comes not from bids but from creative review. The single most common rejection reason is exaggerated or absolute claims, so write substantiated copy. Regulated categories need extra clearance. Build review time into your launch plan — a first review typically takes hours to a few days, longer in peak periods.

• Edits trigger re-review Changing a creative or landing target usually triggers a full re-review and can pause delivery for days. Bid and budget tweaks generally do not.
• Stage creatives up front Prepare several compliant creative variants before launch and swap to a standby when one fatigues, rather than editing a live creative into re-review.
• Fewer live edits win On this channel, staying live and not interrupting yourself is itself an advantage — both the auction and the review favor accounts left alone to accumulate history.

How do you measure ChatGPT Ads?

Turning high click-intent into provable value depends on getting conversion data back. The platform appends an OpenAI click identifier to your landing URL and writes a first-party cookie; you match conversions back to the click by that identifier. A server-side conversions feed is the reliable path: it survives cookie loss, covers in-app and off-site checkout, and is the prerequisite for the conversion-optimized (CPA) bidding OpenAI opened in 2026.

Independent reads suggest visitors from AI conversations convert well — Adobe Analytics reported AI-referred retail traffic converting materially better than ordinary traffic. But before your own samples are large, the cleanest way to prove the channel caused the lift is a geo holdout: run comparable regions with and without the channel on, and read the difference.

Is ChatGPT Ads right for you? A five-question fit check

A quick self-assessment. The more of these you can answer yes, the more the channel deserves a serious test. If only a few, start low-risk on your best-fit product line.

1. Do your users research this category in ChatGPT? The larger the addressable market and the more conversations touch your topic, the more chances you have to surface.
2. Can it play a mid-funnel, consideration role? ChatGPT Ads shine at getting onto the shortlist, not at closing an impulse purchase. Research-heavy, higher-ticket categories fit best.
3. Do you have an English landing page that converts on arrival? A multi-step signup leaks the high-intent click. Keep the path from ad to value short.
4. Can you wire server-side conversion tracking? Without it you only see clicks, not installs or revenue — and you cannot unlock pay-per-conversion bidding.
5. Can you commit a full quarter and stable, compliant creative? Roughly 90 days lets the account, benchmarks, and review history mature — which is where the durable value is.

How to run ChatGPT Ads without a US entity

Because self-serve is US-only, most brands going global cannot use the dashboard directly. The real comparison is not "self-build vs outsource" — it is an ordinary agency versus a managed service. The difference is how much learning cost and review downtime you carry, and whether the first-party data stays in your own account.

• Self-build Reasonable if you have a US advertiser entity and an in-house team. You own the ramp, the benchmarks, and the review downtime.
• Ordinary agency Absorbs some of the learning curve, but data ownership and benchmarks often live on the agency's side.
• Managed service Compresses the ramp, absorbs review downtime, and — done right — keeps the account and first-party data yours.

Two things to settle before you start, whichever path you pick: cross-border conversion data must be handled per regime (PIPL, EU/UK GDPR, and California CCPA/CPRA differ on consent and transfers), and brand-adjacency risk in AI answers is not yet third-party verified — bake compliance and brand-safety monitoring into the plan.

Where Ceres fits, honestly

Running ChatGPT Ads well is an operating discipline — account setup, conversion plumbing, creative review, staging, and steady optimization — more than a bidding trick. That is the loop a paid-search manager would own day to day.

Ceres is a managed AI marketing team for indie founders and small teams. An AI Growth Officer coordinates specialists, and the one that owns this channel is the Paid Ads Manager. Through the ChatGPT Ads integration, it connects to your own advertiser account, reads campaign and ad-group performance — impressions, clicks, spend, CTR, CPC, CPM — on a schedule, and drafts changes with the numbers attached. Every spend change is approval-gated: you approve it with one click, and nothing touches your account until you do. There is no autonomous spend mode.

Two honest caveats. First, ChatGPT Ads is a young channel; treat early performance reads as directional, not settled. Second, Ceres reads and drafts — it does not run your budget on its own; you connect your own ChatGPT Ads account and spend stays on your payment method. If you want the monitoring and drafting handled while you keep the final say, you can start the free trial (14 days, no card), or read how a small team can grow SaaS without a marketing team.