Advertise in ChatGPT is an advertising feature OpenAI has started testing, letting brands buy placement that appears directly inside ChatGPT conversations, instead of the old-style banner on the side of the screen. The key difference is that ads are “embedded” within the answers the AI recommends, based on the context of what the user is discussing — not fired off randomly the way typical ads are. This is still early days, and real pricing and results data isn’t clear yet. Worth watching for now, but no need to rush in with budget.
(Note: this article focuses on Advertise in ChatGPT — there’s no directly related numerical data from the provided research, so it’s written qualitatively rather than citing statistics.)
What ads in ChatGPT look like right now
From what’s visible so far, ads don’t come as typical banners. Instead, they show up as product/service cards inserted below the answer, right where it’s relevant to what you asked.
For example, ask about running shoes, and the main answer comes first, followed by a sponsored card with a link to the store. The placement doesn’t blend into the answer content itself — it’s a clearly separate block you can click through to see the product.
The upside is that it appears based on the actual conversation context, not fired off randomly like typical ads.
When website traffic disappears because people stop clicking out of AI
Many site admins have started noticing the same thing: traffic from Google is dropping even though people are still searching for the same things — because they’re asking ChatGPT instead, and getting a complete answer right in the chat without needing to click anywhere.
This is what’s known as the zero-click problem — people get the full information they need, with no reason to click through to a website like before. Stores that rely on SEO alone are starting to feel like no matter how good their content is, they can’t compete with a ready-made answer inside AI.
The question that follows is: if SEO can’t drive people to a website like it used to, how does a business show up in front of someone who’s actively deciding to buy? This is the gap Advertise in ChatGPT steps into — not competing with the AI, but embedding itself right inside the answer.
Where OpenAI is positioning this ad product in its overall business
Right now OpenAI makes money from ChatGPT Plus/Pro, the API for developers, and Enterprise for large organizations — all subscription or pay-per-use models. Advertise in ChatGPT is a fourth leg that’s entirely different, since the revenue comes from brands, not directly from users.
Opening this channel comes alongside the ongoing pressure of high AI-running costs — subscriptions alone may not be enough. But the risk is just as clear: if ChatGPT’s answers start smelling like ads, the trust it has built up could be shaken.
This is exactly the thing worth watching — how OpenAI balances revenue against the neutrality of its answers, given that ChatGPT’s core selling point from the start was giving trustworthy answers, not search results with paid placements mixed in.
From experimental feature to a full advertising platform
At the start, OpenAI rolled out the ad feature as an experiment, limited to a small group of users, and it was just small sponsored products or shopping cards inserted into answers. Back then it felt more like a pilot test than a real product.
Now, the picture has changed a lot — it’s become a more open advertising system, easier for advertisers to access, with more refined targeting drawn from conversation data, no longer just showing up at random.
| Factor | Experimental version (initial) | Current version |
|---|---|---|
| User base | Limited to a select group | More open |
| Ad format | Sponsored product / shopping card | Full advertising system |
| Targeting | Basic | More refined, based on conversation context |
| Advertiser access | Limited | Open |
The shift from “let’s try it out” to “running for real revenue” is a much clearer direction now.
What kinds of businesses can this actually help in practice
Let’s break it down into real-world scenarios — easier to understand than a spec sheet.
Someone searching for a restaurant nearby → gets a sponsored answer inserted into the response, similar to what we’re used to seeing from Google Ads, just moved into the conversation instead.
Someone shopping for something, like “good running shoes under $60” → ChatGPT shows a product card with a purchase link. This is where agentic checkout comes in, since the AI can carry the whole thing through to a completed transaction without opening a new website.
The most interesting part is small brands without an SEO team — previously they had to fight hard for Google rankings, but this system uses conversation context to determine the target audience instead, letting them compete on “relevance” rather than “SEO budget.”
That makes it clear why more refined targeting is the main selling point of this round.
If not ChatGPT, what are the alternatives?
Google Ads is still the market leader for reach, since people are searching there every day anyway. Meta Ads is strong on targeting, built on years of accumulated social behavior data.
ChatGPT Ads is new, with a much smaller user base, but its distinguishing feature is that targeting comes from real conversation context, not old click history.
Perplexity, if it launches a serious ad program, would be another option in the AI-native category, but right now its scale doesn’t compare to the three major players.
On brand safety, Google/Meta have had content-control systems in place for much longer, while AI platforms like this one still have to prove how well they can keep ads from blending into genuine answers.
| Factor | Google Ads | ChatGPT Ads |
|---|---|---|
| Reach | Much wider, daily search base | Smaller, still early stage |
| Cost per click | Highly competitive, expensive in some categories | No clear pricing data yet |
| Targeting accuracy | Based on search/click history | Based on real-time conversation context |
| Brand safety | Older, established content-control systems | Control systems still unproven |
Pros and cons to weigh before putting money in
Before deciding to commit budget to Advertise in ChatGPT, you need to look at both sides fully, because this is a new channel where the rules aren’t settled yet.
The clear upside is reaching people who are “genuinely asking” — not just scrolling past — unlike ads in a typical feed. Competition in this space is still low, since most brands are still hesitant and watching from the sidelines, giving early movers a shot at less crowded space. Another point is that it ties into the agentic commerce trend, where AI helps make purchase decisions for people — if a brand positions itself in this system early, it gains a long-term advantage.
On the downside, real measurable performance data is still thin, making it hard to judge whether it’s worth it. There’s a risk of being seen as making AI answers biased because ads are inserted, hurting trust. And the measurement/analytics systems aren’t as mature as long-established ad platforms.
Pros
- +Reaches people who are genuinely asking, not just scrolling past
- +Competition is still low, space isn't crowded
- +Ties into the long-term agentic commerce trend
Cons
- −Real measurable performance data is still thin
- −Risk of AI answers being seen as biased because of inserted ads
- −Measurement systems aren't as mature as established ad platforms
How much it actually costs before you see results from ads in ChatGPT
The ad spend itself is just the starting number — the real cost is hidden behind the scenes. You need to organize product content into structured data that the AI can read and understand, not just reuse the same nice photos and captions you used on social media.
Marketing teams or agencies need to learn an entirely new system, from how the AI selects answers to how conversions are calculated — none of which works like Google Ads or Facebook Ads. Expect to lose time in the early stages.
Even harder is that attribution isn’t settled yet — it’s difficult to measure how many people who clicked from an AI answer actually made a purchase, compared to existing channels that already have clear dashboards.
In the early stage of a new platform, results tend to be highly volatile — the budget you put in may not deliver consistent results. Go in expecting this to be a long-term investment to position yourself ahead of competitors, rather than expecting immediate returns.
What kind of business should try this now, and what kind should wait and see
Made for
- E-commerce businesses that already have a product feed ready — easy to build on, low-cost to test
- Brands whose customers search with high intent, like looking for specific products or comparing before buying
- Marketing teams with a separate experiment budget apart from the main budget, able to absorb early-stage risk
Think twice
- SMEs with limited budget should set aside a small amount to test first, rather than committing fully from round one
Skip this one
- Businesses that need certain, measurable ROI before investing — wait until attribution is more settled
- Brands whose image is sensitive to being seen as "paying to insert answers" — should watch the situation first
Questions to answer before shifting budget here
Before putting budget into Advertise in ChatGPT, ask yourself three questions: does our customer base actually search via AI chat, does the team have the bandwidth to track results in real time, and how much brand damage would a misstep cause?
There’s no single answer to this question for every business.
One angle worth considering: new advertising channels like this usually have an “early adopter window” where costs are lower than once the market gets crowded — so the real question isn’t just “are we ready,” but “how long until competitors move in to fight for the same space.”
If your business falls into the “consider” group above, setting aside a small test budget now might be worth more than waiting. If you’re in the “not for” group, hold onto this question and revisit it every quarter.
Take a look at your marketing budget today — is there 5-10% you could set aside to test a new channel?