ChatGPT Ads Are Here. The Measurement Isn't.
- Heidi Schwende
- Jan 19
- 8 min read
Last September, I wrote about OpenAI's $25 billion advertising bet. Back then, people asked if I was being dramatic.
On January 16, OpenAI confirmed they're testing ads in ChatGPT for U.S. users on the free tier and the new $8/month Go subscription. Ads will appear below responses, labeled and separated from answers.
Sam Altman posted on X the same day:
"It is clear to us that a lot of people want to use a lot of AI and don't want to pay, so we are hopeful a business model like this can work."
So here we are.
What's Actually Happening
Free users and Go subscribers will start seeing ads at the bottom of ChatGPT responses in the coming weeks. If there's a relevant product or service tied to your conversation, a sponsored listing will appear below the answer.
The paid tiers stay ad-free
Plus ($20/month)
Pro ($200/month)
Business, and Enterprise customers won't see ads.
Two-tier experience: pay to avoid ads, or accept them as the cost of free access.
Users under 18 won't see ads. Neither will conversations touching politics, health, or mental health. You can dismiss ads and turn off personalization entirely if you want.
OpenAI made a point of explaining what ads won't do. They outlined five principles:
Mission alignment:
Advertising supports making AI accessible, not replacing their core mission.
Answer independence:
Ads won't influence ChatGPT's responses. Answers stay optimized for usefulness, not commercial outcomes.
Conversation privacy:
Conversations stay private from advertisers. User data won't be sold.
Choice and control:
Users can turn off personalization and clear ad-related data.
Long-term value:
No optimization for time spent or engagement loops.
Those are the promises. Whether they hold over time is a different question.
Why Now
OpenAI hit a $20 billion annualized revenue run rate by late 2025. They still spend more than that. Microsoft's financial filings show OpenAI lost over $11.5 billion in Q3 alone. They've committed to $1.4 trillion in infrastructure deals. Only about 5% of ChatGPT's 800 million users pay for subscriptions.
Meanwhile, Google pulled in $74 billion from ads in Q3 2025. Meta made $50 billion. That's the business model OpenAI is eyeing.
The person running this initiative is Fidji Simo, OpenAI's CEO of applications. Before OpenAI, she ran Facebook's advertising business, then served as CEO of Instacart. She knows how to build ad products. Her involvement signals this isn't an experiment they expect to fail.
Sam Altman has publicly said he "hates" ads and called combining them with AI "uniquely unsettling." But he also said in late 2024 that OpenAI would "try ads at some point."
That point arrived.
As Jeremy Goldman, an analyst at eMarketer, put it:
Ads aren't a distraction from the gen AI race; they're how OpenAI stays in it.
As I noted in December, OpenAI's internal documents project $1 billion from "free user monetization" by 2026, growing to nearly $25 billion by 2029. The announcement confirms the timeline is real.
What It Looks Like
OpenAI shared two examples of ad formats:
Image credit: openai.com
The first shows a ChatGPT response with Mexican dinner party recipe ideas. Below the response, a sponsored product recommendation appears for a grocery item (hot sauce). Clearly labeled as "Sponsored" and visually separated from the answer.
Image credit: openai.com
Image credit: openai.com
The second shows travel advice about Santa Fe, New Mexico. A sponsored lodging listing appears below the response. This example also shows a follow-up screen where users can start a direct chat with the advertiser's bot to ask questions about the listing.
The placement is deliberate. Ads aren't inside the response or interrupting the conversation. They show up at the end, after you've gotten your answer.
The format matters for one specific reason: ChatGPT isn't a search results page. When someone asks "What should I cook for a dinner party?" they're not scrolling through ten blue links. They're getting a direct answer. An ad below that answer is a different animal than a sponsored result in Google.
Whether that context makes ads more valuable or more intrusive depends on execution.
The Measurement Gap
Here's where executives need to pay attention.
OpenAI has not announced buying models. No CPM, no CPC, no CPA structure. No targeting details beyond contextual relevance to the conversation. No measurement framework. No access to reporting or attribution tools.
This is a test, not a platform.
I've been saying for months that the measurement infrastructure for AI advertising doesn't exist yet. Perplexity learned this the hard way. They launched ads in late 2024, charged premium CPM rates, then stopped taking new advertisers a year later because they couldn't prove ROI. Advertisers couldn't track CTRs or ROAS with any reliability.
OpenAI's September 2025 job posting mentioned building real-time attribution pipelines. That tells you where they are in the process. They're hiring people to build the measurement systems. Those systems don't exist yet.
So if you're wondering whether ChatGPT ads will show up in your analytics dashboards; not anytime soon. There's no Google Analytics integration announced. No conversion tracking framework confirmed. No API connections to your CRM or attribution tools.
The technical infrastructure being developed includes campaign management tools and attribution reporting. But being developed and available to advertisers are not the same thing. Until OpenAI launches a self-serve platform with tracking capabilities, you're flying blind on performance.
One note on the OpenAI-Google Cloud partnership announced in mid-2025: that deal is strictly about computing infrastructure (access to Google's TPUs and data centers for running AI models). It has nothing to do with advertising or measurement. There's been no announcement about integrating Google's ad measurement tools with ChatGPT ads. Given that ChatGPT competes directly with Google Search for ad revenue, whether that kind of partnership would ever happen remains an open question.
What This Actually Means for Marketers
This isn't something you can act on today. OpenAI has been clear: this is a limited test, not an open marketplace.
But it signals where conversational AI is heading. Someone asking ChatGPT for laptop recommendations is expressing clear purchase intent in natural language. That's valuable targeting real estate. The question is whether that potential translates into measurable results. And right now, there's no way to know.
My advice is to watch this closely, but don't chase it yet. The platforms want your ad dollars before they can prove they work. That's the playbook we've seen with every new channel. The marketers who wait for measurement to catch up make better decisions than the ones who throw budget at unproven inventory.
The Trust Question
OpenAI spent as much time explaining what ads won't do as what they will. That's telling.
Simo addressed it directly:
As we introduce ads, it's crucial we preserve what makes ChatGPT valuable in the first place. That means you need to trust that ChatGPT's responses are driven by what's objectively useful, never by advertising.
They know the risk. ChatGPT's value comes from people trusting the answers. If ads erode that trust, the product loses what makes it useful in the first place.
The guardrails they've announced are more conservative than what we've seen from other platforms at launch. No ads for minors. No ads near sensitive topics. Ads separated from answers. User controls for personalization.
Whether those guardrails hold under revenue pressure is the open question. History suggests that once ads work, the temptation to expand inventory follows. Google's AI Overviews went from 3% ad coverage to 40% in less than a year.
OpenAI says they'll refine the experience based on user feedback. That's probably true. What they don't say is how much feedback it would take to change course if the ads perform well financially.
How the Other Players Stack Up
OpenAI isn't operating in a vacuum. Here's where the major AI platforms stand on advertising:
Google AI Overviews:
The most aggressive and most mature. Google expanded ads into AI Overviews globally in 2025, and they're now testing ads in AI Mode (their ChatGPT competitor). Ads alongside AI Overviews went from about 3% of results in January to roughly 40% by November. The key difference is that Google has decades of ad infrastructure, measurement tools, and advertiser relationships already in place. When you run Google Ads, you can show up in AI Overviews without extra setup.
Microsoft Copilot:
Already running ads in conversational search. Microsoft reports 73% higher click-through rates and 16% stronger conversion rates compared to traditional search. They've launched Showroom Ads (interactive product comparisons), Brand Agents (AI-powered brand representatives in chat), and Copilot Checkout (complete purchases without leaving the conversation). Microsoft has the infrastructure advantage through their existing ad platform.
Perplexity:
They launched ads in November 2024 with premium CPM pricing and big-name partners like Indeed and Whole Foods. By October 2025, they stopped accepting new advertisers. The problem: they couldn't prove ROI. Advertisers couldn't track CTRs or ROAS with any reliability. Their head of ad sales departed in August 2025. The lesson: launching ads without measurement infrastructure doesn't work.
Anthropic (Claude):
No ads. Anthropic launched a brand campaign in September 2025 to advertise Claude to consumers, but they're not serving ads inside the Claude product. No announced plans to do so. For now, Claude remains ad-free.
What does this tell us? Google and Microsoft can make AI advertising work because they built on existing ad platforms. Perplexity tried to build from scratch and hit a wall. OpenAI is somewhere in between: they have scale (800 million users) but no measurement tools yet.
For marketers, this means Google AI Overviews and Microsoft Copilot are the only AI ad placements you can actually buy and measure today. ChatGPT ads are coming but not ready. Perplexity is paused. Claude isn't playing.
What about Google's AI Overviews?
On the organic side, third-party tools like SE Ranking, Otterly.AI, and seoClarity now track whether your brand gets cited in AI-generated answers, measuring presence rate, citation frequency, and share of voice across AI platforms. That's progress.
But for paid ads in AI Overviews? Google's own documentation says it plainly:
Google Ads currently doesn't offer segmented reporting when ads show within Search AI Overviews.
Your AI Overview ads get lumped into "Top Ads" reporting. You can't isolate performance. You can't opt out. You can't target them specifically.
Google says they're still learning and actively thinking about what the future of reporting looks like.
Even the company with $74 billion in quarterly ad revenue and decades of measurement infrastructure hasn't solved this yet. That should tell you something about the timeline for ChatGPT.
Bottom Line
ChatGPT ads are real. The measurement isn't.
If you're running paid media, this doesn't change your 2026 budget today. There's no platform to buy, no targeting to configure, no performance to optimize. This is OpenAI testing whether ads can exist in a conversational interface without breaking trust.
But it does change how you think about 2026 planning. AI platforms are monetizing. The marketers who figure out how to show up in these environments, both organically and through paid placement, will have an edge when the measurement catches up.
The thing you can actually do right now? AI search optimization. You can't buy ChatGPT ads yet. You can't measure them. But you can build the content, authority, and brand signals that get you cited and recommended when someone asks ChatGPT a question in your category. The brands showing up organically in AI responses today are building an advantage that's hard to replicate. When paid placement finally works, they'll have both channels. Everyone else will be starting from zero.
For now, the smart move is preparation, not ad spend. Get your tracking systems ready for when measurement catches up. Invest in content that AI systems want to cite. Understand how your brand appears when prospects ask ChatGPT for recommendations.
When the measurement catches up, you'll be ready. Until then, don't let anyone sell you on spend you can't track.
Comments