Google’s AI Overviews started showing ads in 2024. Perplexity announced theirs at the end of that year, but it took OpenAI’s early 2026 announcement to get everyone’s attention.
AI company Anthropic saw its chance, and just three weeks after OpenAI’s ad launch, it used nothing less than the wildly popular American Super Bowl to target its rival and reassure users that Claude will remain ad-free.
Whichever side you’re on as a user, LLM advertising has arrived. Marketers who ignore it would be like SEOs ignoring the existence of Google Ads.
Brands need to fight the war for attention on an increasingly large battlefield that brings with it new opportunities to generate visibility, but challenges as well.
In this article, we’ll summarize the LLM ad possibilities and difficulties out there to help you navigate this new marketing channel and make sure you don’t miss a trend that could define 2026.
LLM Advertising: An Overview
| Platform | Has ads? | Who can see ads? | How to run ads | Known available metrics |
| AI Overviews (AIO) | Yes | Users in the U.S., Australia, Canada, India, Indonesia, Kenya, Malaysia, New Zealand, Nigeria, Pakistan, Philippines, Singapore, | By buying Google ad inventory – through Pmax and AImax. You cannot buy ads specifically for AIO, nor can you opt out of them as a Google advertiser. | No segmented reporting available for AIO. |
| AI Mode | Yes – ads in test phase | US users | By buying Google ad inventory. You cannot buy ads specifically for AI Mode, nor can you opt out of them as a Google advertiser. | No segmented reporting available for AI Mode. |
| Gemini | No | |||
| Perplexity | Not really… | US users | Buy directly from the sales team, but Perplexity is currently not accepting new advertisers. | CPM (Cost Per Mille) |
| ChatGPT | Yes – ads in test phase | Free and Go subscribers above 18 in the US. Free users can opt out in exchange for stricter daily conversation limits. | Via an ad agency or a direct invitation from OpenAI. “ChatGPT ad pricing is impression-based. Advertisers need to commit less than $1 million over several weeks to join OpenAI’s advertising pilot.” This form will allow you to be added to the waitlist. | CPM (Cost Per Mille) |
| Copilot | Yes | Every Copilot user above 13, in some regions above 18, who can see Microsoft ads across its services. | By buying Microsoft ad inventory. You cannot buy ads specifically for Copilot, nor can you opt out of them. | ImpressionsClicksClick-through-rateConversionsConversion rateCost per click (CPC)Return on ad spend (ROAS) |
| Claude | No |
AI Overview Ads
Google launched ads in Gemini-powered AI Overviews (AIO) in October 2024 “to help people discover new brands and make informed purchasing decisions”. Initially just for mobile users in the US, ads in AIO expanded to desktop devices in May 2025 and to 11 other countries in December 2025.
The ads carry the “Sponsored label” for transparency, are limited to English, and exclude “sensitive verticals like adult, alcohol, gambling, finance, healthcare, politics, and more” (Source). Also, only Text or Shopping ads from existing Search, Shopping, and Performance Max campaigns are currently eligible to be shown in AIO.
On top of that, not all AIO queries trigger ads.
- The query needs to show buying intent.
- The ad(s) need to be relevant to the user’s query.
- The ad(s) need to be relevant to the content generated by the AIO.
Google gives the following example:
It is not possible to directly pay for ad placement in AI Overviews. When you advertise in Google, Google’s systems decide whether your ads will be shown above, below, or within an AIO.
What you can do is:
- Leverage PMax to display across the wider Google inventory
- Use Google’s new AI Max for Search Campaigns to optimize your ads for long queries and the new ways in which people search multimodal Google.
- Follow Google’s recommended best practices for getting ads in AI Overviews.
You also can’t opt out of having your ads appear in AIO.
What about other Gemini-driven features?
Google is currently testing ads in AI Mode in the US. “Where relevant, ads may appear below and integrated into AI Mode responses.”
However, when it comes to standalone Gemini tools, Google representatives have spoken against adding ads on multiple occasions.
In December 2025, Dan Taylor (VP of Global Advertising said that “There are no ads in the Gemini app and there are no current plans to change that.”
One month later, DeepMind CEO Demis Hassabis shared at the World Economic Forum in Davos: “We don’t have any plans to do ads at the moment. We’re focusing on the core experience and the core technology of being a better assistant.” DeepMind is the developer of the Gemini models.
While these are strong statements, they come from a privileged revenue position, and it remains to be seen whether Google will hold onto them if users increasingly move their purchasing research (and decisions) to LLM platforms.
Measuring AIO Ad Performance
Currently, Google Ads doesn’t offer segmented reporting in AI Overviews. Ads that appear in AI Overviews are reported as “Top Ads.“
Perplexity Ads
Perplexity soon followed Google. In November 2024, it launched “sponsored follow-up questions” in the US. These appear at the top of Perplexity’s recommended next or related questions to ask and are marked “sponsored”:
At the time of the launch, Perplexity assured users that ads would not influence the content of answers and that “answers to Sponsored Questions will still be generated by our technology, and not written or edited by the brands sponsoring the questions.” It also won’t share personal user information with advertisers.
Interestingly, the company also stated that it wants to offer brands the opportunity to advertise so it can “avoid duplicating the SEO industry where people are implementing arbitrary tactics to improve their rankings at the expense of user utility.”
Advertisers have reported Perplexity advertising prices ranging from $30 to $60 CPM (Cost per Mille or cost per 1,000 impressions).
Note that there is no self-serve portal. Perplexity advertisers need to buy directly from Perplexity’s sales team. However, if you want to experiment with ads in Perplexity, you’re out of luck for now. Perplexity paused accepting new advertisers in October 2025, and Head of Publisher Partnerships Jessica Chan said that expanding ads to Perplexity’s AI browser Comet is not on the roadmap.
Exact reasons for the pause are unclear, but it’s safe to assume Perplexity isn’t generating the ad revenue it was hoping it would. On top of that, several media buyers told Digiday that the ad program focuses too much on just raising brand awareness and lacks scale, demonstrated ROI, and a clear way forward.
Measuring Perplexity Ad Performance
Perplexity’s privacy commitment makes it hard for advertisers to get granular ad performance data. The platform doesn’t provide:
- User demographics
- Ad click-through rates
- Retargeting options
- Ad conversion rates
With Perplexity’s ad program paused, there’s no decision to be made at the moment.
Brands that do want to get in front of highly-educated tech, finance, law, and healthcare workers (the core demographic of Perplexity users) can instead consider the company’s Publisher Program and optimize for organic appearances in the LLM-driven search engine.
ChatGPT Ads
OpenAI decided to start testing ads in ChatGPT in January 2026 and launched that test just a few weeks later. During this test phase, ChatGPT ads only appear for free users and ChatGPT Go subscribers above 18 years old in the US, not for those with pricier subscriptions. Ads also won’t appear for “sensitive or regulated topics like health, mental health or politics.”
Free tier users can opt out of ads in exchange for getting fewer daily free messages.
ChatGPT shows ads based on the topic of the conversation, past chats, and past interactions with ads. Advertisers, however, won’t have access to this data. In fact, OpenAI won’t give advertisers access to any type of user data. Users are also able to turn off personalization and clear data used for ads at any time.
They even get some control over the ads they’re shown, as users can dismiss ads and share feedback about them.
OpenAI also claims not to optimize ChatGPT for time spent in the tool. It’s an eyebrow-raising statement as any user knows how “fond” ChatGPT is of asking follow-up questions and suggesting next in-tool steps whenever supplying an answer. But we can be generous and subscribe to OpenAI’s mission to provide a helpful user experience.
As for the ads themselves, they will appear in clearly labeled “sponsored boxes” at the bottom of ChatGPT’s answers. They won’t influence the content of those answers.
For now, there’s no self-serve advertising platform yet for advertising in ChatGPT. OpenAI is working with ad agencies, such as WPP Mediaand Omnicom Media, to test ads in ChatGPT, and according to Digiday, it’s also reaching out to brands directly to join its pilot.
Brands can also fill out a formto be alerted as the advertising program expands.
ChatGPT ad pricing is impression-based. Advertisers need to commit less than $1 million over several weeks to join OpenAI’s advertising pilot, and OpenAI is targeting a premium cost-per-thousand-impressions (CPM) of around $60.
As for the future of ChatGPT ads, OpenAI plans to develop new formats and buying models.
Measuring ChatGPT Ad Performance
Given OpenAI’s strict commitment to user privacy, advertisers don’t get a lot of data to work with. They’ll only receive aggregate information about ad performance, such as the overall number of views or clicks their ads are getting. OpenAI won’t offer conversion tracking, nor data on purchase behavior or demographic breakdowns.
It’s a bold move given its ad pricing and a show of confidence in the value of purchasing visibility in ChatGPT conversations.
| “As we evolve our advertising program over time, protecting user privacy and safety will remain central to how we design and scale, including guardrails to prevent narrow ads targeting. We will continue to be deliberate about who we allow into the advertiser program and build protections to reduce the risk of scams and other harmful or misleading ads” – OpenAI. |
Copilot Ads
Microsoft announced Copilot ads in March 2025 and launched new interactive ad formats “entirely designed with Copilot in mind.”
Showroom ads tend to appear at the bottom of Copilot answers. They show rich sponsored content with images and product information when users express buying intent in their question to Copilot. In the future, Microsoft plans to add brand agents to these so that users can “engage directly with a virtual brand representative” through the ads.
Showroom ads were launched in April 2025 and are still in active rollout with global implementation planned.
Dynamic filters allow Copilot users to refine their search by interacting directly with the ads instead of typing another question into Copilot. They’re currently only available in English-language markets.
Other ad types can appear between the top links button and the answer.
When taking into account all ad types, any user who sees Microsoft advertising across its services and is older than 13 or 18 (depending on the region) can see Copilot ads.
Just like with AI Overviews, it’s not possible to buy ads specifically for Copilot features. Advertisers buy Microsoft Ads, and then Microsoft decides whether these appear in Copilot (Edge sidebar and Windows Copilot) alongside other Microsoft Platforms like Bing Search and the Microsoft Audience Network.
Aside from introducing Copilot advertising, Microsoft also made it possible for brands to use Copilot for ad asset creation on the Microsoft Advertising Platform, the Microsoft Advertising Editor, and through its API.
What’s more, advertisers can use Copilot in Microsoft Monetize, a key component of Microsoft’s advertising platform, to streamline their ad operations and get conversational access to their performance metrics.
Measuring Copilot ads
Measuring the performance of your Copilot ads is a lot easier than measuring the performance of ChatGPT and Perplexity ads, as Microsoft does not share the same strict privacy commitment.
Through Microsoft’s Advertising platform, brands can measure:
- impressions
- clicks
- click-through-rate
- conversions
- conversion rate
- cost per click (CPC)
- return on ad spend (ROAS)
This makes paid advertising plus organic Copilot optimization a powerful marketing combination.
No Ads on the Horizon for Claude
Anthropic, the company behind LLM platform Claude, has made an official statement saying that “Claude will remain ad-free.” The company generates revenue through enterprise contracts and paid subscriptions.
If you want to reach the enterprise users Claude targets, your best bet is to optimize your site and content for Claude.
Benefits of Buying AI Ads
The rise of LLM ad options presents new opportunities for brands.
Early Adopter Advantage
Brands that get in now have the opportunity to buy prime real estate in LLM conversations and gain visibility fast, where they might otherwise not (yet) appear organically.
They’re also able to experiment and fine-tune their approach so they’re in the best position to reap results and scale their campaigns as AI advertising matures.
Increased relevancy thanks to a variety of targeting options
Rather than targeting specific keywords, advertising in LLMs allows brands to reach users based on context and consumer search behavior.
- Someone looking for onboarding tips might be interested in dedicated onboarding software.
- Someone looking for ways to generate more customer reviews for their eCommerce store might be interested in your review and testimonial collecting tool.
Your LLM ads can be seen by people who likely need your product or service, even if they’re not specifically looking for it. Imagine an HR manager asking an LLM about a better way to handle their employees’ work schedules, and the LLM includes an ad for your employee management software to its answer. Chances are, they’ll get curious about your product.
By targeting context and behavior, advertisers can bypass the lack of personal user data with platforms like ChatGPT and Perplexity and target those prompts that have a higher likelihood of coming with purchase interest or intent.
And because LLMs understand what users need, they provide broad keyword matching that is likely to perform better than exact keyword matching.
AI Ad Creation Tools
AI tools allow businesses of all sizes to create high-quality ads, campaigns, and A/B tests at scale. A lot of ad platforms have these tools built in, and some even require you to use them to be able to advertise in LLMs.
- Google Ads has AI Max for Search.
- Perplexity generates its own sponsored follow-up questions and answers based on information provided by advertisers.
- Advertisers can use Microsoft 365 Copilot and Copilot in Microsoft Advertising to create ad creative and campaigns.
Users can create ads with ChatGPT, but OpenAI hasn’t yet launched a separate ad creator.
Challenges of Buying AI Ads
Alongside opportunities, advertising in LLMs also comes with its own challenges.
Measuring
Measuring the performance of LLM ads is hard for multiple reasons:
- LLM analytics tools are in the beginning stages and don’t fully offer clear and reliable tracking yet.
- LLM ad platforms are committed to user privacy. The aggregate ad data they’ll deliver will be less granular than what other digital ad platforms are providing.
- LLM ads will fragment the buyer journey even further, making it harder to track.
Those aren’t reasons to ignore LLM ads, though. As Flow Agency founder Viola Eva says: “Most new ad platforms have bad analytics when they launch first. That’s not a reason to stay on the sidelines. On the contrary: brands that show up first in LLM advertising have a competitive advantage that may carry through as the channel’s popularity increases.” If anything, brands that start now will have a solid foundation to build on as LLM ad platforms mature.
Marketers need to expect limited ad data, given the privacy commitments of many of these LLM platforms and the fact that long prompts often have too much personally identifiable information to be shared.
Eva recommends experimenting with a designated budget, measuring what you can, relying on self-reported attribution, and scaling if you get traction.
Budget
When Perplexity launched its ads in 2024, CPMs were hovering between $30 and $60. OpenAI is, as mentioned earlier, targeting a CPM of about $60.
In comparison, Google Display Network ads are estimated to average between $1 and $3 CPM, Meta ads between $6 and $20 CPM, Google Search ads between $11 and $20 CPM, and Netflix ads between between $20 to $65.
LLMs are pricing their ads at a premium, and that might be a reason for some businesses to hold off for a while.
On the other hand, getting in now gives you the chance to experiment and understand the field before everyone else does. On top of that, advertising on a platform like ChatGPT gets your brand in front of users right when they’re expressing buying intent.
User Trust
People want to believe that LLMs provide neutral answers. There is a chance that seeing ads in those answers will undermine that trust. It reminds me of reading a blog post in which the author has incorporated affiliate links.
Two things can happen:
- Either those products benefit from the trust and authority the author has gained…
- Or they create distrust in both the product and the author.
The latter is more likely to happen when 1. the recommendations are bad and 2. products are pushed in a way that feels unnatural.
The same risk exists with ads in LLM answers. Will the authority these platforms have gained reflect positively on advertisers, or will the ads break down some of the trust users have in tools like ChatGPT? Time will tell.
Prepare to Play With AI Advertising
If you’re ready to start experimenting with LLM ads, here are a few ways in which you can increase your chances of success:
- Clean up your product, service, and feature pages. Think clear and concise titles, appealing images, and up-to-date pricing.
- Optimize your content for visibility in organic LLM responses.
- Strengthen or build partnerships with platforms that (can) drive business for you. If they start advertising, you could benefit too.
- Optimize for trust by generating more (real!) positive reviews and testimonials.
- Target context over demographic profiles and create ads for specific problems.
Remain flexible. Now is not the time to build a rigid LLM ad system. Instead, prepare to test, track, and adapt. That includes freeing up some advertising budget to start experimenting with advertising in LLMs.
Ready to Start Experimenting With Ads in LLMs?
The introduction of LLM advertising allows brands to reach their target audience when it’s most likely to be interested in their offerings.
Lack of concrete metrics and high CPMs (cost per thousand impressions) definitely create challenges for marketers, but getting on board now allows you to test, experiment, and show up as platforms mature.
At Flow Agency, we’re determined to try LLM ads in all platforms relevant to our clients. We recognize the importance of being early adopters and want you to benefit from what we learn.
Sounds interesting? Reach out today to discuss how we can help you build a sustainable LLM optimization strategy.
