Everything Brands Need to Know About ChatGPT Ads: Insights From an Early Advertiser

[tldr]

In summary: What you need to know about ChatGPT ads

• Insights from an early ChatGPT advertiser show an unpolished product that was difficult to prove as a performance channel, but OpenAI has since added many of the capabilities they were waiting for.
• Early access required large budgets + a tolerance for rough processes and infrastructure. OpenAI has since eliminated its large minimum spend requirement and expanded access through its self-serve Ads Manager.
• Campaign controls during the original test were minimal, with limited reporting and no automated bidding. Today, advertisers have access to CPC and conversion-optimized campaigns, conversion tracking, custom audiences, and automated bidding.
• Performance during the advertiser's early test fell short: several thousand dollars in daily spend drove very few orders.
• The platform has evolved quickly, although OpenAI still considers ChatGPT Ads a beta product and has not published broad performance benchmarks.
• Brands should prepare now by investing in GEO, evaluating whether the platform is ready for a test, and carefully monitoring developments in the platform.

[/tldr]

ChatGPT ads are here here.

And the platform looks considerably different than it did when we first published this article in May 2026.

In just a few months, OpenAI has added conversion tracking, conversion optimization, automated bidding, custom audiences, product feeds, additional ad formats, and more robust reporting. It has also expanded ChatGPT Ads well beyond its initial U.S. test.

To help you know exactly what to expect and how to prepare, Right Side Up recently caught up with an early ChatGPT ads tester—a senior performance marketing leader at a large consumer brand.

Their experience offers a useful look at where ChatGPT Ads started—and, perhaps more importantly, a benchmark for how quickly the platform is changing.

What ChatGPT ads are like today, in short

The short version?

When our advertiser first tested ChatGPT ads, they were very much in their “hard hat required” phase. The platform was early, unpolished, and difficult to prove out as a performance channel.

Four months later, the hard hats are still needed, but the building looks more put-together.

OpenAI now offers CPM, CPC, and conversion-optimized CPC (oCPC) campaign objectives. 

Advertisers can track conversions using the OpenAI Pixel and Conversions API, use first-party customer lists for targeting and bid adjustments, and let ChatGPT automatically adjust bids through its Maximize results strategy.

In other words: several features our early advertiser identified as prerequisites for serious performance investment are here.

In this article, we’ll go over the advertiser’s experience with running a ChatGPT ad campaign and what has changed since. We’ll cover:

• How brands are getting into ChatGPT ads today 
• How campaigns are run (controls, bidding model, reporting)
• Why the earlycampaign fell short of expectations
• The path for ChatGPT ads to become a widespread, viable channel 
• What brands should be doing today to prepare and test

How brands are getting into ChatGPT ads right now

When our advertiser entered the pilot, the brand proactively reached out to OpenAI.

They advertiser described the process as very early-stage, with requirements evolving over the course of the relationship.

What was required for ChatGPT ads during the early pilot

Participation initially required meaningful budget and tolerance for ambiguity:

‍

That experience is now largely historical. OpenAI has since opened self-serve Ads Manager access in the U.S. and several other markets, with additional advertisers able to access the platform through OpenAI's sales team and partners. The company says the platform remains in beta and is still in the early stages of scaling.

What the setup process looked like

The early setup process was not polished, to say the least.

As mentioned earlier, the advertiser initially had to send spreadsheets to OpenAI with campaign inputs, and the experience felt closer to “send us the files and we’ll upload them” than a modern ad platform.

Over time, that improved. A lightweight UI became available, and OpenAI continued sending product update emails every couple of weeks.

Still, the advertiser described the overall experience as “the wild west.” Not surprising for a new ad platform, but important for brands to understand as context for the campaign results discussed below.

Today, Ads Manager supports campaign creation and editing, billing, account management, performance reporting, bulk uploads, conversion measurement, and other capabilities you'd expect from a self-serve advertising platform. It's still a beta, but the buying experience has definitely moved beyond the spreadsheet era.

A comprehensive look at how ChatGPT campaigns work

Our advertiser gave us all the details on how campaigns functioned during the early pilot, from what controls were available to how bids worked to the limited reporting capabilities at the time. Here’s what they experienced—and what has changed since.

What campaign controls were available?

The advertiser had control over two primary inputs:

1. Keywords
2. Creative

That was it, more or less.

At the time, there was no real audience targeting, meaningful geo-targeting, conversion-based optimization, or automated bidding.

OpenAI’s position, according to the advertiser, was that ChatGPT’s conversational context and high-intent environment reduced the need for granular targeting.

For performance marketers used to deep controls with Google or Meta ads, the experience felt limited.

That has changed a fair bit. Advertisers can now target countries and platforms, while custom audiences allow them to upload customer or prospect lists using email addresses or phone numbers. Those audiences can be included, excluded, or used to adjust bids for specific ad groups.

And what the advertiser called “keywords” has evolved into what OpenAI calls “context hints.” These describe the conversations, needs, topics, or use cases where an advertiser's product may be relevant. 

Important note: These context hints aren't exact-match keywords or guarantees that an ad will appear in a specific conversation. Instead, OpenAI claims to combine them with conversational intent, landing pages, ad copy, targeting selections, and other signals to determine relevance.

Creative mattered more than expected

One interesting difference: OpenAI seemed to emphasize creative variation more than traditional paid search platforms.

Instead of submitting one ad with several headline and description variations, the advertiser created multiple distinct creative versions for different campaigns, events, and categories.

That aligns with OpenAI’s current public creative guidance, which recommends building “for coverage” with a high volume of distinct ad variations so the system can match ads to a wider range of conversational contexts.

If this remains best practice, it’s a noteworthy shift for marketers in terms of creative structure compared to classic keyword-based search.

Brands may need more message diversity, more contextual angles, and clearer descriptions of:

• What they offer
• Who it’s for
• When it’s useful
• Why someone should care in that specific moment

Since then, OpenAI has introduced product-feed campaigns for retailers, allowing advertisers to upload catalogs and create product ads at scale. Product carousel formats can surface multiple items within an ad placement rather than relying solely on a single static card.

Interestingly, OpenAI says product-feed advertisers have been among the platform's strongest performers to date—something worth watching, especially for ecomm brands.

The buying model: CPM, CPC, and conversion optimization

During the test, the platform was primarily CPM-based.

That made the channel feel less like paid search and more like display or social, despite the high-intent promise of the environment.

Today, advertisers can now choose among CPM campaigns for reach, CPC campaigns for traffic, and conversion-optimized CPC (oCPC) campaigns for downstream actions. 

OpenAI has also introduced automated bidding through its “Maximize results” strategy. For eligible campaigns, ChatGPT Ads can automatically set and adjust bids to generate as many clicks or conversions as possible within the available budget.

It does not currently guarantee a particular CPA, CPC, or ROAS, so in that sense it’s still far away from the mature bidding systems marketers are used to on Google and Meta. 

Reporting during the early pilot was extremely limited

The advertiser had access to basic platform metrics:

• Impressions
• Clicks
• Spend

Compounding the lack of a proper bidding system, there was little visibility into what was actually driving results. 

Instead, the team used UTMs and had to strictly rely on Google Analytics and internal order data to understand downstream performance.

Measurement is another area where OpenAI has made progress. Ads Manager now reports impressions, clicks, spend, CTR, average CPC, average CPM, and attributed conversions when conversion measurement is configured. Advertisers can also export reporting data for analysis.

More importantly, advertisers can now send conversion events using the OpenAI Pixel, Conversions API, or both. OpenAI has also begun supporting third-party measurement integrations.

Campaign results, and 3 reasons it fell short

As a reminder, these reflect an early version of ChatGPT Ads, and a current advertiser’s experience is likely to be very different.

For context, the brand spends roughly $1 million per day on Google Ads and generates thousands of orders. 

ChatGPT ads were receiving a few thousand dollars per day in spend but producing very few orders.

Based on the results driven by its other search programs, the advertiser felt it would be reasonable to expect more than a handful of daily orders from ChatGPT ads.

So why was the performance so disappointing? The advertiser pointed to a few likely reasons.

1. The ad placement felt less organic than search

In traditional search, users often click top placements without deeply distinguishing between paid and organic results.

In ChatGPT, ads appeared below the response and were clearly labeled as sponsored, and visually separated from the response. While the transparency works in favor of user trust, it can also affect click behavior 

That separation remains intentional today. OpenAI says ads are designed to remain separate from ChatGPT's answers and not influence them. 

However, the format itself is evolving: users may now encounter ad units featuring multiple items, including product carousel experiences, rather than only the static format our advertiser tested.

2. The format still felt rough

The advertiser described the ads as visually unpolished.

‍

As expected, the formats have already started evolving. OpenAI has continued refining static cards while adding product-oriented formats and testing additional ways for advertisers to surface relevant products inside the experience.

3. There was no automated bidding

We’ve mentioned this already, but this was the biggest missing piece.

Without automated bidding, the advertiser felt the platform was closer to “set budgets and let the chips fall where they may” than a mature performance channel.

That was a tough sell for senior marketers responsible for efficient growth.

OpenAI has since starting addressing this gap with oCPC campaigns and automated bidding.

That's an important development because it means one of our advertiser's biggest requirements for revisiting the channel now exists—although the system is still new and lacks the years of optimization data behind established platforms.

What needs to improve before ChatGPT ads become a must-test channel?

When we first interviewed the advertiser, they were clear: they would need to see more infrastructure before recommending the channel broadly for performance marketing.

Let’s take a look at some of these initial gaps—and if and how each has been addressed.

‍

Add custom audiences, expanded geo-targeting, and product feeds to the list, and ChatGPT Ads is closer to a real performance platform than it was in May.

“Closer” is the key word here: OpenAI itself says ChatGPT Ads remains in beta and doesn't yet have broad performance benchmarks across advertisers, industries, or campaign types. 

So while the platform is significantly better than it was in May, it’s still far from the competitors it’s trying to catch up to.

Should brands test ChatGPT ads now?

Our early advertiser's recommendation was that brands requiring immediate ROI should wait. It’s still mostly true, but given the product changes since then, we'd soften that conclusion a tad.

ChatGPT Ads may now be worth testing for brands that:

• Have enough test budget and volume to generate meaningful learnings
• Have a broad ideal customer profile (ICP)
• Are early adopters in channels and/or want an early learning advantage over competitors
• Can tolerate a less mature performance environment than Google or Meta‍
• Have strong conversion tracking and a clear event to optimize toward
• For ecommerce brands, have a product catalog that could take advantage of feed-based campaigns

Even then, be wary of the platform’s limitations: You're testing an advertising platform that's roughly six months into its life—not one with two decades of optimization history.

How brands should prepare for ChatGPT ads

Whether you're ready to test ChatGPT Ads today or plan to wait, brands should absolutely prepare for the broader AI search and advertising ecosystem now.

The advertiser’s strongest recommendation: get your AI search foundation in order.

1. Invest in GEO visibility

Paid and organic AI visibility will likely become increasingly connected, much like paid search and organic search evolved together.

[tldr]

Related article

Need advice on how to build or boost your GEO strategy? Read our comprehensive guide.

[/tldr]

As such, marketers need to make sure AI systems can understand, retrieve, and accurately describe their products and brand. 

That means strengthening things like:

• Website and product content
• FAQs and comparison content
• Third-party mentions and reviews

For ecommerce brands, there's now an additional paid-media reason to get product data in shape: ChatGPT Ads supports structured product feeds that can power ads across a catalog. 

Organic and paid product visibility remain separate today, but clean, detailed product data is becoming useful on both sides of the AI discovery ecosystem.

2. Think about creative for conversational context

ChatGPT ads are designed for conversational moments. 

And when the time comes, brands will need creative that maps to customer intent, not keywords.

For example:

• “I’m comparing options”
• “I need help choosing”
• “What’s the best solution for X?”
• “What are the tradeoffs between A and B?”

In these instances, contextual usefulness will likely dictate how successful a creative is.

OpenAI's own guidance reinforces this. The system considers conversational intent alongside ad copy, landing pages, context hints, and other relevance signals.

That means marketers should think less about exact-match keyword lists and more about the situations and decisions that make their product useful.

3. Prepare measurement expectations internally

Whenever you’re ready to pilot ChatGPT ads, keep in mind that it may not look like a mature Google Ads test. 

In that case, it’s essential to clarify the learning agenda internally:

• Are we testing volume?
• Conversion quality?
• Incrementality?
• Creative resonance?
• AI discovery behavior?

Then make sure your measurement backbone is solid before you launch. With the OpenAI Pixel and Conversions API now available, brands can measure downstream actions and give conversion-optimized campaigns the signal they need.

4. Take advantage of the new targeting and optimization tools

This platform is still moving very quickly, but advertisers now have considerably more to work with.

Custom audiences, for example, let brands use first-party customer or prospect lists and for campaign inclusion, exclusion, or bid adjustments. Advertisers can bid more aggressively for high-value audiences or suppress existing customers from acquisition campaigns.

One important limitation as of now: custom audiences need at least 25,000 matched users before they can be used, and OpenAI recommends lists of at least 100,000. 

How ChatGPT ads could change the marketing mix

If ChatGPT ads mature, they could become a third major high-intent advertising channel alongside Google and Microsoft.

That matters because paid search has become increasingly expensive; CPCs on Google have steadily risen over the past decade. 

But ChatGPT ads may not simply become “paid search 2.0.”

The bigger opportunity is AI-native advertising: showing up while users are comparing options and making decisions inside a conversational interface.

What that could mean for your marketing mix:

• Paid search budgets may expand into AI answer engines.
• SEO and GEO may become even more tightly linked to paid strategy.
• Creative may need to become more contextual and educational.
• Measurement models may need to account for AI-assisted journeys.
• First-party audience data may become another important signal layered on top of conversational intent.

The takeaway: Don’t ignore it, but don’t overhype it

When we published this article in May 2026, our conclusion was that ChatGPT ads deserved marketers' attention, but not necessarily their budget. It may still be the case for most brands, but with more nuance now.

The platform has the ingredients of a major future channel:

• High-intent conversational moments
• More specific context than traditional search‍
• Self-serve campaign management, conversion measurement, and optimization‍
• First-party audience targeting

For now, we'd still treat ChatGPT Ads as an emerging test channel rather than a proven replacement for mature performance platforms. But for the right brand, there's a much stronger case for actually running that test than there was in May.

Strengthen AI visibility, build (or strategize) conversational creatives, define a learning agenda, get your conversion tracking in order, and watch the platform roadmap closely. Then, when the infrastructure matches your needs, you’ll be ready to act.

[tldr]

Need help with ChatGPT ads?

We’ve got talent running ChatGPT ads for clients right now.

Want to learn more about the channel and evaluate if it could be a good fit for you? Let’s chat.

(And if you need help strengthening your GEO strategy in the meantime, we’ve got you covered too.)

[/tldr]

‍