We've Been Testing ChatGPT Ads And These Are the Results We Are Seeing - Part II (3 Month Update)

Three months ago, we began testing ChatGPT Ads with a number of clients. Since then, the platform has evolved rapidly, with new campaign building tools, metrics, bidding options, and conversion-focused features. We've rebuilt campaigns multiple times as new features have been released and are seeing conversions. 

If you've read our previous articles on how to set up ChatGPT Ads and our initial impressions of the platform, this is the next update in our series, with new insights and experiences after three months of ongoing management.

ChatGPT Ads Campaign Setup Is Easy, and New Tools Are Available to Help

As explained previously, setting up a ChatGPT ads campaign is easy. The platform now also offers tools to further support campaign creation. This includes a detailed bulk-upload spreadsheet and a structured process to develop campaigns, ad groups and individual ads, as well as a “custom audiences” option. 

There are a number of differences between the ChatGPT Ads-recommended structure and the structure we originally developed on our own.

Feature	Characteristics & Differences
Ad Groups	The platform’s recommended workflow encourages creating a higher volume of ad groups compared to typical Google Ads setups.
Context Hints	The platform generates shorter and more direct context hints than those used in our initial manual campaigns.
Ads	The platform suggests creating significantly more ads than standard campaigns, with copy that is shorter and more direct.

We are currently testing both approaches: ads developed through the platform’s recommended process and ads written independently based on our own research and experience. Over time, this should help us determine whether the platform’s shorter copy and higher volume of ad variations improve performance.

Product Campaigns Are Accessible 

The platform now also provides the option to upload a product feed and create a dedicated product campaign. Although the setup guidance indicated that advertisers may need a catalog containing approximately 1,000 products, indicating making it an option primarily for businesses with large catalogs, we were able to launch a product campaign with significantly fewer items. 

You Can Now Set Up Campaigns for Conversions, With Bid Targets 

One key recent development was the introduction of a new conversion-focused campaign objective. Previously, campaigns were primarily designed to generate clicks. The new objective allows advertisers to create conversion-optimized campaigns to reach users who are more likely to complete a desired action. At first, existing campaigns could not simply be switched to the new objective. We had to rebuild participating client campaigns from scratch using the updated conversion campaign type. Since then, a new ‘campaign cloning tool has become available to simplify the work involved.

The new conversion campaign setup also asks advertisers to specify how much they are willing to bid for a conversion. The platform frequently recommends bids that are more aggressive than we would initially select. When a lower bid is entered, advertisers may receive a warning that the campaign “may not deliver.”

That does not necessarily mean the recommended bid should be accepted. We are starting with bids that reflect each client’s business goals and adjusting them based on results. As more conversion data accumulates, this will help us establish more realistic bidding ranges by industry and campaign type.

Conversion Tracking Remains a Challenge

ChatGPT Ads conversion tracking tools are evolving quickly, and the introduction of conversion-optimized campaigns is a key positive development. However, tracking and attribution are not yet as reliable or transparent as they are on more established advertising platforms.

This makes it important to compare platform-reported results with website analytics, customer relationship management data, booking systems and other sources. For example, several of our clients have experienced an uptick in brand search in Google Ads and ChatGPT organic traffic in client accounts after starting to test ChatGPT ads, even without the platform attributing conversions. As conversion measurement evolves, advertisers should currently treat the available conversion metrics with caution.

Conversions Are Coming In, But At Higher Cost Than Google Ads

After several months of testing, we are now beginning to see conversions across client accounts. These results have generally not appeared immediately after campaign launch. There has often been a lag between setting up a campaign, building sufficient visibility and beginning to generate conversions.

Overall, click-through rates are lower in ChatGPT, and cost per conversion is higher than Google Ads. Of the industries we are currently testing, travel has produced the strongest early performance so far, aligning well with people’s usage of ChatGPT to help research destinations and plan trips and itineraries. 

One observation we’ve had is that Open AI’s suggested CPA bids are significantly higher compared to what platforms like Google Ads are achieving, giving some insights into the higher average cost per conversion advertisers are currently getting. As you work through campaign setup, you will see some guidance on CPAs you can set to be competitive in the ad auction. For example, the ChatGPT Ads UI suggests a $600 Target CPA bid to a client who is currently paying around $100 per conversion on Google. 

Google Ads vs. ChatGPT Ads: How Do They Compare?

	CTR	CPC	Cost/Conversion
CLIENT 1	ChatGPT 1.17%  Google 6.2%	ChatGPT $2.78 Google $6.57	ChatGPT $16  Google $1.68
CLIENT 2	ChatGPT 0.82%  Google 6.12%	ChatGPT $2.69 Google $0.18	ChatGPT $230 Google $197
CLIENT 3	ChatGPT 1.19%  Google 1.30%	ChatGPT $2.34 Google $3.12	ChatGPT $311 Google $103
CLIENT 4	ChatGPT 1.87%  Google 1.75%	ChatGPT $1.11 Google $1.3	ChatGPT $201 Google $50

Reporting Is Improving but Remains Limited

ChatGPT Ads has also introduced an Overview reporting feature that can be integrated into external reporting tools. This makes it possible to aggregate basic metrics such as impressions and clicks alongside data from other advertising platforms. That is a useful improvement for agencies managing campaigns across multiple channels.

However, the available analysis remains rudimentary compared with Google Ads. Conversion data is not yet fully available through the reporting integration, limiting our ability to produce complete cross-platform reports. There is also very little visibility into what is driving impression volume, where ads are appearing or which types of conversations are generating the highest engagement.

For example, an advertiser may see a high impression share without having a detailed way to analyze where that visibility is coming from or whether it is reaching the most valuable users. These limitations make it more difficult to diagnose performance and identify specific optimization opportunities.

The Platform Requires Frequent Monitoring

One of our clearest observations after three months is that ChatGPT Ads requires a very hands-on approach. New features are being released regularly, and some updates require advertisers to make significant changes to existing campaigns. Campaign structures, objectives, bidding options and reporting tools have all evolved during our initial testing period. This rapid evolution requires frequent account monitoring and adjustment. 

Three Months In: Our Overall Assessment

Three months into testing, ChatGPT Ads remains an early-stage advertising platform with both potential and limitations. Campaign development has become more sophisticated, and conversion-focused optimization is now available. At the same time, reporting remains limited, conversion tracking requires improvement and bidding recommendations need to be evaluated and recalibrated carefully.

Most importantly, we are beginning to see early results. The platform is likely to become more competitive as additional advertisers enter the market. We are glad to be testing it now at a less competitive cost, while there is still an opportunity to learn how campaign structure, ad volume, creative style, bidding and industry type influence performance and to position client campaigns accordingly.

We will continue monitoring results, testing different approaches and sharing what we learn as ChatGPT Ads evolves.