Back in March, OpenAI announced the launch of ads on ChatGPT. After initial tests in the US, the ads platform has now rolled out in the UK.
With its simple set-up process and ability to launch campaigns with small budgets, many advertisers are likely to be taking up the opportunity to test the platform. We recently conducted a small-scale test of ChatGPT Ads to better understand how the platform works, evaluate initial performance, and identify areas for future experimentation. While the sample size was relatively small, the results provided some interesting insights into this emerging advertising channel.
Why test ChatGPT ads?
ChatGPT enables advertisers to connect with users who are already exploring topics, products or services relevant to their business. Furthermore, a recent survey from YouGov found that ChatGPT was the preferred AI brand in the UK. 30% of respondents chose ChatGPT over the likes of Copilot and Gemini.
As the audience is there, the question is:
"Can ChatGPT become a viable media channel for driving awareness, traffic and leads?"
To help answer that question, we launched a test campaign promoting Equimedia's digital marketing services.
How ChatGPT advertising works
The current ChatGPT Ads platform differs from other advertising platforms in several ways.
Objectives
The platform offers the choice of optimising towards Reach, Clicks or Conversions. Additionally, for e-commerce advertisers, there is an option to run a Product Feed campaign.
Targeting
At present, targeting is primarily based on contextual themes, similar in some respects to Google Performance Max campaigns. Rather than targeting specific keywords, advertisers select broader themes and categories that align with their products or services.
Demographic targeting is limited with no age or gender targeting and geo-targeting is available at country level only (city, postcode or radius targeting is not currently available).
One additional option is the ability to use first-party data and create custom audiences. These can be used for inclusion or exclusion.
Budget Management
We also observed some differences in campaign budgeting.
- Lifetime budgets did not pace evenly across the campaign duration
- Daily budgets provided more predictable spending control
For advertisers testing the platform, daily budgets may offer greater control during the learning phase.
Measurement
Tracking was implemented using a website pixel, with both page view and lead events configured.
This allowed us to monitor traffic quality and understand how users interacted with the website after clicking an ad.
The Results
Across three ad variations, the campaign generated:
- Average CTR between 0.75% and 1.70%
- CPCs ranging from £1.95 to £2.08
- Average CPM of £26
Interestingly, click-through rates were more comparable to what we might expect from Meta campaigns, while CPMs were closer to LinkedIn benchmarks.
We also reviewed data in GA4. Users arriving via ChatGPT demonstrated:
- Engagement rates higher than our PPC traffic
- Engagement time comparable to organic search visitors
While the dataset remains small, these early indicators suggest that ChatGPT users may arrive with stronger intent than other paid media channels.
Results were segmented by device which showed:
- Around 80% of impressions were served on mobile devices
- Desktop represented only a small proportion of inventory
- Desktop CTR averaged just 0.4%
This suggests that ChatGPT advertising should be approached with a mobile-first mindset.
What's Next?
Currently, the ads platform is very simplistic as it is still in Beta. But as OpenAI develops the ads offering, I would expect some major changes to the ads platform over time. Future updates that we would like to see include:
- Insights into the prompts (or context themes) which are triggering ads (similar to search query reports)
- Built in GenAI to help advertisers with creating ads
- Additional metrics for reporting e.g. Reach and Frequency
- Segmentation of results by demographics
- Granular geo-targeting
User feedback on the ads experience will no doubt be a key factor in how the ad offering evolves. OpenAI will be keen to find ways of increasing revenue through ads whilst maintaining and growing its base of users.
Our Early Verdict
While the platform currently lacks the reporting depth and targeting sophistication of Google, LinkedIn or Meta, it offers access to an audience actively seeking information and solutions. Early engagement metrics suggest traffic quality could be strong, but caution is advised given the limitations of the insights offered.
Are you interested in exploring emerging advertising channels like ChatGPT? Get in touch with the Equimedia team to discuss how AI-powered advertising could complement your wider digital marketing strategy.
Newsletter sign up
Post Information
| Author | Amisha Rathod |
| Channel | Media |