A few weeks ago, I tested ChatGPT Ads by launching a campaign for Parkour3. In this article, I shared what you need to know before getting started with ChatGPT Ads. Now I’m sharing what I learned: the campaign results, what fell short, and the opportunities worth exploring next.
ChatGPT Ads is OpenAI’s advertising platform. It displays ads beneath ChatGPT’s responses, clearly marked “Sponsored” and matched to the topic of the conversation. For now, these ads are shown only to users on the Free and Go plans.
ChatGPT now reaches more than one billion weekly active users. Its advertising platform reached a $1 billion annualized revenue run rate within 200 days of launch and now offers self-service access in more than 60 countries, with tens of thousands of advertisers already using it.
I ran a three-week campaign to promote our HubSpot implementation services for manufacturing companies, with a budget of nearly CA$500. I divided it into three ad groups, each testing a different messaging angle:
- Unreliable sales pipelines
- Manual tasks and user adoption challenges
- A broader focus on CRM migration for manufacturers
All three ad groups directed visitors to the same landing page.
The campaign generated approximately 15,800 impressions and 110 clicks, with an average click-through rate (CTR) of 0.70% and an average cost per click (CPC) of $4.51. “Forecast Your Revenue” was the top-performing ad, with a CTR of 1.72%. The broader CRM migration angle drew just four clicks from nearly 750 impressions.
Conversions were less encouraging. The campaign produced only one form submission over the entire period, and it wasn’t a qualified lead: the person was simply curious, with no plans to implement HubSpot.
The first thing that stands out is the price. My average CPC hovers around $4 to $5, whereas comparable terms on Google Ads for a similar intent cost between $12 and over $50, depending on the exact term.
For example, “hubspot onboarding” can reach $59 per click at the upper end of the Google Ads range. ChatGPT Ads’ lower costs don’t necessarily mean it’s a better platform; they may simply reflect less competition among advertisers in our industry for now.
My second observation: direct messaging outperformed more creative or catchy copy. “No more unreliable pipelines” and “Forecast your revenue” clearly state the problem or benefit, and both were among my strongest performers. Other advertisers have reported similar results, with straightforward ads sometimes generating more than twice the CTR of more polished alternatives.
Third, context hints worked better when I treated them as a detailed brief rather than a field to fill in. Describing the buyer, identifying their stage in the purchase journey, and including examples of specific wording yielded better results than a vague description.
The final limitation is also one of the most important: for now, these ads reach only users on the Free and Go plans. Professionals using paid plans fall outside that audience, so we may be missing some of our most relevant potential buyers. That could help explain why the campaign’s only conversion wasn’t qualified, though one submission isn’t enough to establish a pattern.
For my next campaign, I’ll focus the budget on revenue predictability, the strongest-performing angle from this first test, rather than dividing it across three different messages. I also want to test whether highlighting our HubSpot Elite Partner status helps build trust.
On the landing page, I’d like to add a customer case study with a measurable outcome, such as implementation time, time saved, or sales team adoption. That would give visitors concrete evidence of what our work can deliver.
For now, I can’t say whether ChatGPT Ads is a good fit for us. Three weeks and one unqualified conversion aren’t enough to draw a firm conclusion. I’ll need longer, more controlled tests before deciding whether this channel deserves a permanent place in our advertising mix.
Are you considering digital marketing consulting or training? The Parkour3 team has been supporting B2B companies in this process for over 15 years. Let’s start with a conversation.