The Failure of ChatGPT Ad Targeting

Successful Software requires more than just good programming.

I recently conducted an experiment to see if buying advertisements within ChatGPT would help drive users to my event seating assignment software, PerfectTablePlan. To put it bluntly: the targeting is abysmal.

📊 The Raw Data

Between 05-Aug-2026 and 28-Aug-2026, I tracked the following performance metrics:

• Total Expenditure: £289.52
• Total Clicks: 2,988
• Average Cost Per Click (CPC): £0.10
• Average Time on Page (via Google Analytics): 7s
• Bounce Rate: 69%
• Total Installs: 14
• Conversion Rate (Click $\rightarrow$ Install): $0.46\%$
• Cost per Conversion: £20.68

Interestingly, while ChatGPT suggested a bid of approximately £2 per click, I found that I could still generate significant traffic with bids ranging from £0.07 to £0.15.

Based on the lifetime value of a customer and the typical install-to-purchase ratio for my software, these ads are nowhere near profitable.

🔍 Comparative Analysis

To determine if the issue lay with my landing page or the traffic source, I compared ChatGPT's paid ads against Google Ads and organic (free) ChatGPT referrals during the same window.

The disparity is stark. Because the organic referrals and Google Ads performed well, it is evident that the problem is the quality of the paid ChatGPT traffic.

🛠️ The Investigation Process

I attempted to diagnose why the paid traffic was so poor. Here is the logic flow of my investigation:

Ruling Out the Obvious

I used a task list to track my hypotheses:

• OpenAI is overcharging for clicks they aren't sending. (Ruled out: OpenAI's reports align with Google Analytics).
• The ad creative/copy is ineffective. (Ruled out: I used the same ads that have worked on Google Adwords for years).
• The traffic is almost entirely bots. (Investigated below).

The Bot Analysis

I consulted digital ad fraud specialist Dr. Augustine Fou, who helped me implement fouanalytics.com to monitor user behavior.

The findings were revealing:

1. Bot Presence: About 30% of the traffic exhibited "some" or "many" red flags for being bots.
2. User Interaction: While 88% of visitors moved their mouse, only 10% actually clicked anything on the page.

I used Easy Data Transform to create a histogram of the time spent on the landing page (binned by 1 second), which further highlighted the lack of engagement.

📉 Final Conclusions

It is clear that OpenAI is capable of sending targeted traffic, as their free referrals convert at a rate an order of magnitude higher than their paid ads.

This leads me to wonder: is OpenAI so desperate to close their financial gap that they are prioritizing quantity of clicks over quality? I have no idea in what context my ads are being displayed, and I see no realistic path to improving the conversion rate enough to make this profitable.

Expert Verdict

"Best practices confirmed by this experiment — always use real conversions as the measure of success. The problem with ChatGPT ads was NOT bots; it was low value visitors (ones that arrived on the landing page but did nothing else and left). It's not entirely clear WHY these visitors are so low in value (e.g. were the ChatGPT ads simply targeted very poorly?) Invest your ad budgets in sources that not only yield more clicks to your site, but most importantly more actual conversions." — Dr. Augustine Fou

---

Related Tools:

• Easy Data Transform: For cleaning and merging data without coding.
• Hyper Plan: A flexible visual planner.
• Perfect Table Plan: The fastest way to create seating plans.