ChatGPT is now big enough to matter as an ad channel, but not big enough to prove it can drive sales. New data cited by Digiday shows advertiser adoption is climbing fast while revenue per visit and conversion rates remain far behind Google, leaving marketers with a familiar question: is this a performance channel yet?
Adoption is growing fast
Adthena tracked 1,082 distinct brands running ChatGPT ads across the U.S., Australia, the U.K. and the EU in June. That figure jumped to 3,741 in July and 6,333 in August, with projections putting September at roughly 6,830 advertisers. OpenAI is also reaching fewer than half of eligible users in its most established markets: the U.K. leads at 45.7%, followed by the U.S. at 43.4% and Australia at 43.3%, while Mexico sits at 27.8%.
Advertising only launched on the platform this year, so some of the gap between usage and monetisation is expected. ChatGPT handles about a fifth of Google’s search queries, but its ad revenue is still a fraction of Google’s.
The advertiser mix is broadening
Shopping brands accounted for 38% of U.S. ad spend on ChatGPT in April, but that share fell to 20% by September, according to Sensor Tower. In the same period, financial services rose from 2% to 13% of spend, software from 13% to 18%, and travel and tourism from 5% to 9%. The platform is no longer a retail-only experiment; it is drawing demand from service, software and travel advertisers.
The conversion gap remains the problem
Scale does not equal performance. Research from Maximilian Kaiser and Christian Schulze, based on ecommerce websites, found that paid ChatGPT visits generate $0.89 in revenue per session for U.S. advertisers, compared with $2.45 for paid Google Search visits. The conversion gap is just as stark: 0.37% for ChatGPT traffic versus 1.17% for Google.
New Street Research models commercial chats on ChatGPT rising from 45.9 billion this year to nearly 380 billion by 2030, with the share of chats carrying ads climbing from 20% to 28%. But the model also carries a caution: OpenAI now says CPC and outcome-optimized bidding account for most campaigns, and New Street believes its earlier estimate of a $5 CPC may be too high if monetisation is weaker than expected.
Measurement issues add to the uncertainty. Farhad Divecha, group CEO of Accuracast, said integrations with Magnite and Pubmatic have not solved the underlying tracking problems. “Some are teething problems because it’s a new system that has to be set up on advertiser systems, but some problems are down to attribution being messed up,” he told Digiday.
What marketers should do
- Treat ChatGPT as a test-and-learn channel, not a replacement for proven search budgets.
- Measure revenue per visit and conversion rate separately from clicks and impressions.
- Push for log-level attribution before scaling spend, because last-click data may be unreliable.
- Take the $5 CPC assumption as a model, not a benchmark; actual outcomes may be weaker.
For now, the smart move is to keep budgets small, demand conversion proof, and treat early ChatGPT ads as an audience experiment with performance potential, not a guaranteed performance engine.
Source: Digiday
