Key points
- Adthena tracked 1,082 brands advertising on ChatGPT in June and 6,333 in August, with September projected near 6,830 across the U.S., Australia, the U.K. and the EU.
- A working paper under review found paid ChatGPT visits earned $0.89 per session versus $2.45 for paid Google Search, with conversion at 0.37% versus 1.17%.
- Sensor Tower says shopping fell from 38% to 20% of U.S. ChatGPT ad spend between April and September, as finance and software gained share.
- In an Emarketer and Semify survey of 75 SMB search marketers, 56% named measuring ROI from AI visibility as their top challenge.
Why it matters for marketers: Budgets are moving into ChatGPT faster than proof of return, so marketers should treat spend as a test with strict measurement and not assume parity with search.
ChatGPT's advertising business is growing quickly in brand count and category mix, but evidence of performance is still thin. Adthena data cited by Digiday shows distinct advertisers rising from 1,082 in June to 6,333 in August. Sensor Tower estimates that shopping's share of U.S. ad spend dropped from 38% in April to 20% in September, while financial services, software and travel all gained.
Performance is the open question. A working paper by Maximilian Kaiser and Christian Schulze, still under review and based on ecommerce site data, found paid ChatGPT traffic converted at 0.37%, against 1.17% for paid Google Search. Revenue per session was $0.89 versus $2.45. Reach is also uneven: Similarweb and Graphite data show fewer than half of ad-eligible users see ads even in the U.K., the U.S. and Australia.
Measurement adds friction. Farhad Divecha, group CEO of Accuracast, said new partner integrations with Magnite and Pubmatic have not fully fixed tracking, and that attribution problems persist. Adweek reports a similar picture on the organic side: in an Emarketer and Semify survey of 75 SMB search marketers, 76% called AI search visibility very or extremely important, yet 56% named ROI measurement as their top challenge.
For marketers, the practical step is to run ChatGPT ads as a contained test with clear conversion tracking and compare results against paid search on the same terms. Watch for whether OpenAI expands ad delivery in more markets, and whether measurement partners close the attribution gaps. Until then, the numbers suggest curiosity is ahead of proof.
Sources
This brief is our own summary. Read the full reporting at the source.