ChatGPT’s advertising business has hit $1 billion in annualised revenue. OpenAI announced the figure at the end of August 2026, around 200 days after ChatGPT Ads launched. For marketers, the big number isn’t the story. The story is the attribution gap.
Paid AI clicks are already hard to track
OpenAI has built a hard wall between the answer and the ad. It is cleaner than Google’s approach, but it is a headache for the way most marketing teams measure performance. Agencies are already seeing clicks reported in OpenAI’s dashboard that Google Analytics doesn’t fully recognise.
We are moving from an era where we could count every website click to one where we have to interpret the value of mentions. Yet most companies still judge AI visibility by watching for referral traffic in Google Analytics. If a channel doesn’t show up in your analytics, does your business treat it as real?
The bigger blind spot is unpaid AI mentions
Earned AI visibility and paid ChatGPT ads are two different fights. Treating them as one budget line misses the shift, and better attribution tooling won’t fix that.
If paid, deliberately trackable clicks are slipping through the gap, think about the mentions nobody is paying for or tracking at all. These are the moments when ChatGPT, Perplexity or Google’s AI Overviews simply describe your brand to a buyer, with no click involved. Zero-click search is getting worse, not better. For B2B companies, those unpaid descriptions shape shortlists long before a prospect visits your website.
Authority matters more than attribution
Attribution is important, but authority is the actual game. If you are struggling to measure the impact of AI, shift your focus to three areas.
Optimise for information gain
If your content only summarises what’s already on the web, you are teaching AI how to replace you. Original data, proprietary frameworks and genuinely new perspectives make you a primary source. The same logic Google uses to reward original content over summaries is likely to extend to how AI systems choose what to cite.
Track AI mentions as a leading indicator
If your referral traffic from AI looks like a ghost town but your brand mentions in Perplexity and ChatGPT are climbing, you are winning share of model. Measure the mentions, how you are described and which competitors appear alongside you, and stop judging AI search by clicks alone.
Build the narrative, not the hack
Google’s own guidance is clear that content chunking and llms.txt files don’t build lasting visibility. Earned media, original research and consistent expert commentary do, and brand mentions now track AI visibility more closely than backlinks. That is slower work, but it compounds.
The bottom line for B2B marketers
If you aren’t the source an AI system needs to cite to be accurate, you will be invisible, whatever your attribution dashboard says. Measure what you can, but invest in being the source.
SOURCES
Questions
What is the AI attribution gap?
The AI attribution gap is the difference between how much AI assistants influence buyers and how much of that influence shows up in analytics. Most AI influence happens when an assistant describes or recommends a brand with no click, so it never appears as referral traffic.
What is share of model?
Share of model is how often and how prominently a brand appears in AI answers for the questions its buyers ask, compared with competitors. It is the AI-search equivalent of share of voice.
What is information gain?
Information gain is the new knowledge a piece of content adds beyond what already exists online, such as original data, a proprietary framework or a genuinely new perspective. Content with high information gain is more likely to be treated as a primary source.
Nick Clark is the founder of Datum Point, a B2B consultancy for AI visibility and presence. He has spent 25 years in B2B, technology and corporate comms and marketing, including as CEO of Nelson Bostock Group and Publitek. First published on LinkedIn on 4 September 2026. Follow Nick on LinkedIn.