Data
Open datasets on ChatGPT advertising and AI search. Every number links to where it came from, every table downloads as CSV or JSON, and everything is free to reuse with attribution.
ChatGPT Ads Index
- ChatGPT Ads benchmarksEvery publicly reported ChatGPT Ads CPM, CPC, CTR, conversion rate, ROAS and minimum spend, deduplicated and linked to its source.
- ChatGPT Ads timelineDated changelog of ads in ChatGPT: announcements, pricing changes, country launches, formats, tools and policy.
- Brands advertising on ChatGPTAdvertisers publicly reported as running ChatGPT ads, with category, market and the source that reported them.
AI Search Reference
- AI search engine referenceHow ChatGPT, Gemini, AI Overviews, AI Mode, Perplexity, Claude, Copilot and 11 more AI engines find, cite and send traffic.
- AI crawler directoryEvery known AI crawler and user agent: operator, purpose, robots.txt behaviour, user-agent string and IP list.
- Who blocks AI crawlers? robots.txt studyOriginal analysis of the robots.txt and llms.txt files of the 10,000 most popular websites.
- AI referrer domains for analyticsReferrer hostnames and UTM parameters for AI assistants, with a ready-to-paste GA4 regex.
- AI assistant user numbersOfficially stated user counts for ChatGPT, Gemini, AI Overviews, AI Mode and other AI engines, dated and sourced.
Research
- Who blocks AI crawlers? What 10,000 robots.txt files revealPublishers block training, not answers; half of all blocklists leak; Cloudflare writes the most common AI policy.
- The $4 click: what 184 data points reveal about ChatGPT AdsCPC anchored to OpenAI's recommended bid, CPM and CPC converging, and a $250K club turned $25-a-day long tail.
- The opt-out asymmetry: you can block the small AI engines, not the big oneThe AI surfaces with the most users are the ones publishers can't refuse and can't measure.
How this data is built
Secondary datasets are assembled from public reporting, official documentation and practitioner case studies. Sources are searched systematically, figures are extracted into a fixed schema, deduplicated, and linked back to the earliest or most authoritative source. Primary datasets, such as the robots.txt study, are collected directly. The full process is documented in the methodology.
Questions
Is this data free to use?
Yes. Every dataset is published under Creative Commons Attribution 4.0, which allows any use, including commercial, as long as you credit Kuldeep Paul, kuldeep.cc and link to the dataset page.
Where does the data come from?
The robots.txt study is primary data I collected by fetching the top 10,000 websites. The other datasets aggregate public reporting, official documentation and company statements, with every figure linked to its source.
How often is the data updated?
On a rolling basis. Each dataset shows its last update date, and the CSV and JSON downloads always match what the page shows.
Can I download the data?
Yes. Every dataset page has CSV and JSON downloads, and each file includes the licence and a link back to its page.