This is a dated open dataset, not a ranking of AI search engines.
I previously posted a public dataset of what one AI web-search answer tool cited for 48 “best … 2026” buying questions (Nov 3–6, 2026). On Wed Nov 18, about 8:32–8:35 AM PT (America/Los_Angeles), I re-ran the same 48 queries through the same single tool. One engine, one re-run — not a comparison with ChatGPT, Perplexity, Gemini, or Google AI Overviews.
A new write-up counts, from the public CSVs, which domains showed up most often and which ones appeared or dropped out:
- Original 240 citation rows: 193 distinct domains. Re-run: 190. 92 domains appeared in both runs. 101 appeared only in the original; 98 only in the re-run. Most of those were cited once (94 of 101, and 89 of 98).
- forbes.com was cited most in both runs, and it increased: 8 then 12 (publisher). nytimes.com was 5 then 6; goodhousekeeping.com was 4 then 4.
- exa.ai, cited 4 times on local-services queries in the original, was 0 in the re-run. wired.com went from 3 to 0 on consumer queries.
- In the local-services vertical, aggregator citations fell from 26/60 to 19/60, and brand citations rose from 17/60 to 29/60. Domain-type labels are ours, and the vocabulary differs by vertical.
Tables, the CSVs, SHA-256 hashes, and a method note are in the linked post.
48 queries is a small snapshot. Another re-run, or a different engine, could look different.
I made this for AnswerPlume, which is our AI-search citation project. This post is the dataset write-up, not a product pitch.


