Everything I test in live sprints - prompt targeting, citation tracking, what actually moves an LLM's answer - lands here first, before it goes near a client deck.
In a University of Toronto replay, ChatGPT's cited pages overlapped by only 15% between two runs of the same question on the same day. One answer tells you little about your visibility. A share of answers over several weeks tells you more.
Flagging the answer is worth doing, but it won't fix the next one. What works is finding where the error came from and putting clear, specific facts where AI tools look.
For getting cited in ChatGPT or Google's AI answers, no. Google's own guidance and two large studies point the same way.
Some of the work is shared. But what decides whether an AI cites or names you differs from what ranks a page in Google, and two studies show where.
A University of Toronto study of 158,099 AI answers found that page features explain 2.8% of the citation gap between pages on the same site. What ChatGPT's own search retrieves explains 32%.
The guides I compared cover the same six things. What separates a £1,500 retainer from an £8,000 one is how much of each you get.
Conductor's 1.08% covers May to September 2025. Contentsquare reports 0.2% and First Page Sage 6.2%, and the gap comes down to who was measured and when.
For two years, the answer to "do AI crawlers render JavaScript" has been a flat no, for every bot except Google's and Apple's. A detailed server-log analysis published this week says OpenAI's crawlers may have just started.
New research testing roughly 10,000 AI citations found a real effect from ranking yourself first in your own comparison content: a small lift in recommendations, a far larger one in mentions.
Three separate studies, checked against their own primary data, point the same way: brand equity and existing marketing spend explain most AI visibility. Citation-specific tactics explain a small, measurable slice of it.
No articles match your search.