I have tended to avoid the conversation about AI citations, primarily because 90% of the chat on the subject is either scaremongering, ragebaiting or a snake oil sales pitch dressed up as a framework or a tracking tool.
The real world stats on how AI (primarily AI overviews) is affecting click numbers is very real though. So I sat down and actually read the recent research instead of skimming the LinkedIn hot takes and worked out how to take data that is already there and work out a plan to stand the best chance of being cited by AI while not obliterating what remains of organic traffic as traditionally the 2 do not work hand in hand.
There's a layer underneath your keyword research called "fan-out queries". They have been spoken about for a while, but very much missused in research. They are invisible little sods and some times not what you might expect to see. Before ChatGPT, Gemini, perplexity or AI Mode ever touches a web search, it quietly decomposes your prompt into a handful of sub-queries it reckons will get you a fuller answer, fires them all in parallel, then stitches the results together into a frankenstein answer. Sometimes good, sometimes terrible. But, if I am truly honest they are getting better day by day.
You never see it happen. Neither does your keyword tool.
Finding them yourself doesn't take much faff at all. Mark Williams-Cook's QueryFan and Mic King/iPullRank's Qforia both do it well.
I went further down the rabbit hole than I meant to and wrote up what the research actually says, including the bit that should make you think. Trying to cover ALL the fan-out queries on one page or even worse, many very thin pages is the wrong move. The data is clear, partial coverage wins.
I built a small tool that piggybacks on the real work above (because I'm not reinventing what Mark and Mic already nailed) and lets you check your existing pages against the gaps. You may like it, you may not.
However, central to this tool is common sense and interpretation of the data and use results where they clearly make sense to do so.
richvoller.com/blog/fan-out-…