The skill
Install once in Claude. It audits your AI citation share across the engines and drafts citable content on demand.
What's inside
Three files, one job: get quoted by the engines
Built for marketers, not search engineers.
Install guide
A short step-by-step for setup, plus a no-install option if you cannot add skills.
Plan B prompt
A single prompt you paste into any Claude chat to run the same audit and drafting without installing anything.
How it works
Three steps, from a prompt log to citable content.
Step 1Bring your input
Either a log of buyer prompts run across the engines, or a target query you want to win. No log yet? The skill builds the prompt library and the run instructions first.
Step 2Claude does the work
It computes your share of voice, citation share and sentiment per engine and maps each gap to a content, credibility or entity cause, or it drafts the answer capsule, triples, table and FAQ for your target query.
Step 3Get a clear output
A visibility audit with a fix for every gap, or a set of citable components with a checklist pass.
There is no live connection to the engines. You run the prompts and bring the log, the skill does the analysis and the drafting. That keeps the numbers yours and reproducible.
How to get cited by AI answer engines
Getting cited by AI is not search engine optimization with new keywords. The pages answer engines cite often do not rank in Google, brand mentions matter more than backlinks, and answers shift on every run, so you measure presence as a percentage, not a position. Here is how the skill audits your citation share and engineers content that gets quoted.
Why AI citation is not the same as SEO
The pages AI engines cite usually do not rank well in Google. About 90 percent of ChatGPT's cited pages sit at position 21 or worse, and only around 12 percent of AI-cited URLs rank in the top 10. At the same time, brand mentions across the web correlate with AI visibility roughly three times more strongly than backlinks. So the job is not to chase a keyword ranking, it is to be the named brand the model already knows and to publish passages it can quote cleanly.
How to measure AI visibility
AI answers are non-deterministic, so a single run tells you nothing. The skill applies one consistent audit logic over a library of 30 to 100 non-branded buyer prompts, each repeated 60 to 100 times per engine, with ChatGPT, Perplexity, Gemini and AI Overviews kept separate. It reports visibility as percentages and trends, never as a rank position, and tracks three things, share of voice, citation share and sentiment. Top brands appear in roughly 55 to 77 percent of category runs, so that band is the benchmark to read your own presence against.
How the skill diagnoses a citation gap
For every prompt where a competitor is cited and you are not, the skill inspects the sources the engine actually used, then assigns one of three causes. A content gap means there is nothing extractable from you to answer the prompt, and the fix is structured, citable content. A credibility gap means competitors are cited from third-party sources you are absent from, a Reddit thread, a review site, a YouTube video, and the fix is earning presence there. An entity gap means the models do not yet recognize you as a known brand, and the fix is consistent named presence across credible sources.
How to engineer content that gets cited
The skill drafts the components that get extracted. The core unit is the answer capsule, a 40 to 60 word direct answer placed under a question-format H2, leading with the answer, carrying a number or named source, and containing no links, so it can be quoted in isolation. It adds semantic triples written as plain subject-predicate-object claims, comparison tables and real FAQ blocks, and it front-loads this into the first 30 percent of the page, where most citations are drawn from. It also flags promotional language to cut, since that correlates negatively with citation, and it tells you not to waste time on an llms.txt file, which no major engine uses.
Where this fits in your strategy
Keep it in proportion. AI referrals are still about 1 percent of web traffic, though they tend to convert far better, so this is a shift in emphasis you start now at low cost, not a reason to abandon search. The honest framing is that this is a re-weighting of SEO's priorities, entities and mentions over links, passages over pages, not a separate discipline you start from scratch. For the full evidence, including the studies on click collapse and the mention-versus-backlink gap, read AI Engines Cite Pages That Don't Rank in Google.