Citation-source strategy for GEO advisory: the methodology

Citation-source strategy is the part of GEO advisory most platforms don't talk about — because most platforms don't do it. They tell you which prompts you're invisible in. They show you which competitors are louder. What they don't tell you is where to go to get cited.

Citation-source strategy answers that question. It's the methodology by which we work backwards from an LLM's actual answer — past the prompt, past the score — to the specific publications, comparison articles, podcasts, schema-rich pages, and structured-data sources the model pulled from. Then it's the plan for getting your business into those sources.

This piece is the methodology. The companion piece, a worked example, applies it to a real category.

Article structure (in progress)

1. What a citation source is. The distinction between a "mention" (LLM names you) and a "citation source" (what the LLM read in its training data or live retrieval to know about you). Why the latter is the lever.
2. How LLMs select citation sources. A non-technical model of training corpus weighting, retrieval-augmented generation, freshness signals, and entity disambiguation. What changes between models. What stays stable.
3. Mapping the citation-source landscape for a category. The five buckets we always check: (a) trade press and broadsheets, (b) comparison articles and listicles, (c) authoritative directories, (d) Q&A communities and forums, (e) the business's own structured content (FAQPage, glossary, methodology pieces).
4. Identifying gaps. Where AI is citing your category but you're absent. Where AI is citing you but framing you wrong. Where AI is citing nothing — and the opportunity to be the canonical source.
5. Prioritising moves. The 12 type-tags we use (PR, Content, Schema, Owned, Off-site) and how to sequence them by leverage and time-to-impact.
6. Measuring movement. What to expect from re-measurement at 12 weeks, 6 months, and 12 months. How to attribute score changes to specific citation-source moves.
7. The honest limitations. What we don't know. What changes between LLM updates. Why we don't guarantee outcomes.