Written by Roman ChornovolCitation and source analysis is the practice of identifying which external sources and narratives an AI model draws on when constructing answers about a category, so practitioners can act on the inputs that shape the output rather than reacting to the output alone.
Also known as: citation and source analysis, source analysis
What It Is
When an AI engine answers a question about your category, recommending a tool, describing a market, or framing a problem, that answer isn’t generated from nothing. It’s assembled from sources the model has learned to trust: publishers, forums, analyst reports, documentation, and the accumulated consensus those sources represent. Citation and source analysis is the discipline of tracing that chain: which sources are being drawn on, what narrative those sources carry, and how heavily any single frame is weighted in the resulting answer.
The output you see (a mention, a recommendation, a framing) is a symptom. The sources and the narrative consensus behind them are the cause. Citation and source analysis works at the level of cause.
Why It Matters for AI Search
AI search doesn’t rank pages. It recommends from a learned story about what your category is, who the credible players are, and what problem they solve. That story is built upstream, in the sources the model absorbed before you ever checked your visibility score. A brand can be mentioned in an AI answer while an entirely different company’s frame defines the recommendation. Mentions and narrative share aren’t the same thing.
This distinction has direct consequences. Optimising for appearances, adding schema, seeding Q&A pages, chasing citations as an end in themselves, addresses the surface. It doesn’t change whose frame the model uses to build the answer. If the sources the model trusts are telling a story anchored in a competitor’s positioning, tactical content won’t move that narrative. Source analysis reveals the gap between where you appear and whose story is actually winning.
The Mavel Framing: Narrative Share Over Presence
Most visibility tools count whether you appear in AI answers. Citation and source analysis asks a prior question: whose frame is the answer built on, and why? AI is downstream of consensus. The model reflects the narrative that the most trusted sources have already settled on. Understanding that consensus means reading the sources, not just the outputs.
Mavel treats citation and source analysis as the foundation of narrative intelligence. Rather than reporting a presence score, the goal is to trace the actual sources and citations shaping the AI’s frame, identify the narrative those sources carry, and surface where a brand’s representation is absent, thin, or wrong. That way, the work that follows targets the inputs that move the recommendation, not the score that describes it.