Why does my priority list look different from a colleague's?

Priority tiers are calibrated to the audience description, not fixed by the content. The same 12 core concepts appear regardless; what changes is which ones the analysis judges as load-bearing for each reader.

The gap list is the same; the priorities are not

The same 12 core concepts appear for every audience. ContentGrapher's gap diagnosis is driven by the page content, not the audience description. What differs is which concepts the pipeline judges as load-bearing for the specific reader that was described. A concept essential to one reader may be important to another, even on the same page. Neither result is wrong.

A concrete example

"RAG-based retrieval" is a concept identified on every ContentGrapher analysis of the same page. In analyses run without an audience specification, it is consistently classified as important. In analyses where the audience is fully specified as an expert content strategist working explicitly on LLM retrieval gaps, it is elevated to essential in 4 of 5 runs.

The pipeline's judgment: an expert working explicitly on LLM retrieval gaps should treat the RAG mechanism as foundational, not background context. For that reader, a gap in this concept is not a supporting detail; it is a structural failure.

How the same concept is framed differently

The concept "RAG-based retrieval" is the same in both cases. What differs is why it matters to each reader.

No audience specified: "Understanding that AI systems use retrieval-augmented generation explains precisely why structural completeness and concept coverage determine whether content gets cited."

Expert B2B content strategist, task: identify LLM retrieval gaps: "Expert B2B strategists working on LLM retrieval gaps need to know that ContentGrapher's analysis is grounded in how RAG systems actually select and surface content chunks."

The gap has the same label. The reason it matters to the reader is calibrated to who they are.

How to compare results with a colleague

If you and a colleague analyzed the same URL with different audience inputs, the priority distributions may not match. Check whether your audience inputs were the same. If they differ intentionally, each priority list is correct for the reader it describes. If they differ accidentally, re-run Phase 2 with matching audience inputs to get comparable results.

Re-running Phase 2 with a new audience specification uses one analysis credit. Phase 1 does not need to re-run.