Why does my analysis show the same core concepts as last time?

The concept set reflects your page content, not your session history. A stable concept list across runs is a quality signal, not a cache artifact.

The gap diagnosis is driven by your page, not your session

The concepts appearing in your analysis are there because Phase 1 found them on the page, or found them structurally absent. This is not a cached result from a prior run. Phase 1 re-runs fresh on each analysis: it reads the current content of the page, builds a new concept map, and produces updated boundary classifications.

If you see the same concepts across sessions, it is because the same concepts are present, or absent, on the same page.

Stability is the expected result

In a 30-run controlled study, the same 12 concepts appeared in every single run across every audience condition tested, without exception. The pipeline found the same things because the same things are present on the page. A stable concept list across sessions is evidence that the analysis is working correctly, not evidence that something is stuck. These figures are from a controlled study on a single URL; counts on your page will vary with your content.

Two stacked panels comparing concept rows for Intermediate and Expert audience conditions. Both panels show six concepts with identical boundary badges and integration state badges. Priority tiers differ only for HNSW index, which rises from Important to Essential for the Expert audience. A divider between panels reads same 11 concepts, same boundary decisions, priority tiers shift.

What the level does and does not change

The knowledge level does not change which concepts appear. The level changes two things: how those concepts are prioritized (which are classified as essential, important, or useful) and how the writing guidance prescribes fixes for each gap. Changing the level between sessions will not produce a different concept list, but it will produce different priority tiers and different writing guidance.

Phase 2, which produces the priority assignments and writing brief, requires an analysis credit. Phase 1, which produces the concept list and boundary classifications, is free.

Two-column reference card divided by a vertical audience label. The left column lists Phase 1 outputs fixed by page content: concept list, boundary decisions, and integration states. The right column lists outputs shaped by the knowledge level: priority tiers, writing brief emphasis, and structural recommendation weighting. A footnote explains that changing the level re-runs Phase 2 only.

If you want the gap list to change

The only way to get a different concept list is to change the content on the page and re-analyze. Phase 1 will detect the change and produce an updated concept map. If you added or removed a major concept from the page, the next Phase 1 result will reflect it.