코트

Three numbers per situation: how big the situation is (a ceiling), how strongly it already leads to the category (connection signal), and the difference (unconnected searches). The bigger the difference, the sooner content should make the connection.

Only searches that name an audience (the recipient, companion or user) are counted, and their monthly volume summed. It reads who people search for, and with whom, straight from the queries.

A CEP is a situation that many search signals point to together, not a single keyword. This shows how much related-search, search-path, search-cluster, SERP text, PAA and AI Overview data was collected, and which of it served as evidence for each CEP's scene (when, where, why, with whom) and its nano context, KBF, constraints and avoidances.

Dossier extraction: scene 7W 100 · attributes 110 · detailed situations 10 (71 failed pointer checks)

Where each CEP showed up

CEPs surface in different places depending on the topic and situation: some in search paths, some in clusters or related searches, some in AIO and PAA. Left: share of the CEP's measured query volume found in each source (discovery). Right: how often each source was cited for the scene 7W and attributes. 'none' means that source never reached this CEP.

For each CEP's queries we checked whether category brands are mentioned in Google's SERP and AI Overview, weighting the AI Overview body most. Unlike paths (the brand consumers searched next), this is the brand search AI offers as the answer for that situation. Situations where no brand appears are where content and GEO can claim the spot first.

Of 10 CEPs with SERPs checked, 9 have no category brand in the AI search answer at all.

• Any brand share / response-priority judgment ("how a brand is doing") holds only where brand linkage was actually measured for that CEP; check the measured ratio in Evidence readiness just below. Where many CEPs are unmeasured, treat brand-related conclusions as provisional.
• Search reflects familiarity and purchase-consideration more strongly than unaided recall; in person-level validation, recognition alone drove search only 1.22x (near-negligible), while familiarity/consideration were clearly larger, reaching 7.0x (smartphones) / 5.3x (autos) when all 5 positive brand attitudes were held (Dotson et al., Google). This correlation is confounded by brand size (bigger brands move on many metrics at once), and validation was done at brand-total level.
• Existing/habitual buyers don't disappear from search; the same study found brand owners searched their brand 2.5-3.5x more than non-owners (Dotson et al., Google). But much of that is non-need-expressing (brand-name/navigational/support queries), so this map's need-type CEP distribution skews toward new-entry/switch-consideration demand.
• Brand share is relative within the measured set; an unqueried brand x CEP pair is [no data], not zero.
• Paths are aggregate graph flow, not one person's timeline; "transition" is a modeled statistic.
• CEP power's denominator is classified queries, not people; it isn't directly comparable to survey-based CEP power (denominator = people); the same person's 10 queries count as 10 votes on this map.
• Queries that pick by naming a brand/origin are closer to a search mode than a situation, and differ in kind from other CEPs - they're bucketed separately to avoid contaminating need-type CEPs, so when counting "number of entry situations," we recommend excluding this item and the unclassified long tail.
• Before decisions like budget/resource allocation, we recommend pairing this with survey-based recall measurement (person-level CEP power / share of mind).
• The code discarded 68 AI-suggested item(s) lacking evidence (evidence is enforced).
• Of 220 qualitative attributes: 28 measured (13%), 123 consensus (56%), 61 derived, 8 assumption. Measured means the condition exists as search queries, not that it drives purchase. Consensus is the social common sense Google has organized, not a fact check.