Signals
What is the market asking for beneath job titles? Enrichment subset only — domains/themes excluded.
Filters (above) apply to every panel below except Skill Co-occurrence Matrix and Normalized Skills, which stay full-snapshot (the skill vocabulary is too large to bitmask cheaply). Summary above always describes the full snapshot.
Behavioral signals
Bar length = rate — the most accurate way to compare a dominant signal against a long tail. Closed vocabulary (max 6 per posting). Rates sum >100% — multi-label.
Role score profiles (0–10)
Ridgeline of the four enrichment score axes; each ridge scaled to its own peak — compare shapes and medians, not heights. Dot = median.
Signal density per posting
Composition — each posting counted once; bars sum to the enriched subset. Closed vocabulary caps at 6.
Hard gates
Tile area = share of total gate mentions (not of postings — one posting can carry several gates). Bounded vocabulary. Enriched subset only.
Gate load — how many barriers per role
Postings by number of distinct hard gates carried. Roles with 3+ simultaneous gates are the most constrained searches.
Normalized skills
Inferred categories — directional, not canonical. Enriched subset only.
AI exposure by function
Current snapshot — cross-sectional only. Trend requires ≥2 captures.
Archetypes — prevalence vs function concentration
Each dot is one archetype: right = more common market-wide, up = more concentrated in a single function (100% = found only in one function). Bubble size = posting count. Bottom-right = common and spread across every function; top-right = common and function-specific. Inferred categories — directional. Enriched subset only.
Signal concentration by function
Each row (signal) sums to 100% across functions — shading shows where that signal's postings concentrate, not each function's overall signal rate. Reveals which functions actually drive each behavioral signal.
Skill co-occurrence matrix
Viz delta: symmetric matrix instead of force network. Diagonal = postings mentioning the skill; off-diagonal = postings mentioning both; 0 = never co-listed. Inferred labels — directional.
Outcome systems by function
Each row (outcome) sums to 100% across functions — shows which teams are actually accountable for each outcome. Bounded outcome schema. Enriched subset only.