CRO & Stakeholder Negotiation Suite

This skill exists to stop: auditing UX by feel ("looks off") and optimizing conversion with dark patterns — instead of pointing at specific elements with sourced rules.

🔀 Division of labor with ux-review: this skill covers CONVERSION + NEGOTIATION; general UX quality/usability/heuristics → use ux-review (with a manual heuristic checklist).

📁 Source note: [sage] = upstream Sage repo (github.com/xoai/sage, public) — optional deeper reading; this skill runs fully on the rules inlined here. A step marked MUST READ points at a file in your own project (e.g. an event registry) — if it is missing, stop and ask instead of improvising.

🤖 0. HOW TO USE (agent workflow)

A. AUDIT a page/screen ("audit this landing"): run the 3-question filter (§3) on the flow's main screen → sweep the 15-topic table (§7), only reporting rows with real findings, each pointing at a concrete element → check triggers/popups (§4) and visual safety (§5) → enumerate the non-happy paths (row 15: every async/AI/network step must have a specified failure, empty and low-confidence state). Output table: # | exact location | disease | fix | severity. Generic remarks ("UI feels dated") are invalid.

Severity scale: launch-blocking = breaks or coerces the core arc of the flow (§3 three questions unanswered, dark pattern, unspecified failure branch, payment/registration wall before any value) · should-fix = measurable conversion or trust leak with a known fix · nice-to-have = polish. Rows that don't apply to the surface type (e.g. hero/social proof on an in-app result screen) are listed once as "swept, n/a".

Spec-only input (no screenshot / URL): still run the full audit; write "exact location" descriptively and tag it [UNVERIFIED]; close with the list of what you need to see. For a scored deliverable use templates/ux_audit_scorecard.md (1–9 per topic, P0/P1/P2 map to launch-blocking/should-fix/nice-to-have).

B. NEGOTIATE with stakeholders ("boss dislikes it", "marketing wants popups", "brand demands the color"): identify which group in the fear map (§2) → speak THEIR language of concerns → never oppose the goal, change HOW it's reached (move the trigger, re-zone the space). Output: a concrete dialogue script (feeling→goal question, trade-off option) per §2 templates.

C. VERIFICATION plan ("test with users before dev"): assemble the 5-user guerrilla plan (§6): single task, observation criteria, evidence capture.