Executive business review generator (PB-CAD-01)

Follow ${CLAUDE_PLUGIN_ROOT}/skills/customer-360/references/playbook-protocol.md with these specifics.

When it applies

An EBR/QBR due within 30 days (cadence.next_ebr_due_at, or a matching meeting in ~~calendar), or renewal ≤ 180 days with no EBR in two quarters, or on request. Choose the variant: renewal when contract.days_to_renewal ≤ 180; expansion when invoked from expansion-whitespace or the user asks for an expansion EBR; else standard.

Inputs beyond the Customer 360

• roi_inputs_json: the loaded hourly rate and any outcome measurements, from ~~crm business-outcome fields and ~~documents (business case, prior EBR, finance export). Never assume a loaded hourly rate or an attribution share; if absent, the ROI table reports the outcome as unvalued with a proposed measurement. Ask the user once whether a rate exists before running; record the answer in the action's data_quality_flags.
• retention_context_json: the account's segment row from ~~data warehouse mart_retention_nrr_grr when present; else the segment benchmarks only (and say the peer comparison is benchmark-based).
• feature_audit_json: run feature-depth-audit as a sub-step (narrative mode) and pass its feature_audit block.
• expansion_brief_json: for the expansion variant, the expansion_brief from expansion-whitespace; else {}.
• Prior EBR material from ~~documents (last deck, notes) so the "where we started" slide uses the customer's own words.

Prompts

1. ${CLAUDE_PLUGIN_ROOT}/prompts/adoption/feature_depth_audit.prompt (narrative mode)
2. ${CLAUDE_PLUGIN_ROOT}/prompts/expansion/executive_ebr_generator.prompt: 8–10 slides, every number sourced, one customer quote max, no pricing anywhere, decision asks with owners and dates.

Artifacts

• PB-CAD-01.executive_ebr_generator.action.json with the ebr block
• PB-CAD-01.ebr_narrative.md: the one-pager
• PB-CAD-01.ebr_script.md: slide-by-slide, with title, claim, talk track, ask the room, visual, sources, risk if challenged
• PB-CAD-01.ebr_roi_table.md: rows with formulas, low/base/high, measured flags, unmeasured outcomes with proposed measurement
• PB-CAD-01.ebr_prework.json: three unknowns, missing attendee, risk to pre-empt, duration
• PB-CAD-01.executive_ebr_generator.comm_1.md: pre-read email to the most senior attendee (approval required)

When the session offers a slides artifact type or ~~documents, offer to build the deck from the script: the script's visual line per slide is the brief for each slide. Keep the script as the source of truth; the deck is a rendering.

Proposed writes

• ~~crm: cadence.last_ebr_prepared_at = today; task "EBR: <account>" with the artifact location; one measurement task per unmeasured outcome, owner and date from the ROI table.
• ~~calendar: if no EBR is on the calendar, a proposed hold with the attendees from ebr.audience including the missing attendee.
• ~~email: draft of the pre-read.
• ~~documents: the narrative, script, and ROI table as one document.

Rules specific to this playbook

• Every number on every slide has a sources[] path. No source, no number.
• Peer comparison uses only the retention context and benchmarks provided: quartile language, never invented percentages.
• The talk track is spoken language in the second person; the customer did the work, the product enabled it.
• Include the strongest negative theme when its urgency is medium or higher, paired with the status of the fix.