Setup & Customization Checklist

1. How this skill is invoked

This is a sub-skill, not a triggered skill. It has no trigger of its own. Your parent lead-scoring skill loads it when the account's stage is Closed Won (paying customer), after running the ACV assessment. It scores and returns.

• {{PARENT_SCORING_SKILL}} exists and: (a) runs the ACV assessment before delegating here, (b) branches customers into this skill instead of prospect scoring, (c) handles the post-scoring steps (CRM sync, tags, alerts) after this returns.
• Product usage data is reachable by the agent (product event history / webhook log / usage tables). Without it, Signal Stack #2 — the highest-leverage signal — is blind and every account will look shallow.

2. The one section you MUST rewrite: product depth

{{DEPTH_SIGNALS}} and {{AUTO_CREATED_ARTIFACTS}} are the heart of this skill and are 100% product-specific:

• List the 3–5 artifacts in YOUR product that a customer only has because they deliberately built them (integrations connected, workflows authored, API keys used, teammates invited...).
• List everything your onboarding auto-creates — and exclude it. The original deployment learned this the hard way: auto-created objects made every account look like a power user until they were explicitly excluded from depth scoring.

3. Calibration decisions

• {{INACTIVITY_CAP}} — default 30 days → cap at second-lowest tier. Match to your product's natural usage cadence (a weekly-use product needs a longer window than a daily-use one).
• Senior Stakeholder Exception — the "VP+ at a quality company scores Gold regardless of seats" override is aggressive by design. Confirm the company-quality bar ($10M+ funded / 100+ employees / known brand) matches your market.
• Persona lists — tune strong/weak to your buying committee.
• Tier definitions — the Diamond bar ("could 3–5x in ARR") should reflect your actual expansion economics.

4. Behavioral invariants (do not remove)

• Never re-research ACV dimensions the parent already assessed.
• Inactivity cap beats every positive signal — churn risk outranks expansion.
• Auto-created artifacts never count as depth.
• Output is always tier + short reasoning covering all four dimensions (depth, persona, headroom, ACV ceiling), then control returns to the parent.