setup-customization-checklist.md
Setup & customization checklist
Work through this once before enabling the skill; revisit when your GTM motion changes.
Placeholders (all required unless noted)
| Placeholder | What to set it to |
|---|---|
{{PRODUCT}} |
Your product's name as it appears in memory notes and alerts |
{{CRM}} |
Your CRM (HubSpot, Salesforce, Attio, …) |
{{SCORING_SKILL}} |
The account-scoring skill this feeds. The champion move enters it as a first-party, high-intent signal — the scoring skill owns tiers, gates, MQL routing, and outreach downstream |
{{LINKEDIN_LOOKUP}} |
A live LinkedIn profile lookup (e.g. an Apify profile-search actor) that returns the full experience section — the verification rules depend on it |
{{MQL_OWNER}} |
The person/queue that owns champion-move tasks and alerts |
{{DIGEST_CHANNEL}} |
Channel for the monthly digest |
{{COMPETITORS}} |
Your direct-competitor list; moves into them are logged, never scored |
{{WATCHLIST_MAX}} |
Default 5 per account. Raise only for very large committees |
{{REFRESH_WINDOW}} |
Default 90 days between watchlist rebuilds |
{{USAGE_WINDOW}} |
Default 90 days of usage for the "materially active" bar |
{{SELF_SERVE_THRESHOLD}} |
Default <50 employees AND ≤$10M raised — below it, moves are FYI/future pipeline, never MQLs |
{{TIER_FLOOR}} |
The floor a confirmed champion move scores (default: one tier above mid). Must NOT override your scoring skill's hard gates |
Policy decisions to make deliberately
- Per-person usage attribution. Identify where your product records usage by individual user (events, credits, seat activity). Org-level totals cannot support the heavy-user leg — if you only have org-level data, the heavy-user leg degrades to the champion leg only; say so in the digest.
- Tag + stage conventions. Create a
champion-movetag (or your equivalent) and confirm which stage your MQLs enter. The tag + an active carried-forward watchlist entry are what keep non-customer companies in the monthly scan set. - Outreach stays human. This skill never sends. Confirm your scoring skill's outreach steps are queued-for-approval unless you have explicitly opted into auto-send there.
- Prior-owner etiquette. Decide how the prior relationship owner is surfaced (recommendation inside the owner's task is the default) — not as an automatic sender.
- Never-prospect list. Confirm standing exclusions (advisors, partners, government, VC/PE) live where the scoring skill's gates can see them.
Behavioral invariants (do not remove when editing)
- MOVED requires affirmative evidence from the experience section, plus the wrong-person guard (origin company must appear in the destination profile's history).
- The B.0 qualification gate runs before any scoring: realized value AND not a recent closed-lost. Warm familiarity is not advocacy.
- One move → one MQL, ever (the
MOVED → processedmarker is the dedup record; never delete it). - Multiple movers to one destination = one signal, scored once.
- Carried-forward entries survive watchlist rebuilds.
- UNCERTAIN never fires the MQL flow.