lead-scoringlisted

# Account Qualification ## Pipeline contract **Stage 10 of 20.** Reads `08-leads.csv`, `02-client-profile.yaml`. Writes `10-scored.csv` (adds `screen` pass/fail/review, `screen_reason`, `fit_score` 0-100, `fit_components`). - Run Level 1 on every row, Level 2 on rows that pass. Level 3 belongs to [`/lead-qualification`](../lead-qualification/SKILL.md) and [`/account-notes`](../account-notes/SKILL.md). - This stage scores FIT only. Intent is stage 9's data and is combined in stage 11 — keep them in separate columns. - Missing data stays blank and scores as unknown, never as zero. Put what was missing in `fit_components`. - Weights come from the profile's hard/soft split. Propose weights to the human once per run; do not invent new ones mid-run. Files live in `runs/<run>/`. Column definitions: [`/prospecting-system`](../prospecting-system/SKILL.md). Standalone use: ignore this block. ## When to Use - Evaluating a new lead list before launching campaigns - Prioritizing which accounts to target first from a large TAM - Building qualification workflows for SDRs or AI enrichment pipelines - Deciding whether to continue pursuing an account after no initial response ## Framework ### The Qualification Stack Account qualification happens at three levels. Each level adds confidence but also adds cost (time, data credits, manual research). Match the depth to the deal value. ``` Level 1: Automated Screening (seconds per account) Filters: firmographic + technographic data