# ICP Lead Scorer (Claude Code command)

Stop working your lead list top-to-bottom. This command scores any CSV of leads
against your ICP and re-ranks it so the first call of the day is the best call
of the day.

**Install:** save the block below as `.claude/commands/score-leads.md`. Define your
ICP in `CLAUDE.md` first (grab the template from the
[Claude Code Prospecting Kit](https://www.salescadia.com/free/claude-prospecting-kit.md)).

Run it like: `/score-leads leads.csv` (any CSV with at least company + title columns)

```markdown
Score and re-rank the lead list in $ARGUMENTS against our ICP (see CLAUDE.md).

Process:
1. Read the CSV. Tell me which columns you found and which ICP inputs are
   missing (e.g. no industry column) BEFORE scoring.
2. Score each lead 0-100:
   - Title match to our buyer titles: 0-30
   - Company fit (industry, size): 0-30
   - Buying signal present in the data: 0-25
   - Data completeness (email, phone, domain): 0-15
   Show the rubric you actually used at the top of the output.
3. Deduplicate on email (case/whitespace-insensitive) and on company+name.
   Report how many dupes you merged.
4. For anything scoring 80+, use web search to verify the company still matches
   (right size, still hiring, not shut down). Note "verified" or "stale".
5. Write scored-leads.csv: original columns + score + one-line reason,
   sorted descending. Print the top 10 as a table.

Rules: never invent values for empty cells. A blank stays blank and costs
completeness points. If the file has under 5 usable rows, say so and stop.
```

**Why the rubric is weighted this way:** in our own outbound data, title match and
timing signals predicted replies far better than company size did. Steal the
weights, then tune them against your last 200 sends.

---

From [Salescadia](https://www.salescadia.com) — our engine scores every sourced
lead continuously and prioritizes sends by predicted response, not list order.
See your ICP built live: [salescadia.com/gtm](https://www.salescadia.com/gtm).