Lead Scorer

Why this exists

A lead list of 80 businesses is not a pipeline, it is a procrastination device. The real question is always the same: which five do I contact this week? Answering it by gut means you contact the ones with the nicest names.

This is the scoring layer from my own prospecting pipeline, rebuilt as a skill anyone can point at a CSV. The key design decision: the rubric is yours, not mine. What counts as a hot lead for a brand designer (no website, just launched) is a dead lead for an accountant. The skill makes you write your sales judgment down once, then applies it consistently to every list you feed it. No API keys, no SaaS, it runs entirely inside Claude Code.

How it works

1. It makes you write the rubric first What you sell, your hot signals, warm signals, cold signals, and hard disqualifiers. Saved as lead-scoring-rubric.md, reused on every future list, edited as your judgment sharpens.
2. It reads the list and says what it can judge CSV, markdown table, or pasted list. It maps your columns to your rubric signals and tells you which signals the data cannot answer, instead of scoring blind.
3. It scores with reasons, not just numbers 0 to 100 per lead, tiered hot, warm, cold, or skip. Every score names the specific signals behind it, plus one concrete next action. Disqualifiers are absolute.
4. It is honest about thin data A lead it cannot judge gets flagged "research" with the missing piece named, never a guessed score. And if every lead scores hot, it tells you your rubric is not filtering.

Step by step (for first-time users)

1. Export or collect your lead list From Google Maps research, a directory, an event attendee list, wherever. A simple CSV with names and websites is enough to start.
2. Open Claude Code in that folder Launch it where the CSV lives.
3. Type /lead-scorer leads.csv First run, it builds your rubric with you: what you sell, what makes a lead hot for YOUR business, what disqualifies one outright.
4. Optionally let it check websites If your rubric cares about web presence, it can fetch each lead's public site and judge what is visible. Only with your go-ahead, only URLs already in your list.
5. Work the ranked file top-down leads-scored.csv arrives sorted, highest score first, original columns intact. Contact the hot tier this week, research the flagged ones, archive the skips.

What each lead gets

Honest take

What it does well: Consistency and honesty. Fifty leads scored by the same written rubric beat fifty scored by mood, and the reasons column means you audit the rubric, not the mystery number. The "research" tier is the feature I am proudest of: most scoring tools hide thin data behind a confident-looking score, and this one refuses to. The summary observations also earn their keep, spotting patterns like "everything without a website scored hot" that tell you where to hunt next.

What it does not do: Find leads. You bring the list; this ranks it. It does not enrich data from paid services, does not connect to a CRM, and absolutely does not contact anyone. The scores are also only as good as your rubric: a vague rubric produces vague scores, which is why the skill makes you fill it in before scoring anything.

When to use it: Every time a lead list is bigger than your outreach capacity, which is every lead list. Score it, work the top tier properly with researched, personal outreach, and let the bottom tier go without guilt.