Build research-backed cold outreach sequences
cold-outreach-sequenceskillsetup L2★293
BrianRWagner/ai-marketing-claude-code-skills ↗Causal-lift measurements
outreach-campaigning+10pp vs no-skill baselinewith-skill 77% · baseline 67%
Measured by running the task with and without this artifact, K=5, graded by deterministic checks — no LLM judging.
What it does
Build multi-touch cold outreach sequences with research signals
Best for
Individual cold outreach at small scale (5-50 prospects) where research and personalization per prospect is viable and each message needs to reference a named signal to clear inbox clutter.
Inputs
- · Prospect name, company, title
- · Research signals (funding news, LinkedIn posts, job postings, hiring, product launch)
- · Sender positioning (what sender does, for whom, with what result)
- · Platform (LinkedIn DM, email, or both)
Outputs
- · 4-touch sequence (connection request + first message + 2 follow-ups)
- · Tier 1/2/3 variants based on research signal strength
- · Objection pre-emption angles
Requires
- · Web search (funding, news, LinkedIn posts)
- · Optional: positioning-basics output (sender context)
Preconditions
- · Research signals gathered (do not write Tier 1 without named signals)
- · Sender positioning known
Failure modes
- · Writes Tier 1 without research signals (defaults to generic tier 3)
- · No specific reference to signal in opener (prospect thinks 'nice try')
- · Repeats value prop in follow-ups without new angle (stales fast)
- · 'Following up' message with zero new information (auto-delete)
Trust signals
- · Signal classification (recent news/funding strongest, LinkedIn post strong, company stage medium, role-only weak)
- · Personalization tier rules (Tier 1 = named signal, Tier 2 = segment template, Tier 3 = volume with signal warning)
- · Formula for connection request (no pitching, prove research, one sentence, conversational)
- · Formula for first message (thanks + bridge to relevance + light value + soft question)