Personalize outbound sequences with signal evidence, not mail merge

The problem

Your sequences use {{first_name}} and {{company}}. Reply rates are under 1%. Reps say personalization does not scale. They are right, if personalization means manual research.

The system

1. Signal-to-copy pipeline. Buying signal → evidence extraction → first-line generation.
2. QA sampling. 10% of generated copy reviewed by a human. Failures feed back to prompt tuning.
3. A/B testing. Reply rates tracked by personalization type. Winning patterns promoted.
4. Scale via automation. 500 personalized first lines per day, each referencing a real signal.

Field notes

Evidence-based personalization at scale is an engineering problem, not a headcount problem. The signal engine feeds the copy engine.