What problem does it solve? Raw reply rate hides whether a cold email campaign is actually working — a 5% reply rate full of unsubscribes and "not a fit" responses is worse than a 1% reply rate from interested buyers. This Skill classifies every reply from a Smartlead campaign and computes the positive reply rate, the metric that actually predicts pipeline. ## Core Features & Use Cases - Reply Classification: Fetches all replies from a Smartlead campaign via its API and classifies each into 11 labels (positive_interested, positive_soft, positive_referral, neutral, negative, unsubscribe, OOO, bounce, etc.) using Claude. - Rate Aggregation: Computes positive reply rate, positive share of replies, hostile rate, and unsubscribe rate, excluding OOO and bounces from denominators, with B2B benchmarks for interpretation. - Action Flagging: Surfaces positive replies needing immediate human response, referrals to follow up, and hostile replies that signal deliverability risk. - Use Case: After a campaign has run 14+ days, score it to decide whether to scale, iterate, or kill it — and compare two experiment variants using the same cutoff date. ## Quick Start Ask the AI to score the replies for Smartlead campaign 12345 and report the positive reply rate with a breakdown by classification.
