Every new case published, every offense type added to the taxonomy, every scoring change. The record of what changed and when.
Self-audit tool: paste any cold outreach message and get a two-pass forensic analysis. Heuristic regex pre-screen followed by a full Claude API pass against the 30-offense taxonomy. Results include per-offense span annotations, a weighted score, percentile rank against the 300-case reference set, and a materially rewritten version.
Annual research report covering the 40% decline in reply rates since 2019, the AI volume acceleration, the quality gap between top-quartile and median senders, and the buyer's view. Four sections with bar charts, stat pulls, evidence tables, and research citations woven into prose.
Each /tag/:offense page now includes the coinage block, example with why-it-fails explanation, research validation stat, co-occurrence data from the 300-case reference set, fix prescription, and a sidebar with all offense navigation.
Dedicated pages for each industry (SaaS, FinTech, Cybersecurity, MarTech, HR Tech, Real Estate Tech, DevTools) and each channel (LinkedIn InMail, Cold Email, SMS) with aggregate stats, pattern analysis, and case indexes.
Ranked leaderboard across four dimensions: worst scores, most offenses per message, most common offense types, and performance by channel.
LinkedIn InMail · SaaS · SDR. Auto-fail triggered by MERGE-FAIL. Four offenses identified: MERGE-FAIL, FAKE-RESEARCH, VAGUE-ICP, PREMATURE-ASK.
Email · FinTech. Four offenses. Generic growth compliment as opener, buzzword-dense value prop, FOMO-BAIT close.
New offense type added after appearing independently in three cases. Definition: asserting the recipient is experiencing a specific problem without evidence they are. Severity: Major.
Email · Cybersecurity. First case to trigger the new PAIN-ASSUMPTION offense type. Also the first Cybersecurity case in the archive.
Email · HR Tech. 0.8/10 - the lowest score in the archive. Five offenses across a 4-sentence message. Every element failed simultaneously.
Updated scoring methodology: any message containing a MERGE-FAIL offense is capped at 2.0/10 regardless of other content quality.
Site launched with 4 cases, 6 offense types, and the initial scoring rubric. FOMO-BAIT and REPLY-ALL-ENERGY added in the first week.