A/B/N Experiments
Every message is a hypothesis under test
We never ship a message we haven't measured. Winners graduate into production playbooks. Losers become learnings.
Redacted
Version 12
Angle · Learning Per Touch
Reply14.8%
Meeting5.2%
Accept62%
Positive41%
Redacted
Version 18
Angle · Compounding Pipeline
Reply11.2%
Meeting3.9%
Accept55%
Positive33%
Redacted
Version 22
Angle · Signal, not Volume
Reply9.6%
Meeting3.2%
Accept51%
Positive28%
Redacted
Version 27
Angle · The Cost of Guesswork
Reply12.7%
Meeting4.4%
Accept58%
Positive36%
Redacted
Version 31
Angle · The GTM Flywheel
Reply15.6%
Meeting5.8%
Accept64%
Positive44%
Redacted
Version 9
Angle · Founder POV
Reply4.1%
Meeting1.2%
Accept32%
Positive12%
Redacted
Version 14
Angle · Case Study Lead
Reply5.3%
Meeting1.6%
Accept38%
Positive15%
Winning Hypotheses
Opening with an industry-specific cost-of-inaction stat lifts reply rate 2.4x on P0 accounts.
Replacing product language with 'operating system' framing raises meeting acceptance 38%.
A 3-line message with one hypothesis outperforms 6-line messages on VP-level personas.
Current Experiments
Does referencing a peer company in the same funding stage beat industry-level social proof?
Does a Loom-style async video CTA outperform a direct meeting CTA on CFO personas?
Does a 'learning readout' offer beat a 'strategy call' offer on RevOps buyers?
Rejected Angles
Personalised weather / sports openers — no lift, adds noise.
Multi-question opening lines — depress reply rate 40%.
Product feature lists in first touch — depress positive-reply ratio.