Learning velocity
Every experiment we're running for you
Entropy runs continuous, hypothesis-led experiments across outreach, messaging, ICP, content, engagement and AI workflows — so every week your engine gets sharper.
Outreach Experiments
Winner
Test benchmark-led openers vs. question-led openers on CFO persona
Learning
Benchmark openers outperform 2:1 at n=412
Result
+47% reply rate
Winner
Founder-to-founder voice vs. SDR voice on Series B AI accounts
Learning
Founder voice lifts positive replies 3.1x
Result
+210% positive reply ratio
Rejected
Async video CTA vs. calendar CTA on VP Operations
Learning
Calendar CTAs win in ops-heavy verticals
Result
-32% response rate
Messaging Tests
Winner
'Learning velocity' angle vs. 'ROI' angle on Series B founders
Learning
Learning velocity wins in AI · SaaS
Result
+38% meeting acceptance
Winner
3-line message vs. 6-line message on VP-tier
Learning
3-line format wins for VP+
Result
+29% reply rate
Winner
Regulatory tailwind opener on FinTech CFOs
Learning
Regulatory framing lifts reply 47%
Result
Confirmed at n=284
ICP Research
Complete
Buying committee mapping for enterprise AI
Learning
5-node committee is standard; CIO gates budget
Result
Reduced sales cycle 28%
Active
Series A vs. Series B GTM readiness signals
Learning
Series B is the tipping point for outbound receptivity
Result
In progress
Content Experiments
Winner
Teardown format vs. thought leadership on LinkedIn
Learning
Teardowns drive 4x inbound
Result
+128k views on first teardown
Winner
Carousel length: 6 vs. 10 vs. 14 slides
Learning
10-slide is the engagement peak
Result
+42% engagement
Winner
Founder POV vs. brand voice on B2B posts
Learning
Founder POV drives 3.2x engagement
Result
Adopted across book
Engagement Patterns
Winner
Weekend sends vs. weekday sends for founder persona
Learning
Weekend outperforms Monday 1.6x for founders
Result
Adopted for founder-tier
Winner
Reply-to-comment engagement window
Learning
First 3-hour window captures 78% of engagement
Result
New response SLA set
Campaign Optimization
Complete
Variant promotion criteria (statistical confidence bar)
Learning
n=200 + p<0.05 is the safe promotion bar
Result
Standardised across campaigns
Active
Campaign cool-down window between sends
Learning
Testing 4 vs. 6 vs. 8 day windows
Result
In progress
AI Workflow Improvements
Winner
AI-assisted variant generation vs. human-only
Learning
AI-assisted lifts iteration speed 3.4x with equal quality
Result
Adopted in workbench v2
Active
AI enrichment for buying committee signal detection
Learning
Signal detection precision at 84%
Result
In progress