The Growth Architect: Psychology-Driven Marketing for AI Products

Most marketing advice is generic. "Write good copy." "Know your audience." "A/B test everything." It's not wrong—it's just not useful when you're launching an AI product into a market that doesn't fully understand what AI can do.

This framework comes from months of testing what actually works for AI-native products. The core insight: psychology trumps features.

The Psychology-First Approach#

Before you write a single line of copy, understand this: your B2B buyers fall into four psychological profiles, each requiring different messaging:

The mistake most AI companies make: they lead with NT messaging (innovation, technical superiority) when 40% of their market wants Guardian messaging (safety, proven results). Research on B2B buyer psychology confirms 72% of buyers expect role-specific content.

The Growth Architecture#

Stop thinking about "marketing" and start thinking about growth systems. A growth system has three components:

1. Viral Loops#

K-factor = (Invitations Sent × Conversion Rate) / Cycle Time

If K > 1.0, you grow exponentially. Most companies never calculate their K-factor. Do the math.

Optimization levers (Kurve's deep-dive on K-factor):

• Invitations: Embed sharing in the onboarding flow (typically +20-50% based on implementation)
• Conversion: Dual-sided incentives (can improve rates 2x over single-sided)
• Cycle time: Immediate value delivery (reduces time-to-share significantly)

2. Network Effects#

Not all network effects are created equal. Diagnose which type applies to your product:

• Direct: Each user adds value for all users (collaboration tools)
• Indirect: Two-sided marketplace dynamics (platforms)
• Data: More usage = better product (ML-powered features)
• Social: Status/reputation mechanisms (communities)

For AI products, data network effects are usually strongest. More usage → better model → more value → more usage.

3. Funnel Optimization#

Stop optimizing what's working. Fix what's broken first.

Baseline metrics (B2B SaaS):
- Visitor → Lead: 2-5%
- Lead → Trial: 15-20%
- Trial → Paid: 15-25%
- CAC:LTV target: 1:3+

Find your biggest constraint. If your visitor-to-lead rate is 0.5% when the benchmark is 2-5%, that's your bottleneck—not your trial-to-paid rate. 2025 benchmarks show PLG products achieve 9% visitor-to-trial while sales-led B2B averages 1-2%.

The AI Product Playbook#

AI products have unique challenges. People don't understand what AI can do. They're afraid of being replaced. They've been burned by overpromising AI hype.

Education-First Funnel#

Awareness  → "What is [AI capability]?"
Consideration → "How does it work?"
Decision   → ROI calculator + case studies
Success    → Implementation playbooks

Don't lead with features. Lead with jobs-to-be-done. Not "Our AI uses RAG with vector embeddings" but "Answer any question about your documents in 10 seconds."

Trust Building Ladder#

1. Free tools/calculators → Prove competence
2. Case studies with metrics → Prove results
3. Pilot program design → Reduce risk
4. Enterprise rollout plan → Scale success

The ladder matters because AI trust is low. You can't skip steps. A company that's never used AI won't sign a 6-figure contract on your first call—no matter how good your demo is.

Cognitive Bias Application#

Use psychology ethically:

The key word is ethically. Dark patterns destroy trust. In AI products, trust is everything.

30-60-90 Day Roadmap#

Days 1-30: Foundation#

• Implement tracking for K-factor components
• Baseline personality segmentation in analytics
• Launch one quick-win experiment

Days 31-60: Experimentation#

• A/B test messaging by psychological profile
• Launch viral loop v1
• Implement psychological triggers at value moments

Days 61-90: Scale#

• Roll out winning variations across channels
• Optimize K-factor toward 1.0
• Expand to new segments with proven playbook

Related Frameworks#

This connects to several concepts in our AI Dictionary:

• Feedback Loops — Growth systems are closed-loop control systems
• Agent Orchestration — Multi-agent systems for automated marketing workflows
• Cognitive Dataflow — Parallel testing and optimization pipelines

See also: YAML to Agentic Runners for how we automate marketing workflows with CDO.

---

Further Reading#

• First Page Sage: B2B SaaS Funnel Benchmarks 2025
• Kurve: K-Factor & Viral Retention Guide
• WDG Agency: MBTI Buyer Personas for B2B
• Brian Balfour: Four Fits Framework
• Cialdini: Principles of Persuasion
• Product-Led Growth Collective
• Reforge: Growth Strategy
• BJ Fogg: Behavior Model
• Estha.ai: Embeddable AI Widgets for Viral Growth

Adapted from our internal Growth Architect system. For implementation support, see our consulting services.