---
id: ins_skill-as-unit-of-distribution
operator: Kyle Poyar
operator_role: Founder, Growth Unhinged; former OpenView
source_url: https://www.growthunhinged.com/p/claude-skills-gtm-and-pricing
source_type: essay
source_title: How to Use Claude for GTM and Pricing
source_date: 2026-04-29
captured_date: 2026-05-02
domain: [pmm, gtm, ai-native]
lifecycle: [tooling-config, ai-workflow]
maturity: applied
artifact_class: framework
score: { originality: 4, specificity: 4, evidence: 4, transferability: 5, source: 5 }
tier: A
related: []
raw_ref: 
---

# The unit of distribution for operator expertise is the Skill, not the prompt

## Claim
Fifteen years of GTM and pricing expertise can be packaged as Claude Skills, not prompts. Skills are versioned, portable, and composable; prompts are not. The strategic implication for any team building a knowledge substrate: ship Skills with decision ledgers, not prompt libraries.

## Mechanism
Prompts are ephemeral text; Skills are addressable artifacts with metadata, versioning, and composition. That makes them distributable, auditable, and improvable, the properties operator expertise needs to scale beyond the original author.

## Conditions
Holds when: the host platform (Claude or similar) supports Skills as first-class objects.
Fails when: the workflow is one-shot enough that the Skill overhead exceeds the prompt's lifetime.

## Evidence
> "I turned 15 years of GTM expertise into Claude skills you can use."
· Kyle Poyar, Growth Unhinged, 2026-04-29

## Signals
- Internal expertise published as Skills with named owners, not as Notion prompt libraries.
- Skills carry version numbers and changelogs.
- Decision ledgers (Brier-scored or similar) attached to each Skill's outputs.

## Counter-evidence
For rapidly evolving tactical knowledge that changes weekly, a Skill's versioning overhead may be too heavy, a living doc still wins.

## Cross-references
- (none in current corpus)