---
id: "claim-70-20-10-development-loss"
type: "claim"
source_timestamps: ["§ 2. Build a Distributed Apprenticeship Pipeline"]
tags: ["pedagogy", "experiential-learning", "metrics"]
related: ["concept-knowledge-cliff", "entity-center-for-creative-leadership"]
confidence: "medium"
testable: true
speakers: ["Jenny Fernandez"]
sources: ["reskilling"]
sourceVaultSlug: "hbr-seg-reskilling"
originDay: 10
articleStem: "hbr-sig-51-talent-strategy-ai-transformation"
sourceUrl: "https://hbr.org/2026/06/your-talent-strategy-has-to-keep-up-with-your-ai-transformation"
sourceTitle: "Your Talent Strategy Has to Keep Up with Your AI Transformation"
---
# Eliminating Entry-Level Roles Removes 90% of the Professional Development Model

**Claim (confidence: medium · testable: true).** Relying on the Center for Creative Leadership's **70–20–10 framework** (see [[entity-center-for-creative-leadership]]), the author argues that because 70% of professional development comes from on-the-job experience and 20% from relationships with experienced colleagues, hollowing out entry-level roles destroys **90%** of the traditional development model. Only the 10% derived from formal training remains — vastly insufficient for building AI-era skills like resilience, adaptability, and social influence. This drives the [[concept-knowledge-cliff]] and is the assumed premise of [[prereq-70-20-10-framework]].

**Enrichment / verification.** The **70–20–10 numbers themselves are validated** — CCL popularized the model and it is widely used in HR/L&D design. The **90% loss figure, however, is interpretive, not empirical**: it assumes that almost all 70–20 (experiential + relational) development happens *inside entry-level roles*. In reality, mid-career lateral moves, cross-functional projects, and stretch assignments can also deliver 70–20 learning. A more precise statement: eliminating entry-level roles **severely undermines the main early-career channels** for experiential and relational learning. Directionally correct (major damage to experiential learning), quantitatively speculative — hence **confidence: medium**. This is the single most contestable number in the source; flag the distinction when asked.