---
type: "synthesis"
sources: ["spine"]
tags: ["experimentation", "lean-startup", "prototyping", "operating-discipline"]
id: "cd-experimentation-operating-mode"
sourceVaultSlug: "hbr-seg-spine"
originDay: 1
articleStem: "hbr-seg-spine"
sourceUrl: "(unified vault: 9 sources)"
sourceTitle: "HBR — Strategic Spine — value thesis & how much to bet"
---
Beneath the strategy debates sits a shared *tempo*: don't plan the transformation, run cheap experiments and learn. Four articles converge on lean-startup-descended practice, differing mainly in fidelity and governance.

- A020's [[concept-minimum-viable-ai]] applies the MVP directly ([[prereq-lean-startup-methodology]], [[action-incremental-ai-rollout]]) — small, low-risk use cases first.
- A098's [[concept-build-to-learn]] and [[framework-half-day-prototyping]] compress it to a 3-hour, 4–6-person workshop ([[action-run-half-day-prototype]]) with existing tools — no infrastructure ([[contrarian-no-complex-infrastructure]]).
- A095's [[concept-controlled-experimentation-ai]] adds statistical rigor — A/B testing, co-pilot vs. solo modalities ([[action-run-ai-experiments]]) — and funds it via [[action-fund-innovation-stage-gates]].
- A061's [[concept-ai-learning-journeys]] adds governance — experiments test technical/enterprise/human viability and exit through [[concept-red-team-scrutiny]] ([[contrarian-learning-vs-validation]]: a POC that fails human desirability is a *successful* learning journey).

The shared reframe is that experiments are for *learning*, not *validating*. The unresolved axis is fidelity-vs-governance: A098's 3-hour demo is exhilarating but, as its own enrichment warns, is not scaled transformation — the bottlenecks are integration, security, and change management. A061 supplies exactly that missing governance scaffold ([[framework-four-portfolio-stages]]). Read together, they form a two-track model: fast prototypes on top of patient platform/data/governance investment ([[cd-roi-is-the-wrong-lens]]).