Enterprise Adoption Ladder
A buyer-side organizational maturity model proposed by wef-ai-in-action-2025 (Accenture, in collaboration with the WEF AI Transformation of Industries community). Distinct from — and complementary to — the substrate-side H1_L0-L7-ladder.
The five phases
| Phase | Name | Marker | Stuck signs |
|---|---|---|---|
| 1 | Initial / ad hoc | AI sandbox experiments; basic data infra; no strategy | Regulatory paralysis, risk aversion, lack of awareness |
| 2 | ”Thousand flowers bloom” | Multiple disconnected pilots; tech-savvy leaders driving MVPs | Use cases disconnected from core business |
| 3 | End-to-end reinvention | Formal AI strategy in one business domain; measurable ROI | Functional silos still intact |
| 4 | Enterprise-level reinvention | Ethical AI board; data products consumable across functions; “enterprise cognitive brain” | Inter-firm collaboration not yet structured |
| 5 | Value-chain reinvention (future vision) | AI extends across partners, suppliers, competitors; “AI as core business enabler” | (no current exemplars) |
How it differs from L0-L7
The two ladders look superficially similar but operate on different axes:
- H1_L0-L7-ladder is a vendor / substrate ownership ladder. It asks: who owns the stack? L0 = human as API; L7 = vertically integrated datacenter + energy + model. The actor is a builder.
- The WEF adoption ladder is a buyer / organizational maturity ladder. It asks: how deeply has this enterprise integrated AI? Phase 1 = experiments; Phase 5 = AI is the operating model. The actor is an adopter.
A firm can sit high on one and low on the other. A traditional bank in Phase 4 of adoption may still operate entirely at L0–L2 of the substrate ladder (rented APIs, no model ownership). A small startup at L6 (open weights on own GPU) may still be in Phase 2 of adoption (one product, no enterprise-wide deployment).
The SME-specific parallel. oecd-sme-ai-adoption-2025 proposes oecd-sme-adopter-taxonomy (Novices / Optimisers / Explorers / Champions) as the SME-specific analogue of the WEF / Accenture 5-phase ladder. Two structural differences:
- The WEF ladder is linear (one phase per firm). The OECD taxonomy is two-axial (complexity × scope), so a firm can be “wide-but-shallow” (Optimiser: many tools across departments, no customization) or “deep-but-narrow” (Explorer: bespoke AI in one niche). This decomposition is richer for the SME population where these two paths diverge — both are real strategies, and the WEF ladder collapses them.
- The WEF ladder is calibrated for large enterprises. Phase 4 (“enterprise cognitive brain”, ethical AI board) describes organisational maturity at scales most SMEs cannot reach. The OECD Champion case studies (Calgary healthcare-tech, Cambridge biotech) are the SME analogue of WEF’s Phase 4–5 but operate at materially different scale, with materially different substrate and finance positions.
What this ladder reveals that L0-L7 hides
- The scaling-gap — most enterprises stall between Phase 2 and Phase 3. L0-L7 has no analogue for this transition; it assumes a stack actor can climb freely if capital is available. The adoption ladder makes visible that capital alone does not move you up; foundational enablers (foundational-enablers) do.
- The organizational vs. technical distinction. Many bear-case arguments (scaling-wall, task-based-framework) operate on the technical side. The 74% gap suggests the bottleneck may also be organizational, which is a different lever set.
- Where the capital-labor-divergence plays out concretely. Phase 4 reorganizations are where roles change, where the chief AI officer appears, where teams reshape.
What this ladder hides that L0-L7 surfaces
- It treats AI as something you adopt, not something you own. The question of who captures surplus is invisible.
- It collapses very different kinds of integration (renting Claude vs. running your own model) into the same maturity step.
- It has no concept of distribution-moat. A Phase 4 firm that runs everything on someone else’s API is not strategically equivalent to a Phase 4 firm that owns the model.
- It is silent on geopolitics (divergent-value-stack-optima) — the same phase looks different in the US / China / EU stacks.
Use in the paper
The two ladders are likely complementary axes in the paper’s framing:
- The substrate-side ladder (H1_L0-L7-ladder) explains where margin pools structurally.
- The adoption ladder explains which enterprises capture how much of that margin.
A firm at L0–L2 (renting capability) in Phase 4 (deeply integrated) is paying surplus upward to its L6–L7 vendors. A firm at L6 (owns model) in Phase 2 (no real deployment) has built a substrate it cannot monetize. The interesting position — and likely the durable one — is high on both ladders.
Data
- scaling-gap-74-16 — the 74% / 16% artifact that motivates the ladder.
- ai-leader-productivity-delta — leader-cohort outperformance numbers.
Related
- wef-ai-in-action-2025 — source.
- H1_L0-L7-ladder — substrate-side parallel.
- scaling-gap — the empirical artifact this ladder visualizes.
- foundational-enablers — what WEF claims moves a firm up phases.
- H2_u-curve-of-value — value-capture question this ladder cannot answer alone.
- distribution-moat — the strategic asset this ladder is blind to.
- capital-labor-divergence — what changes inside the firm at Phase 3–4.
- divergent-value-stack-optima — the bloc-level pressure this ladder ignores.
- taker-shaper-maker — strategic-positioning axis orthogonal to this maturity ladder.
- digital-core — the architectural precondition assumed at each phase.
- fusion-skills — the workforce capability assumed to move firms up phases.
- non-tech-digital-core-synthesis — secondary restatement of this ladder for non-tech firms.
- oecd-sme-ai-adoption-2025 — SME-specific parallel source.
- oecd-sme-adopter-taxonomy — the buyer-side decomposition for SMEs.
- sme-ai-adoption-gap — the data context that motivates an SME-specific taxonomy.