OECD SME Adopter Taxonomy
A four-quadrant typology of SME AI adopters proposed by oecd-sme-ai-adoption-2025. Crossed on two axes — complexity of AI use and scope of AI application — and shaded by a third dimension, digital maturity, that runs diagonally across both. It is the SME-specific analogue of the large-firm enterprise-adoption-ladder.
The two axes
Complexity of AI use (x-axis, five steps from low to high):
| Level | Marker | Example |
|---|---|---|
| None | No AI in any capacity | Family logistics firm on paper-based systems |
| Embedded | Built-in AI features in everyday digital tools, often unnoticed | Word grammar correction; email spam filter |
| Off-the-shelf | Public AI tools (ChatGPT, Gemini, Midjourney) used directly | Tourism SME drafting multilingual marketing copy with Llama |
| Customised | Models trained on proprietary data or fine-tuned for sector | B2B retrieval / generative stack on Cohere; clinical-triage fine-tune |
| Frontier | State-of-the-art systems in business-critical workflows; multimodal / agentic | Consulting firm with multimodal pipelines; agents autonomously managing ad campaigns |
Scope of AI application (y-axis, five steps from narrow to broad):
| Level | Marker | Example |
|---|---|---|
| None | No strategic or operational use | Pharmacy on paper records |
| Isolated | Single use case by one team; PoC | Chatbot for FAQs, no integration |
| Functional | Multiple discrete tasks in a small number of functions | Marketing agency drafting content + visuals, no co-ordinated workflow |
| Cross-functional | Multiple tools across most departments, with shared data + governance | Online retailer running Claude, Midjourney, Shopify-integrated analytics |
| Enterprise-wide | AI embedded in virtually all functions, driving strategy + operations | Insurer with AI across underwriting, claims, retention, pricing |
Digital maturity is the diagonal shading. It is both a precondition for moving up either axis and a consequence of doing so — successful AI adoption raises a firm’s digital maturity, which in turn enables the next step.
The four quadrants
| Quadrant | Complexity | Scope | What it looks like |
|---|---|---|---|
| AI Novices | Embedded / Off-the-shelf | Isolated / Functional | Off-the-shelf LLMs in single workflows; experimental. Built on prior digital-tool experience. |
| AI Optimisers | Off-the-shelf | Cross-functional | Wide variety of off-the-shelf tools across departments; growing co-ordination but no custom models. |
| AI Explorers | Customised / Frontier | Isolated / Functional | Bespoke models or agentic workflows in narrow niches; data-intensive sectors (manufacturing, ICT) dominate. |
| AI Champions | Customised / Frontier | Cross-functional / Enterprise-wide | AI embedded across operations and strategy; advanced LLMs, NLP, computer vision underpinning both product and internal processes. |
Two implicit corners sit outside the taxonomy:
- Non-users — bottom-left, no AI at all. Need awareness-and-information policy, not adoption policy.
- AI-first firms — top-right, AI is the product. Drive innovation, not the policy focus of the SME blueprint.
The policy-relevant population sits in the broad middle of the matrix.
How it relates to the other axes in this wiki
| Axis | What it measures | What it answers |
|---|---|---|
| H1_L0-L7-ladder | Substrate ownership | Who owns the stack? |
| autonomy-slider | Human↔agent control | How much does the agent decide? |
| enterprise-adoption-ladder | Org maturity (large firms) | How deeply has the firm absorbed AI? |
| oecd-sme-adopter-taxonomy | Complexity × Scope (SME-specific) | What is the firm doing with AI, and how broadly? |
| taker-shaper-maker | Position relative to the model | Renting / customizing / training? |
The OECD SME taxonomy is not a fourth orthogonal axis — it is closer to a two-axis decomposition of the same maturity dimension that the WEF/Accenture five-phase ladder collapses into one number. It is structurally richer for the SME population because it distinguishes “wide-but-shallow” (Optimisers) from “deep-but-narrow” (Explorers) — a distinction the linear ladder cannot make.
Mapping to taker-shaper-maker
The mapping is approximate but instructive:
- Novices and Optimisers → Takers. They rent off-the-shelf models; no proprietary customization.
- Explorers → Shapers. They customize models on proprietary data; the model itself is still rented.
- Champions → Shapers or (rarely) Makers. Most stay in the Shaper layer; only a small fraction (often deeply technical, often AI-first) cross into Maker.
Read with taker-shaper-maker: the OECD taxonomy names the population that the WEF typology strategizes for. Most SMEs are Novices/Optimisers — i.e., Takers — and the bull-managerial frame’s pathway is to lift them into Shapers / Optimisers-with-customization. That pathway is exactly the consultancy commercial interest taker-shaper-maker flags.
What this taxonomy adds that L0-L7 misses
- The SME-specific failure mode of “wide but shallow” (Optimisers): comfortable with many off-the-shelf tools across departments, but never crossing into custom or core-business use. Most generative-AI-using SMEs sit here; only 29% report core-activity use.
- The barrier between Functional and Cross-functional scope — likely the SME analogue of the scaling-gap (74% stuck between Phase 2 and Phase 3 on the enterprise ladder). Pre-Cross-functional, AI is decoration; post-Cross-functional, it is operational.
- The Explorer corner as a distinct strategic position: deep customization in narrow niches. Often missing from the WEF/Accenture framing, which assumes maturity = breadth.
What it hides
- No vendor / substrate axis. A Champion running everything on Anthropic’s API and a Champion running its own fine-tuned weights both appear in the same quadrant. The H1_L0-L7-ladder is the complement.
- No surplus-capture axis. The taxonomy describes adoption posture, not who captures value. A Champion paying $1M/month to OpenAI is a different economic actor from a Champion that has built proprietary data assets it monetizes. H2_u-curve-of-value is the complement.
- No distribution / market axis. A B2C Optimiser with millions of users and a B2B Champion with five clients look comparable in the matrix and incomparable in cash flow.
Use in the paper
The taxonomy is the buyer-side substrate of the white paper’s argument. Section 4 of the OECD source ties each quadrant to its dominant barrier (Novices need awareness + skills; Optimisers need data integration; Explorers need compute + talent; Champions need finance + scale). For our paper, the taxonomy is the natural decomposition of the SME population that the U-curve must address — it forces the paper to be specific about which SME population it is talking about when it asks where value lands.
Related
- oecd-sme-ai-adoption-2025 — primary source.
- sme-ai-adoption-gap — the empirical gap this taxonomy structures.
- enterprise-adoption-ladder — large-firm 5-phase parallel.
- H1_L0-L7-ladder — substrate-ownership axis the taxonomy is silent on.
- taker-shaper-maker — strategic-positioning axis the taxonomy decomposes for SMEs.
- H2_u-curve-of-value — surplus-capture question the taxonomy cannot answer alone.
- scaling-gap — the intra-firm pilot-to-scale gap; the OECD taxonomy makes visible an SME analogue.
- foundational-enablers — the four enablers tied to each quadrant.
- diffusion-vs-innovation — the SME taxonomy charts the literal substrate of the diffusion S-curve.
- autonomy-slider — orthogonal axis on human↔agent control.
- digital-core — the architectural precondition implicit in “digital maturity”.
- sme-novice-off-the-shelf-llm, sme-optimiser-cross-functional-stack, sme-explorer-custom-agent, sme-champion-vertical-ai — illustrative case-pattern pages drawn from the OECD case studies.