Concept Taxonomy¶
This taxonomy organizes the 561 concepts in the Measuring GenAI ROI learning graph into 13 categories. Categories follow the three natural clusters in the course description — generative AI & token mechanics, financial/lifecycle cost analysis, and executive reporting — while keeping any single category under 30% of the total (the largest, EFFIC, is 8.9%).
| TaxonomyID | Category Name | Concept Count | % of Total |
|---|---|---|---|
| FOUND | Generative AI & LLM Foundations | 45 | 8.0% |
| TOKEN | Tokenization & Token Economics | 46 | 8.2% |
| INFRA | Compute, Infrastructure & Pricing | 40 | 7.1% |
| EFFIC | Token & Cost Efficiency Techniques | 50 | 8.9% |
| FINRO | Financial & ROI Fundamentals | 45 | 8.0% |
| METRC | ROI Metrics & KPIs | 40 | 7.1% |
| DATOP | Data Collection, Instrumentation & Observability | 40 | 7.1% |
| SDLC | Software Development Lifecycle Cost Optimization | 45 | 8.0% |
| DSLC | Data Science Lifecycle Cost Optimization | 45 | 8.0% |
| VNDR | Vendor, Model & Build-vs-Buy Economics | 40 | 7.1% |
| RISK | Risk, Governance & Hidden Costs | 40 | 7.1% |
| COMM | Executive Communication & Financial Literacy | 40 | 7.1% |
| RPRT | Report Design, Visualization & Capstone | 45 | 8.0% |
Category Descriptions¶
FOUND — Generative AI & LLM Foundations¶
The baseline vocabulary of generative AI systems: what an LLM is, how it is trained and run, and the deployment concepts (context window, latency, quantization, right-sizing) that later cost discussions build on. This is the entry point of the graph — it assumes only general data science knowledge.
TOKEN — Tokenization & Token Economics¶
How text becomes tokens, how providers count and price tokens, and the vocabulary needed to read a usage bill: input/output pricing, context-window cost, caching discounts, and cross-provider token comparison.
INFRA — Compute, Infrastructure & Pricing¶
The compute and infrastructure layer beneath token pricing: GPU/CPU inference, cloud pricing models, autoscaling, storage/network cost, and the monitoring and alerting needed to see infrastructure spend.
EFFIC — Token & Cost Efficiency Techniques¶
Concrete techniques for lowering token and compute cost: prompt optimization, caching, model cascading/routing, retrieval tuning, quantization/distillation, and the benchmarking discipline needed to prove a technique actually saved money.
FINRO — Financial & ROI Fundamentals¶
The financial vocabulary a technologist needs before writing an ROI report: ROI, TCO, NPV, payback period, cost allocation, budgeting, and business-case fundamentals — taught from first principles with no accounting background assumed.
METRC — ROI Metrics & KPIs¶
The metrics that connect technical activity to financial outcomes: adoption, quality, productivity, and cost-per-outcome metrics, plus the measurement discipline (baselines, control groups, statistical significance) needed to trust them.
DATOP — Data Collection, Instrumentation & Observability¶
How to actually capture the usage and cost data an ROI analysis depends on: telemetry, cost tagging, observability platforms, dashboards, and data governance for metrics.
SDLC — Software Development Lifecycle Cost Optimization¶
Where generative AI changes the cost of building software — requirements through maintenance — including AI-assisted coding, code review, testing, and the risks (technical debt, security review of generated code) that accompany those savings.
DSLC — Data Science Lifecycle Cost Optimization¶
The parallel cost story for the data science lifecycle: data collection and labeling, feature engineering, training, evaluation, deployment, monitoring, and retraining — including MLOps and reuse strategies that reduce cost.
VNDR — Vendor, Model & Build-vs-Buy Economics¶
The economics of choosing among models and providers: build-vs-buy, open-source-vs-proprietary, licensing structures, contract risk, and vendor-selection methods (benchmarking, pilots, scorecards).
RISK — Risk, Governance & Hidden Costs¶
Costs that do not show up on a token invoice: governance overhead, compliance, security, bias/fairness risk, skill erosion, and the risk-adjusted TCO framing needed to represent them honestly in a report.
COMM — Executive Communication & Financial Literacy¶
The audience-facing skills that connect technical findings to a finance-background executive: message framing, storytelling with data, executive summary writing, and handling uncertainty and objections.
RPRT — Report Design, Visualization & Capstone¶
The step-by-step mechanics of assembling the final deliverable: report structure, chart selection for financial data, review/fact-checking discipline, and the capstone sequence that produces a complete, executive-ready GenAI ROI report.