SpanForge Pilot Risk Index™
A five-failure-mode taxonomy mapped to each gate in the SpanForge Exit Gate System™. Each gate is designed to prevent one specific failure mode drawn from the governance, data, and lifecycle gaps consistently identified across S&P Global Market Intelligence, Gartner, and McKinsey research.
One failure mode. One gate. Five gates total.
The SpanForge Pilot Risk Index™ is a five-failure-mode taxonomy. Each failure mode corresponds to one gate in the SpanForge Exit Gate System™. The design logic is that each gate, if applied correctly, prevents a specific and well-documented failure pattern from surviving to the next stage.
The five failure modes are drawn from the governance, data, and lifecycle gaps consistently identified in S&P Global Market Intelligence (2025), Gartner (2024, 2025), and McKinsey & Company (2025). They are not theoretical — they represent the actual conditions under which the research shows enterprise AI pilots are abandoned.
“The SpanForge Pilot Risk Index™ is a five-failure-mode taxonomy mapped to each gate. Designed to address the governance and lifecycle gaps identified in the sources cited in this paper.”
What each gate prevents
Each row represents a gate in the Exit Gate System™, the failure mode that gate is designed to prevent, and the mechanism by which it prevents it.
| Gate | Failure Mode Prevented | How the Gate Prevents It |
|---|---|---|
| Gate 1 — Scoping | Wrong problem defined | Forces a written, signed problem statement before any technical work begins. An initiative cannot advance to Proof of Concept without a documented problem definition, confirmed data access, and a sponsor commitment letter. |
| Gate 2 — PoC | Technical infeasibility undetected | Requires demonstrated accuracy on representative data before pilot investment. The gate ensures teams are not advancing on the basis of synthetic or curated data that does not reflect production conditions. |
| Gate 3 — Pilot | No measurable business value | Mandates KPI validation in a real operating environment, not a sandbox. Business value must be evidenced in production conditions, with KPI targets achieved and operational integration confirmed. |
| Gate 4 — Limited Release | No user adoption | Requires measured adoption rates and SLA compliance from actual users before full deployment. Validation occurs with real users, not internal testers, and a defined support model must be in place. |
| Gate 5 — Production | No scalability or governance | Confirms runbooks, monitoring, and owner accountability before removing programme controls. The system cannot move to unmonitored operation at scale without a live monitoring configuration, an approved runbook, and documented ownership. |
Failure mode taxonomy based on patterns identified across S&P Global Market Intelligence (2025), Gartner (2024, 2025), and McKinsey & Company (2025). Gate-to-failure-mode mapping: SpanForge analysis.
Each gate in detail
Gate 1 — Scoping — prevents: Wrong problem defined
Forces a written, signed problem statement before any technical work begins. An initiative cannot advance to Proof of Concept without a documented problem definition, confirmed data access, and a sponsor commitment letter.
Gate 2 — PoC — prevents: Technical infeasibility undetected
Requires demonstrated accuracy on representative data before pilot investment. The gate ensures teams are not advancing on the basis of synthetic or curated data that does not reflect production conditions.
Gate 3 — Pilot — prevents: No measurable business value
Mandates KPI validation in a real operating environment, not a sandbox. Business value must be evidenced in production conditions, with KPI targets achieved and operational integration confirmed.
Gate 4 — Limited Release — prevents: No user adoption
Requires measured adoption rates and SLA compliance from actual users before full deployment. Validation occurs with real users, not internal testers, and a defined support model must be in place.
Gate 5 — Production — prevents: No scalability or governance
Confirms runbooks, monitoring, and owner accountability before removing programme controls. The system cannot move to unmonitored operation at scale without a live monitoring configuration, an approved runbook, and documented ownership.
Used alongside the Gate Readiness Score™
In the SpanForge 4-Week AI Portfolio Diagnostic, a Gate Readiness Score™ and a Pilot Risk Index™ assessment are produced for every active initiative in the portfolio. The Gate Readiness Score™ identifies how close each initiative is to meeting the evidence threshold for its current gate; the Pilot Risk Index™ identifies which failure mode the initiative is most exposed to if it advances without meeting that threshold.
Together, the two instruments give programme leadership and the Gate Authority a structured picture of where each initiative stands and what specifically will happen if it advances prematurely.
Grounded in published research
The five failure modes in the SpanForge Pilot Risk Index™ are not hypothetical. They are derived from the governance, data, and lifecycle gaps that three major research surveys identify as the dominant causes of enterprise AI pilot failure.
Gartner (2024) found that fewer than one in five pilot-phase projects had written, agreed success criteria before development commenced — the condition Gate 1 directly addresses. Gartner (2025) found that 63% of organisations do not have or are unsure they have the right data management practices to support AI — the condition Gate 2 addresses. The S&P Global data showing 46% of proof-of-concepts are scrapped before production reflects the combined effect of the failure modes Gates 3, 4, and 5 are designed to prevent.
S&P Global Market Intelligence. (2025). Voice of the Enterprise: AI & Machine Learning, Use Cases 2025.
Gartner, Inc. (2024). Gartner Predicts 30% of Generative AI Projects Will Be Abandoned After Proof of Concept By End of 2025.
Gartner, Inc. (2025). Lack of AI-Ready Data Puts AI Projects at Risk.
McKinsey & Company. (2025). The State of AI in 2025.