Financial statement fraud can impair investor confidence, distort capital allocation and weaken the credibility of corporate reporting. Existing detection approaches use financial-ratio screens, fraud-risk theories, governance indicators and increasingly sophisticated analytical models, but these approaches are often applied separately. This paper develops a risk-based Input–Process–Output (IPO) framework that connects observable financial and governance signals to a structured review process and measurable investigation outcomes. The proposed framework treats financial pressure, unusual accounting movements, control weaknesses and governance signals as inputs; analytical screening, corroboration, professional scepticism and escalation as processes; and prioritised risk alerts, documented evidence and improved monitoring as outputs. A transparent illustrative logistic model is specified to demonstrate how risk indicators could be combined without presenting simulated values as empirical evidence. The paper synthesises established work on earnings manipulation, fraud-risk assessment and fraud-detection model design. It argues that ratio-based scores should be used as screening indicators rather than proof of fraud, and that model performance must be validated against independently established outcomes, class imbalance and sector-specific conditions. The contribution is a practical, auditable framework for connecting detection signals to investigation decisions while preserving human oversight. The paper is conceptual and does not claim original empirical validation.