From numeric scores to letter grades: A systematic review and data-driven ordinal grading framework for supplier performance evaluation in Indonesian government procurement (SIKaP)

Supplier evaluation is the gatekeeping function of public procurement, yet the instruments that support it are methodologically fragmented and, in regulated settings, too coarse to discriminate at the margins that decide outcomes. The root problem addressed here is a structural mismatch: a continuous, compressed 1–3 measurement scale is used to drive categorical, high-stakes decisions — eligibility, monitoring, and blacklisting — without any defensible rule for converting scores into decision categories. In Indonesia’s Supplier Performance Information System (SIKaP), scores cluster near the ceiling, so two providers differing by hundredths of a point (2.70 versus 2.90) cannot be distinguished defensibly. This study integrates a PRISMA-2020 systematic review (57 of 1,184 records) with the design and data-driven demonstration of an ordinal letter-grade framework (A/AB/B/BC/C/D/E) that maps the SIKaP composite onto interpretable, decision-ready categories with explicit cut-offs, grade points, and procurement action rules. Using secondary and web-mined data — official SIKaP weights, the national blacklist aggregate (~5,200 sanctioned providers), and the documented sub-20% assessment-uptake problem — and a demonstration population of 480 providers calibrated to real SIKaP parameters, the study consolidates a criteria–method–scoring taxonomy, operationalises a criterion-referenced grading scheme that resolves the 2.70-vs-2.90 problem, segments providers into four managerial clusters via k-means, and performs sentiment analysis on mined procurement discourse. The review exposes a clear gap — ordinal letter grading is essentially unused in public-procurement supplier evaluation — and the framework fills it with a transparent, reproducible, regulation-compatible alternative augmented by segmentation and sentiment analysis.