A Reproducible Data-Quality and Exception-Management Framework for Resource-Constrained Organizations

Resource-constrained small and medium-sized enterprises (SMEs) and nonprofit organizations require reliable operational reporting but often lack dedicated data-quality teams, enterprise observability platforms and continuous-audit capacity. This study develops and validates a Reproducible Data-Quality and Exception-Management Framework (RDEMF) that converts a small control library into a governed sequence of detection, risk scoring, root-cause coding, ownership, remediation and closure verification. The empirical demonstration uses the Brazilian E-Commerce Public Dataset by Olist, distributed through Kaggle, comprising 99,441 orders, 112,650 order lines, 103,886 payments, 32,951 products, 99,441 customers, 3,095 sellers and 99,224 reviews. Twelve order-level controls and complementary table-level tests assess completeness, uniqueness, validity, referential integrity, temporal consistency, reconciliation, timeliness and robust statistical anomalies. Overall, 21,715 orders (21.84%) triggered at least one rule; 6,530 (6.57%) accumulated a risk score of four or more. Late delivery affected 7,826 delivered orders (8.11%); 1,359 orders (1.37%) recorded carrier hand-off before approval; 775 orders (0.78%) had no item record; and 381 (0.38%) had an absolute payment-to-item reconciliation difference above R$0.01. Robust outlier rules flagged unusual values or durations but were treated as review candidates rather than errors. A transparent queue simulation, based on stated service-time assumptions rather than observed case handling, reduced mean completion time for high-severity exceptions from 5,275.0 to 381.4 hours under risk-priority sequencing, while total workload remained unchanged. The study contributes an auditable rule schema, entity-by-dimension and co-occurrence heat maps, an exception register, a scoring model, a minimum viable dashboard and a closure playbook. Findings demonstrate that low-cost controls can reveal concentrated reliability risks, but causal claims about remediation performance require prospective organizational pilots.