RFM scoring and X-means clustering of citizen transactions in Saudi E-government through public value theory

Abstract Behavioral segmentation has moved from commercial customer relationship management (CRM) into e-government analytics with little scrutiny of what the move costs analytically. The monetary dimension of recency-frequency-monetary (RFM) scoring rests on a specific premise: that paying more signals being worth more. In commercial settings this is workable. In public administration it is harder to sustain, because fiscal contribution and public value do not stand in the same relation as revenue and customer worth. We develop this argument using a de-identified dataset of 5,244 citizen-level transaction profiles drawn from Saudi e-government service interactions associated with Vision 2030, pairing RFM scoring with X-Means clustering. The monetary term is defined as citizen-to-state fiscal contribution, with welfare disbursements excluded. Four segments emerge from the analysis, separating mainly by fiscal contribution and recency rather than by demographic attributes, and a decision tree validation suggests that the behavioral variables carry most of the discriminatory signal. Read through Public Value Theory, the segments show why CRM analytics cannot be ported directly: the statistical structure that ranks customers by revenue produces an analogous ranking of citizens by fiscal engagement, but the two rankings carry different normative weight. The contribution is conceptual and empirical: the study reframes the monetary dimension of RFM for public-sector use and shows, using de-identified transactional data, how citizen segments require interpretation through public value rather than customer-value logic. When read together, the findings reveal a group that an income-based ranking system would reduce to one low-value category: citizens who are both low fiscal-contribution and high-need recipients, a combination relevant to equitable digital-government designs.