<p>This paper reconceptualizes algorithmic systems as institutional actors, arguing that they increasingly perform core functions of governance, norm enforcement, and rule-making across public and private domains. Drawing on and extending institutional theory, the concept of algorithmic institutions is developed to describe sociotechnical systems that classify, regulate, and adjudicate not through human discretion, but through procedural abstraction. To capture how such systems acquire and stabilize institutional authority, a layered analytical framework comprising four interrelated dimensions is introduced: codification, performativity, adaptivity, and opacity. Through three theory-informed diagnostic vignettes—welfare fraud detection in the Netherlands, algorithmic mobility planning in rural Germany, and dynamic pricing in European supermarkets—the framework demonstrates how these dimensions manifest across diverse institutional settings. Despite sectoral differences, each case reveals a convergent logic of institutional transformation: governance by code rather than consensus. Algorithmic institutions are thus conceptualized not merely as technical infrastructures, but as emerging forms of institutional order that reconfigure legitimacy, accountability, and institutional agency. By framing algorithmic design as institutional design, the paper advances a diagnostic vocabulary for analyzing how algorithmic systems not only support organizational processes, but actively reshape the institutional foundations of contemporary governance.</p>