Algorithmic Reputation Management: A Theoretical Framework for the LLM Era

Large language models and AI-mediated search systems act as new gatekeepers of entity authority, deciding which professionals, organisations and institutions are surfaced as legitimate references. Aggarwal and colleagues (2024) introduced Generative Engine Optimisation (GEO) and the GEO-Bench benchmark to measure how content-level tactics raise an item's visibility inside generative responses; GEO treats citation frequency as the dependent variable and does not theorise the prior variable — reputation — that determines whether an entity is recognised at all. This paper introduces Algorithmic Reputation Management (ARM) as a research field rooted in strategic communication and public relations theory, defined as the deliberate stewardship of an entity's authority and credibility as perceived and reproduced by algorithmic mediators. ARM and GEO share an empirical territory — the answer surface of generative AI — but address different objects and rest on different epistemological commitments: they are adjacent fields, not nested ones. A bibliometric mapping of 293 field articles (2022–2026) confirms the boundary empirically: 31% of the field measures optimisation or citation while only 8.5% theorises prior reputation and 6.1% does both. The paper locates ARM within communication theory, draws explicit boundaries with corporate reputation and source credibility, advances testable propositions on construct validity, and sets a research agenda for the field.