Organizational Relationship Memory: Digitizing Seat-Based Relationship Knowledge So It Survives Personnel Change

In most companies, knowing whom the firm deals with at partner institutions, what post each contact holds, and what the two sides have done together is something individual employees carry in their heads. That knowledge leaves when staff retire or when officials on the other side are reassigned. We propose Organizational Relationship Memory (ORM), a reference data model that treats this knowledge as an organizational asset and anchors it on the seat, meaning a position inside an institution, rather than on whoever occupies it. Persons, positions, institutions, activities, notes, and internal staff assignments (PIC) are linked with both valid time and recording time, so a rotation is handled by adding records, never by overwriting them. The study follows Design Science Research. We profile the NIS data (796 rows) as a data-quality baseline and then test ORM on synthetic data built around one representative persona (P-01), comparing it with a flat CRM and with a CRM that supports timed links over 30 repetitions. ORM keeps the context of every activity (1.00, against 0.20 and 0.27), returns 0.99 of the required successor-briefing items when either the counterpart official or the PIC changes, and lifts current seat holder accuracy from 0.40 to 0.93. The timed-link CRM already retrieves most of the relationship history (F1 0.84 versus 0.97). Part of ORM's advantage comes from information the baselines do not store, from its controlled seat list, and from how the simulation was designed; the results cover one synthetic scenario, have not been reproduced independently, and still have to be confirmed on real data. The work bears on SDG 9 (target 9.5) as an upgrade to organizational information infrastructure and on SDG 16 (target 16.6) as a technical basis for traceable institutional memory; neither outcome was measured, and the SDG 16 link depends on governance controls not yet designed.