Abstract
Digital trace data - passively generated records of human activity captured through sensors, logs, and tracking systems - is increasingly available to management researchers, yet the field lacks a principled framework for using such data to develop behavioural theory rather than merely describe or predict observable patterns. This paper proposes the Digital Trace Theorising (DTT) Framework: a four-phase critical realist methodology for moving from empirical trace data to theoretical propositions about the generative mechanisms underlying observed behaviour. The framework operationalises Bhaskar’s ontological strata (empirical, actual, real) through: (1) Ontological Anchoring, (2) Theoretical Code Construction, (3) Multi-Level Empirical Validation, and (4) Retroductive Mechanism Inference. A core methodological device - ontological granularity tuning - enables researchers to identify demi-regularities (Lawson, 1997): patterns that persist across micro, meso, and macro abstraction levels and that thereby provide empirical warrant for retroductive claims about underlying mechanisms. The framework is demonstrated through an illustrative case: computational sequence analysis of 102,726 movement-tracking records from 2,300 visitors to a professional exhibition. The case produces four theoretical propositions about exhibition visitor behaviour and, critically, shows how the productive rejection of theoretically predicted codes constitutes a methodological strength rather than a failure. The DTT Framework is presented as a generalisable methodology applicable across a range of digital trace contexts in management research, including workplace collaboration, organisational routines, e-commerce customer journeys, and online platform behaviour.