An empirically derived activity-based model that characterizes information seeking as a set of recurrent tactics or behaviours—commonly enumerated as starting, chaining, browsing, differentiating, monitoring, extracting and ending—which users combine and repeat nonlinearly to locate, evaluate and acquire information.

Principle

Information seeking consists of observable, reusable tactics rather than a single temporal progression; supporting or measuring these tactics improves usability and captures search practice more directly than stage-only descriptions.

Demonstration

Illustrative scenario: A researcher begins with a known article (Starting), follows its references to related work (Chaining), browses journal tables of contents (Browsing), filters sources by relevance (Differentiating), sets alerts for new papers (Monitoring), downloads key texts (Extracting) and concludes the search (Ending). The researcher fluidly alternates tactics based on yields and time available.

Misapplication

Reducing the model to a checklist of features to add to an interface without observing user combinations and sequences. The semantic error is treating tactics as isolated UI components rather than interactive practices shaped by task and context.

Consequence

Correct use focuses design and evaluation on enabling common tactics (e.g., chaining-friendly citation links, monitoring feeds); misuse can lead to fragmentary interfaces that provide features in isolation or ignore affective/contextual constraints that shape tactic choice.

Reversal

Where affective states or internal cognitive processes are primary drivers (e.g., uncertainty shaping exploration), or when institutional barriers dictate behaviour, an activity-only view may omit critical explanatory factors; likewise, automated retrieval systems can change the prevalence of certain tactics.

Boundary

Clearly within: observed, repeatable search behaviours during complex information tasks. Boundary case: a hybrid situation where users rely on both tactics and prescriptive workflows. Clearly outside: single-query algorithmic retrieval tasks where no repeated tactics are visible.

Semantic Tension

Action/tactic emphasis ↔ Internal-state or contextual models (Kuhlthau, Wilson): Ellis focuses on what users do; other models focus on why users do it. This tension affects whether interventions target observable features or underlying motivations and constraints.

Synthesis

Ellis’s model reframes search design and evaluation around modular, recombinable behaviours: improving discovery requires enabling the tactics people actually use, not only optimizing abstract retrieval metrics.