The growing “strategy-as-practice” (SAP) body of literature has provided a wide range of theoretical developments for doing research on strategy in an organization as “something its members do”, but still leaves the methodological level as one of the five challenges in the SAP field. Analyzing the resulting fine-grained and contextual data is then a main challenge for SAP research, and little work has yet been done on innovative methodology that is specific to the SAP perspective. One solution may be found in working with organizational members, but the analysis of complex data is nevertheless a weakness of these group-based methods. The objective of this paper is therefore to propose a cognitive group-based methodology to represent what practitioners currently do and the combinations of these individual practices in strategic types. Using Kelly’s theory on “personal constructs” as a basis, we propose to codify practices as “constructs” and to represent them on dichotomous axes opposing extreme practices. We then use the repertory grid technique to cross these practice-based axes. When applied to several practitioners, such axes and the resulting repertory grid highlight the different combinations of practices carried out by practitioners. The methodological procedure consists of four stages, mainly iterative, which are illustrated throughout the paper by examples drawn from our agricultural domain. The method is first based on interviews with farmers about their practices, and reports written up after each interview. The group-based analysis of these practice-based reports then aims at formalizing extreme and intermediary individual practices that are then combined in strategies-aspractices using the repertory grid tool, RepGrid. This procedure was first designed within a working group and was subsequently tested in several working groups in the agricultural domain within each of which we have formalized many and various strategies-as-practice in farmers’ typologies. We propose here a cognitive method that can be used to codify and categorize practice-based data in a grounded and rigorous way. Within a knowledge engineering perspective, it aims at enhancing group-based discussions and researcher creativity when analyzing practice-based data. Beyond this first-level analysis, it has also proved to have a strategizing power, since it has been shown to help participants to extend their representations of individual strategy diversity and to draw conclusions that have consequences at higher levels of organization.