Purpose The purpose of this research is to extend the results of previous studies regarding corporate reputation scales and identify new and specific items relevant for studying global corporate reputation from a customer’s point of view. Design/methodology/approach This research was based on the qualitative projective “Album on Line” (AOL) technique. The authors used a sample of 12 French consumers distributed equally between affective and cognitive scenarios. An individual-difference multidimensional scaling approach (INDSCAL) was applied to display the overall semantic space among generated items. Findings The exploratory AOL approach generated 62 items related to both cognitive and affective orientations characterizing online and offline corporate reputation. The results uncovered six semantic clusters for each scenario. All in all, seven new items could be added in the process of building a new global corporate reputation measurement scale by adding: avant-garde, singularity, exclusivity, savings, return policy, freeness and speed. Research limitations/implications This research makes it possible to propose a new global corporate reputation measurement scale with sound psychometric properties. This scale will be adapted for click and mortars and pure players. This paper unlocks future perspectives by suggesting a causal model that integrates online corporate reputation and its main antecedents and consequences. Practical implications From a managerial perspective, this research offers insights to managers with the main orientations surrounding the components of global corporate reputation. Moreover, the AOL mappings delineate which quadrants the managers would like to be fitted into or avoid, and hence define more precisely which key elements should be stressed or discarded. Originality/value This research outlines AOL, an original qualitative projective technique that can be used to understand customers’ thoughts, which are stocked and collected as images. Moreover, this research intends to analyze the gathered data using both INDSCAL and fuzzy k-means cluster analysis to reduce conventional biases related to subjectivity.