As tasting is a dynamic process, temporal data are collected simultaneously with tasting. Indeed, most newly reported studies involving consumers have been conducted using the Temporal Dominance of Sensations (TDS) or Temporal Check That Apply methods. Concurrent data collection reduces potential bias such as memory, but it attaches great importance to the moment at which the sensations are cited. Thus, it results in measurement of durations possibly made imprecise due to heterogeneity in consumers’ behaviour, which could affect conclusions. A new retrospective method inspired from Temporal Order of Sensations, Pick 3 and Rank and the 3 phases of wine evaluation is introduced in this article. Based on a concept close to dominance, the Attack-Evolution-Finish (AEF) method allows consumers to select one descriptor each at the beginning, at the middle and at the end of the tasting. The results obtained with two different panels carrying out both the AEF and TDS tasks on dark chocolates are compared. The conclusions are very similar in terms of product discrimination. The retrospective task removes the consumers’ heterogeneity in terms of the number of citations, delays and durations and thus requires no data transformation before analyses. In view of these results, rather than promoting a new method, the article raises questions about the level of detail to look at in temporal sensory data.