Entity-to-Brand association measures brand salience in ungrounded memory via open-ended lists, whereas relevance measures the model's explicit willingness to recommend the brand for a specific entity through repeated binary checks.