Choose counts, categories, and the right test
Use a contingency table when each observation belongs to one category in each of two variables—for example, preferred plan by customer segment. Enter whole-number observed counts, with each person contributing to one cell.
The test compares observed counts with those expected if the variables were independent. A small p-value provides evidence against that independence model under the assumptions; it does not establish causation or show the size of the association.
Use a test for means when the outcome is a continuous measurement and that is the question you need to answer. A chi-square test of independence asks about a pattern of categories, not a difference in average time, revenue, or score.