The actual thing you’re noting is not a “revealed preference” but a “displayed behaviour in a specific context.” If you’re not AB testing with controls over a broad set, you’re not revealing anything about behaviour. You’re just sating your own confirmation bias.