Interesting, a questionnaire with 4 questions might be a naive methodology. I would design the experiments so that one customer only answers one type of questi

Interesting, a questionnaire with 4 questions might be a naive methodology. I would design the experiments so that one customer only answers one type of question, and they wouldn't even know they are being part of the experiment. For example: yesterday I bought a coffee, and the price seemed too high, I asked the cashier if they were sure, that the price wasn't right, and that I would like to cancel the purchase. It turns out she had wrongly understood I wanted 2 coffees, but if this is an experiment you can replicate, and train staff to do this once or twice per day, you've got a distributed and blind trial that doesn't rely on predicitons of subjects of their own perceptions.