These surveys don't/can't measure customer satisfaction. They suffer from self-selection bias. If only 2% of your visitors actually submit the form, you can onl

These surveys don't/can't measure customer satisfaction. They suffer from self-selection bias. If only 2% of your visitors actually submit the form, you can only make generalizations about those 2% of your visitors. You cannot extrapolate your biased sample to the population (all your users). All you are hearing about is the vocal minority. These can however be useful for spotting objective problems. e.g. a broken link or missing image.