Each of these methods has its own strengths and weaknesses. They should, of course, be used in situations that make use of their strengths.
Surveys should be used when you are looking for data that cannot simply be observed. For example, you cannot really tell by observation what things your customers like and what they do not like about your store or your advertisements. You cannot tell whether your customers are satisfied with the service that they receive. Surveys are necessary when you are trying to elicit this kind of information.
Observation should be used when you are trying to gain an understanding of what actually happens in your business as opposed to what people say. For example, customers might say that they visit all parts of your amusement park equally, but only observation can tell you where they really go and how much time they spend there. Restaurant customers might say on surveys that they want lower-calorie choices, but observation might tell you that they order the most fattening things on the menu. Observation is necessary when you are trying to elicit more factual information that does not depend on people accurately reporting their actions or their wants.