With regard to research that is being carried out for marketing purposes, there are times when surveys are better and times when observation is better. Observation is generally better for gaining objective data while surveys are better for obtaining subjective data.
Let us say, for example, that you want to know how much time people spend at various places in a store that you own. This would be a case in which it would be best to use observation. If you used a survey, the people might not really remember how long they stayed in any one place. Or they might say that they stayed longer in the parts of the store where they thought they should have lingered (like in the produce section rather than the candy section of a grocery store). Observation is good for objective data like this.
But surveys are better for subjective data. For example, if you want to know what people find appealing or unappealing about various parts of your store, observation cannot help. It could tell you where they spend their time, but it could not tell you what their attitudes were towards the various parts of your store or the products stocked in those areas.
Thus, it is best to use observation when you want to get objective data and surveys for in-depth, subjective data.