The best answer for this is that the proper level of measurement here is ordinal. This is because these possible survey responses are given in order, but do not really have set number values associated with them.
Ordinal data are ones that have some natural order (rank order) to them. In this case, good is clearly not as good as better and best is above both of those.
These categories cannot really be deemed to be interval data because there is no exact interval between them. We do not know how much "better" is above "good" and we do not know how much "best" is above "better." So we cannot give these exact number equivalents. Therefore they are not interval data.
I believe the question is asking for identifying the method of measurement being used when respondent in a survey are being asked to describe the attributes of a particular object in relation to some other objects of the same type choosing from one of three alternatives - good, better and best.
The three alternate responses represent quantities or levels that can be compared with each other and what represents more and what represents less can be identified. But the difference between the items cannot be quantified. We can definitely say that the description "best" represents the most desirable or highest level of the trait or characteristics being described. The description 'better" represents higher level than the "good", but lower than "best". In this way "good" represents the lowest level.
From the above discussion it is clear that the survey uses ordinal method of measurement.