Well, typically that is done by addressing the variance in your population. If you have a population in which there is wide variation and in which you could get a large number of different responses, you need a larger sample size/more surveys. If your population is largely the same, a smaller sample size/fewer survey will get the job done statistically. Yes, it is a "cart before the horse" thing. They want you to know your population variance to choose your sample size but you don't know the variance without a sample of the population.
In the real world, there are also political concerns. A sample/number of surveys that is perfectly acceptable statistically in terms of producing data sufficient to characterize the population may not be acceptable politically to those receiving the data. As an example, in a population of 10,000 with almost no variance 100 surveys may be enough to tell you everything you need to know very accurately, but those receiving the data may not be satisfied that only 100 out of 10,000 things were examined.
We don't really know what you are surveying/sampling or anything about your context so it is hard to give specific advice. In general, sample sizes can be too small, but they can't really be too big (they are just more work).
