Surveys are easy to set up and run, but they’re still underused by most companies. With services such as Surveymonkey or Google Forms, they can be low cost or even free.
The first thing to consider is what you’re trying to do. If it’s to find out numbers and proportions, and you can’t promote the survey to thousands of people, forget it. Only a fraction of people seeing the survey will respond, and a couple of dozen responses will almost certainly have no statistical validity. But many companies have extensive mailing lists or very busy website traffic, and do stand a chance of getting the hundreds of responses needed to make sense. For them, a survey can be a very powerful marketing information tool.
Surveys do not have to find out numbers and proportions though. They can ask for ideas or opinions. For this, even a small number of respondents can be worthwhile. You may be surprised how many people are prepared to give you feedback through a form. Again, this can be very useful, so long as you don’t assume a comment represents anything more than that individual’s contribution. If you send out 100 surveys, and get 10 back, of which 2 quite independently say “how about you make red widgets?”, this may be a helpful idea; but it doesn’t necessarily mean there’s significant demand for red widgets.
Most importantly, don’t bore your respondents with any more questions than are absolutely necessary. There’s nothing wrong with a one-question “survey”, which is really a feedback form. Just make sure they’re quite clear how long the survey is, right at the start. As any interface designer knows, people will be prepared to stick with a process for far longer if they know what they’re letting themselves in for.