These FOUR steps will help you understand how to conduct a survey:
Ask yourself why. The first thing you should do before writing a survey is to figure out why you’re sending it out at all. What’s the point you’re trying to prove, or the question you’re trying to answer? This will change the questions and question types you need to ask in your survey.
For example, let’s say you want to know what people think about milk. The first thing you need to figure out is: why do you want to know what people think about milk? Is it because you are a dairy farmer who wants to know how many new cows to buy this year? Or is it because you are a drug company gauging the demand for a medicine for lactose intolerant people? Or is because you want to market your new soymilk most effectively?
Pinpoint your demographic. To pick the best way to design and distribute your survey, you should figure out who you’ll be asking to fill it out. Decide who your target demographic is before you make decisions about your survey. Have more than one demographic? You might want more than one survey.
For example, as a toy company, the kinds of questions that you pose (and the way that you phrase them) to 7-year-old boys should be different from the ones you ask 40-year-old women—and the way you get kids to fill out a survey is different from the way you would get moms to do so. If you want to know the opinions of 7-year old boys AND moms, you’ll want to send out two separate surveys.
Power up. Once you’ve figured out who you’re asking, make sure you have enough of them! Figuring out the number of people you need to have respond to your survey (known as “sample size” in the survey world) will make sure that any analyses that you run on the data you collect are a strong basis for good decisions. This “sample” of people needs to be big enough to be an accurate estimate of your target demographic.
For example, if you’re asking American men how often they wear ties, enough American men need to respond so that their answers represent what American men think. So how many is enough? Well there are roughly 150 million men in the United States, so to estimate a population of that size you’ll need 385 American men. Not sure how big your population is? In general, we recommend a default sample size of 400 people. (If you want to know more about the mathematical magic behind these numbers, check out these tips on finding your sample size.)
Pick your moment. When you choose to send out your survey can have a big impact on who ends up responding. This can cost you money, as incorrect survey launch timing can decrease the percentage of people who respond to your survey. Survey launch timing can also change the answers you get—which can change the decisions you make. Not sure exactly when the right people might be around? Collecting survey responses for at least a week can make sure you’re pulling in a broad range of people and opinions.
For example, surveys about designer dress rental services will likely yield a bigger response in the days surrounding the Oscars than other times of the year. Or, if you’re trying to reach 18-21 year olds who like to go out to nightclubs, sending a survey out on a Friday night is a bad idea as they’re likely to be out. Surveys can also influence the answers you receive. Identical surveys sent out at 9 AM and 9 PM about how tasty people think dumplings are is likely not only to be answered by different people, but also is likely to yield different answers. Similarly, a survey about how much someone wants to quit their job will yield different answers on a Monday and a Friday.