Have a friend or colleague sanity check it. Before you go live, have someone take the survey as if they were an actual research participant. Do they find it onerous? Too full of jargon? In anyway off-putting? Even a short survey can turn people off—if it contains obscure words, has an odd question sequence, or in any way comes off as unprofessional.
Evaluate the questions for audience fit. Do you need to add any screening questions to make sure they are qualified? For example, if you plan to target working parents for a survey about restaurant behaviors, you may need a screening question that confirms a minimal level of household income.
Think about how you will want to analyze the results when data collection is complete. Using the same example, will you want to see how restaurant behaviors vary based on income level? Number of children? Attitudes about food and exercise? Maybe urban vs suburban location? Once you know how you will want to analyze your data, make sure you have included the right items in the questionnaire. I have seen it happen: someone does a project, collects all the data, and then realizes they forgot to ask something critical for the analysis.
Double check for “bad” survey questions. Questions that are too long, vague or leading. Below are a few examples. In the first example, the question is just too vague. My idea of “sometimes” may be once a week. Yours might be 3 times a week. How could you use this data? Not very well.
The second and third examples are just boring. And onerous, which means you may get a lot of qualified people dropping out of the survey because it was just too painful for them. You can get away with 1 or 2 boring questions in a questionnaire, but don't push your luck.