A key marker of a quality survey, whether that survey is an online form or another mode, is replicability of your results over trials. Replicability of your survey results lends credibility to your organization's research work.
One indication that your survey form may have results based on biased samples is nonreplicable results for the same instrument, fielded with the same survey design, from one time to the next. If the same sample frame produces significantly different results for identical questions, that should perk up our noses to the scent of possibly unreliable results. We should start to look at the criteria used to select our survey sample, and take a fresh look at the questionnaire to see if it is still pertinent to our needs, or the needs of our clients.
This holds true for all modes of survey research, including online survey internet research. Receiving divergent findings in survey results being fielded again and again could lead to differing (and possibly harmful) business decisions. No company should be left with uncertainty about answers and actions when decisions need to be made based upon the research. This is just one reason why replicability of your research findings is a key ingredient to sound data. When in doubt- replicate!