Online surveys are a great tool for gaining insights from your customers or target audience. But how do you know the data you’re receiving is accurate?
Did you survey the right people? Ask the right questions? And did your respondents answer those questions truthfully?
Effective online research starts with the questionnaire design. You likely have a lot of burning questions that you’d love to ask your customers about, but ask too many and you’ll wind up with a survey that could take over 20-30 minutes to complete! Your audience doesn’t want to spend that long taking your survey, and many them will simply quit during the middle of it. Those who do complete it may experience respondent fatigue, where they mentally check-out of the survey and stop providing honest feedback (not intentionally, they just stop thinking carefully about the questions and provide an answer to complete the process). A good rule of thumb is to narrow down your list to only the most important questions and topics, and to keep the survey length to a maximum of 10-12 minutes.
Question order and wording is also very important. Your survey should have a logical flow to it, where you start broad and then drill-down to more specific questions and topics. Conditional logic is a useful tool where you ask certain subsets of your audience specific questions based upon how they answer previous questions. Questions should be worded as simply as possible and avoid the use of jargon or complex/scientific terms. You must also be careful to word your questions in a way that will not lead your respondents toward a particular answer or introduce bias in any way. Remember, these are real people you’re speaking to, not robots. You want to ensure that people completing your survey fully understand and comprehend what you are asking them.
A good online survey isn’t just about asking the right questions; it’s also about talking to the right people. Your sample should be balanced based on the attributes most important to you. Gender and age are the most common demographics to balance on, but ultimately you can balance on whichever attributes are most crucial for your study (e.g. income, geography, etc.). We recommend choosing 2-3 attributes to balance on, as trying to balance on too many attributes can be cost prohibitive and reduce feasibility. Setting hard quota stops is a great way to ensure that no group takes up too much of your sample.
You’ve likely heard the phrase “slow and steady wins the race.” This also applies to marketing research. We recommend starting the data collection slowly (i.e. soft launch) to ensure that the sampling plan is on target and that there are no unforeseen issues with the survey, such as respondents not understanding a question. Once the soft launch data looks good, you can increase the speed of data collection, but still keep it at an appropriate pace to ensure that your quotas don’t fill up too quickly. Otherwise you may wind up in a situation where you quickly fill your quotas for females and older respondents and the final 100 or so completes must be males under the age of 35.
Once you’ve wrapped up the fieldwork, it’s important to check the data for cheaters (e.g. people speeding through the survey or not providing truthful answers) and remove them from the dataset. After that, it's time for the experienced market research analysts to dive into the dataset and work their magic to turn it into meaningful insights for your brand.
Posted by Stewart
Stewart is passionate about all things research. Designing studies, analyzing data, and moderating UX labs, IDIs and focus groups - Stew does it all. When not researching, Stewart chairs the leadership advisory team for the AMA's Professional Chapters Council, and enjoys trying new recipes and exploring the OKC food scene with his fiancee, Jade.