When you want to know what your customers think about your product or brand, there's no better way to find out than to ask them directly. Survey results can give you valuable information for directing virtually anything in your business, from product development to customer service to marketing endeavors. Conducting an effective survey and interpreting the data without bias requires some planning and education.
First, you'll need to decide what type of survey research is appropriate and how to distribute the survey in order to reach your target audience. Once you have the data and have accounted for the limitations of surveys and some common biases, you can perform a survey analysis and develop a hypothesis based on your interpretation of the data.
Next, have your data and hypothesis reviewed by a coworker to catch any flaws you may have overlooked. The final step is to convince the decision makers in your company that your theory is supported (or not) by the survey results. Visuals and charts can help you display the data and your findings in an easy-to-understand way. Although there is a learning curve, the more experience you gain conducting questionnaires or interviews and interpreting survey results, the faster and easier this entire process will seem.
Types of Survey Research
Surveys can be conducted as either a questionnaire or an interview. With a questionnaire, survey respondents record their own answers to a set series of questions, whereas an interviewer records the survey responses in an interview. An interview also allows for improvised questions to gather additional information, whereas a questionnaire is static.
Surveys can also be differentiated by delivery method. Phone interviews, face-to-face personal interviews, focus groups, online surveys and written surveys are all possible ways to distribute a survey and ensure accurate data collection. Each one exposes the survey to some potential biases and limitations, however.
Creating and Interpreting Surveys Without Bias
Creating and analyzing a survey with absolutely no bias is practically impossible, but it's important to at least recognize the existence of bias and take steps to lower its influence. For example, one common bias present in survey results is the self-selection bias. This means you only end up collecting survey results from people who chose to participate. The people who did not choose to participate may have a completely different opinion about your brand than the survey respondents, and thus the data will never truly be comprehensive.
A recall bias also affects survey results. You can't ever be sure that respondents accurately recall their brand experiences. The sooner you can send a survey after a customer interacts with your brand, the more likely recall bias will not be an issue. However, if weeks or months pass, the customer is unlikely to remember specific details, like whether any aspect of your online checkout process was confusing or whether he easily found what he was looking for.
If you fail to ask all the important survey questions, you could have omitted variable bias, which skews your data or leads to an improper conclusion. Sometimes, it's an honest mistake, but sometimes, you don't know what you don't know. Having more than one person involved in designing surveys can help diminish this type of bias. Multiple types of biases exist, so it's worth taking the time to research the major possibilities before conducting survey research.
Keep in mind, in statistically-evaluated surveys, no conclusions are considered unbiased or sound unless the sample-size of the data is at least 25 data points.
Evaluating a Hypothesis
The purpose of a survey is generally to test a hypothesis you have with the resulting data. Think of a theory about your product/customers/etc. that you want to test through your survey.
Once you're satisfied with your hypothesis, it's time to pull in a co-worker who can act as a devil's advocate. Ask this person to find all the flaws in your hypothesis. There is no holding back! You want to correct any mistakes before your report is seen by the company's decision makers.
Limitations of Survey Results
One huge limitation of survey-result data relates back to the self-selection bias. A bunch of people will ignore your survey. Perhaps this means they haven't developed loyalty toward your brand and simply don't care about giving you information. You'll never find out exactly why this is because they won't answer your survey.
Make sure you include all data and don't cherry-pick the results to support your hypothesis. This represents a major bias and isn't the intention of research. Your conclusion should always conform to the data, never the other way around.
It's also important to take into account whether or not your survey is easy to understand. If the questions and answers are too complicated, people may misinterpret them. There's also the possibility of respondents giving false answers or leaving survey questions blank, both of which can skew your results.
Presenting Survey Data
Now, you need to make a visual presentation of the survey data so you can walk other people through your logic and give them all the necessary information they'll need to understand your hypothesis. Just as you should not cherry-pick information when creating your hypothesis, don't exclude data that doesn't seem relevant to your hypothesis. Let other people evaluate all the data and look for any logical fallacies or biases.
Decide which visuals make the most sense for your data: bar chart, pie chart, linear graph, data chart, etc. Google Sheets or Microsoft Excel are common tools for data visualization. How you package the survey data will depend on your delivery method and expectations within your company, but PowerPoint is a popular choice for presentations. A formal written report or more robust infographic may also be appropriate.
If you're nervous about giving a presentation, the key is to be as prepared as possible. Practice your presentation out loud with an audience. Your devil's advocate co-worker would be a great choice. No one knows more about this data than you do, so that alone should place you in a position of confidence.
Be Prepared to Answer Questions
Once you give the presentation or distribute the report, be prepared to answer questions. If your logic wasn't clear, you may be asked to clarify how you reached your conclusions. Make sure you understand how the survey was created and distributed, steps you took to avoid biases and the statistical analysis you used when interpreting the data, as all of these factors may come up as questions.
If everyone seems satisfied with your hypothesis, the next question will undoubtedly be, "How do we achieve that?" Do some preliminary research to develop a plan for executing your hypothesis. Include an estimated budget, important developmental steps, key players and other practical details. It also doesn't hurt to have a few alternative options in the back of your mind in case your initial idea isn't received well.