Surveys are effective tools for market research, customer feedback, and monitoring performance. It helps in coming up with important business decisions too. While surveys usually deal with a lot of simple question and answer forms given to respondents, surveying is actually challenging and if not designed correctly, can lead to biases and misleading results.
Designing a survey is as important as carrying it out. Like all processes, it is important to get the first step right.
How to Design a Survey
1. Pre-Survey Process: Define Goals and Outline Predictions
Before creating a survey, you need to identify the goals you wish to achieve first. Whether it’s for market research, feedback or performance monitoring, you need to identify specific goals you’d like to gain from the planned survey. After you identify specific goals, you need to draw some hypotheses (predictions or assumptions) you think will most likely come up from the survey results. Why do you need to draw up hypotheses up front? So you can learn where your intuition is strong and identify organizational blind spots or “weaknesses”.
This phase, also known as the pre-survey process, streamlines and guides you when designing your survey. This phase will also help you in the evaluation and analysis part once the survey results come in.
2. Structuring Survey Questions and Answer Styles
Limiting answers actually helps you define better the limits of your survey analysis afterward. This also helps you design survey questions better.
There are four main responses to data (or survey questions):
- Categorical – “nominal”; unordered labels like colors, brands, occupations, etc.
- Ordinal – scalable; examples are “strongly agree to strongly disagree”, “never to often”
- Interval – ranges; examples are salary ranges, number of employees, etc.
- Ratio – numbers
A good tip would be to think about how you will summarize answers to survey questions in survey results (or reports) as you design the survey. So, do this first – not afterward.
3. Phrasing Survey Questions and Answers
Avoid leading questions.
Avoid “presuppositions”, or leading people to a biased answer. It is hard to cover all the possible answers to a question this is the reason it is recommended to limit answer styles to define survey results easily. But you need to always allow for neutral or NA responses. This might feel like getting a negative response but it can actually improve data quality.
Avoid compound questions.
Don’t ask two questions in a one statement, split it into multiple questions to easily identify sentiment. Keep questions as short and direct as possible.
Use Simple Language. Steer clear of jargon.
Randomize answers. For categorical answers, it is important to randomize arrangements as respondents usually select the first answer because it’s easiest (first seen). Randomization avoids bias. For ordered answers, keep the order consistent (never to often as 1 to 5, etc).
4. Selecting Respondents
Sampling population is important to ensure that the results you get hold true for your target audience. If you can administer a survey in different channels – phone, website, face to face, etc – do so. If this isn’t possible, consider the channel that is more related to the target audience you wish to analyze. For example, if your target audience are repeat customers, it would be better to carry out surveys in stores. Or carry out the survey electronically to customers who have multiple orders on record.
How to Analyze Survey Results
Once a survey is carried out, analyzing results is the harder task at hand. Analysis involves breaking and grouping data in readable format, where decision makers (CEOs, marketers and researchers) can easily interpret results and infer insights.
Analysis can be done by converting data results into tables or graphs, depending on the answer style set. This is also called visualizing data. (This is why it is important to envision how you want to carry out results DURING designing surveys). Data are usually grouped, calculated in percentages and compared with one another based on survey questions. For example, tables are more useful when you are looking at precise numbers and few comparisons.
When interpreting survey results, it is important to focus on the high points – the largest or most unusual differences. If a result makes you skeptical, try to see if it can be replicated by carrying out another survey.
Usually, big strategic business decisions don’t rest solely on a single survey. Multiple surveys are carried out, its data analyzed and visualized, studied and evaluated, before a firm decision is carried out.
Be Smart in Carrying out Your Business Surveys
Designing surveys is not just throwing a bunch of questions together and sending it out to random people. Surveys can help your business but done incorrectly, it can give misleading results that can create gaps in your decision-making.
We at Infinit Datum have a team of expert market research professionals that can help you carry out surveys from start to finish – from creating a solid survey questionnaire, to maximizing response rates, to choosing the right survey format that will best bring out the data that you need. Contact us if you’d like to know more about how we can help.