mohamed_hassan / Pixabay
The rise of low-cost online survey platforms such as Survey Monkey has led to a commensurate rise in poorly designed surveys. That’s dangerous – it reflects poorly on the brand behind the survey and it can lead to decisions being made using dodgy data. So, here are some tips if you’re planning to design your own survey.
First, pause and reflect on why you’re conducting the survey – what’s your goal and what decisions are you going to make? Write this down in a short statement and constantly refer back to it when designing the survey. This will ensure that your survey gathers all the information you need and doesn’t suffer from scope creep.
Then ask yourself another question – to reach this goal, what information do I need to know? Again, write this down as it will form the skeleton for your survey.
Armed with this skeleton, you then need to write a question (or questions) for each information objective. When doing so, avoid these common pitfalls:
Recommended for You
Webcast, May 8th: Customer Insights That Drive User Engagement
- Write plainly and clearly, avoiding formal language, jargon, and complex sentence structures. This will make the survey more engaging. I find it helps to read each question aloud and if it doesn’t match how I’d speak naturally in conversation, then it needs to be revised
- Make sure your question isn’t leading as this will create a biased response. For example, ask “how do you feel about X?” rather than “how positively do you feel about X?” as the latter implies the answer should be positive
- Don’t ask double-barrelled questions. For example, ask “how satisfied are you with the way in which your query was dealt with and the outcome?” and some people may struggle to answer because the query was handled well, but the outcome was poor. They’ll probably give an answer though and focus on just one of the two aspects. Problem is, you won’t know which one they were referring to
- If you’re using a rating scale, make sure that it’s balanced. For example, a satisfaction scale including ‘extremely satisfied’, ‘very satisfied’, ‘satisfied’, ‘not very satisfied’ and ‘not at all satisfied’ has three positive and two negative options. This will lead to a bias towards the positive
- Make answer option lists are as comprehensive as possible (a small number of personal interviews with the target audience is invaluable when forming lists like these, as you can establish all the potential answers up front). It won’t always be possible to create a perfectly comprehensive list though, so be sure to include an ‘other’ option so that people can write in their answer if it’s not on the list
- Always include a ‘don’t know’ in the answer options as otherwise people may guess an answer just to proceed to the next question. That will taint your data set
Now review the survey against the goals you set at the beginning of the design process to check that it will gather all the information you need. Be disciplined when doing so and remove any questions that aren’t essential to reaching your goal, even if they’d be nice to know. This will keep the survey as short as possible (aim for a maximum of 15 questions as this equates to a completion time of around 5 – 7 minutes) and in doing so, reduce drop out rates and enhance data quality (longer surveys see people randomly answer towards the end just to get through).
Finally, test the survey to check that it flows nicely, isn’t too long and works on mobile devices as well as PCs.
