If we can all agree that the data extracted from customers’ digital footprint such as their activities on the web and social media behavior and activities is one of the greatest developments that came to the data collection and data analysis realm.
Imagine all that data had to be entered by people themselves if they had to write what they like, what website or page they visited today, a week ago, a month ago…and then submit it to cookies, pixel, Facebook, or google team and so on. In his way, we will all end up with either a pile of garbage at worse or a lot of data processing and cleaning at best
Now, this is what happens every time a market research survey or questionnaire is filled up with poorly designed open-ended questions that are expected to bring back the perfect answer in the case of online panels without the luxury of a moderator that can shed some context to these questions.
Researchers would say: if we can’t ask a specific question just the way we want it, what is the point of conducting the research?
Well, this is what I’m going to answer quickly here. Definitely, the data collected from digital footprint is valuable however it’s not inclusive and it does not put a lot of specifications into consideration. There isn’t any clear-cut answer but there are some techniques to follow and implement to limit the quality issues while maintaining the structure of the study:
- The length of the questionnaire: you lose the respondent’s attention with time, you better get your answer as soon as possible before the attention fades away gradually with time, hence keep it within 5-10 minutes.
- Avoid similar questions: where the respondent gets a group of questions where only one word has changed, the respondent will soon get used to the questions and unconsciously start giving the same answers.
- Minimize the number of open-ended questions: you will not only save time on data processing but remember not everyone is good at spelling or know how to express themselves with a clear point.
- Seamless floating design: Keep only important details on the screen, and avoid filling the screen with information, unfortunately, respondents usually do not read everything on the screen, keeping only the necessary part will element this issue, one question per screen will put all the focus on the particular question.
- Ask a data scientist: researchers tend to be biassed toward what they believe is perfect questionnaire design, data scientists, on the other hand, understand the possible variety or answers for the same question, by reverse engineering these answers they can help design better questions.
- Speeders do exist but: always base your length of the interview on the actual data, researchers will estimate the length but it’s just an estimate it should not carry more weight than the actual infield length, take the median time spent on the survey, the median will eliminate outliers.
Conclusion: “understanding a question is half an answer” ― Socrates.