So you have designed your survey, collected your data and now its time to analyze your survey results. So where do you start? What techniques do you use? How to represent your insights? This post will show you how to get the most of your survey.
Starting Survey Analysis:
Before you begin the analysis, you should understand who answered your survey. It is a good idea to get a summary of:
- Total number of responders
- % of incomplete survey results – This can become a big issue if you did not randomized your questions and do not have a lot of replies for questions found at the bottom of your survey.
- Demographic information – If you had any questions or data regarding who took the survey, it is a good starting point to summarize the demographical data in order to set up summary.
Descriptive analysis is simple analysis that describe the results of your survey. Here is simple way of analyzing the survey questions:
- Yes/No Questions – This can be looked at by simply looking at what % of responders replied Yes or No to a question.
- Share of Total – You can calculate share of total or percentage of total. For example, what % of people that live in a certain part of the country, % of people that loved a product or % of people that prefer a certain product or service.
- Average – While index/averages is something that you should be careful with they maybe important when looking at certain metrics. Such as average age of respondent or average satisfaction and so on.
- Cross Tabulation – You may want to look at two or more questions combined to see if you can find insights.
Depending on your survey you may want to look at the correlation between questions. You can calculate the coefficient correlation between certain questions. Keep in mind just because something is correlated does not mean there is a cause.
Segmenting Survey Results
If you have collected enough responders, you maybe able to provide a more detailed analysis split by different segments. Segments may include:
- Demographical – age, household income, job title, and company size are all examples. For example, you may discover that millennials prefer online services vs. baby boomers who still talk to agents on the phone or that Enterprise businesses are less sensitive to pricing compared to small private owned businesses.
- NPS – If the survey was based on NPS score, you may want to look at those that are detractors vs. promoters to see what differentiate those that are happy to those that are in danger of leaving.
- Purchase Behaviour – if the survey was based on customers. You may want to look at the results based on purchase amount, loyalty level and even certain products purchased. The key here is to have that data collected and have enough responders to be able to drill down to the product level.
- Geographical – If you are running a national or international business, it is important to understand cultural differences that could impact your survey results.
Response Over Time
If the survey is repeated, it is key to look at the responses over time. For example, if you do a yearly survey on customer satisfaction, it is important to see if NPS score has improved or gotten worst. If you served the same people, it is important to compare answers to previous years.
In most surveys you are likely to find comments and text imputed under “Other”. For all the different “Other” responses, you can use categorization to group similar comments into the same category. For comments, you can use text mining techniques to analyze comments. It is often the section that can bring the most insight if analyzed correctly.
These are just some strategies that you can use when analyzing surveys. If you are looking to analyze survey results, StrategicDB can help, you can contact us at firstname.lastname@example.org.