It can be difficult to get customer feedback. You need to have strong survey questions that will get customers to share their insights. You also need a system for distributing the survey to the right customers at the right time.
Survey data does not automatically sort and analyze itself. A team is needed to go through the results and identify key trends and behaviors for marketing, sales, and customer service teams.
This post provides the steps on how to analyze survey results like a data professional.
1. Understand the Four Measurement Levels
The four levels of measurement are nominal, ordinal, interval, and ratio. Nominal level questions can be answered with a simple “yes” or “no.” Ordinal level questions can be answered with a ranking, such as “first,” “second,” or “third.” Interval level questions can be answered with a specific number, such as “5 out of 10.” Ratio level questions can be answered with a specific number and a unit of measure, such as “5 miles per hour.”
The four levels of measurement are nominal, ordinal, interval, and ratio.
A nominal scale is a data classification system that does not use any numerical values, only labels. An example of a nominal scale is “Select your car’s brand from the list below.” The choices on the list have no numerical relationship to each other.
The only way to know how many people selected each option is to keep track of how many people chose each option, and which option was chosen the most.
An ordinal scale is a way of representing data in which there is a rank or order to the values. So, for example, if you were to ask someone to “Rank the reasons for using your laptop,” that would be an ordinal scale. The advantage of using an ordinal scale is that it allows you to order the data from highest to lowest (or vice versa).
This type of scale allows you to analyze the mode and median, and you can use cross-tabulation analysis to interpret ordinal scales.
Interval scales show both the order and difference between values. These scales have quantitative value because the intervals between data points remain the same along the scale, but there is no true zero point.
An example of an interval scale is an IQ test. You can use mode, median, and mean to analyze data from this type of scale, and you can use ANOVA, t-tests, and correlation analyses to further analyze the data.
A t-test is a hypothesis test that is used to determine if there is a significant difference between the means of two groups. An ANOVA test is a statistical test that is used to test the significance of survey results. A correlation analysis is a statistical test that is used to determine if there is a relationship between two variables.
In other words, the distance between 0 and 1 on a ratio scale is the same as the distance between 2 and 3 or between 100 and 101. Ratio scales show the order and difference between values, but they don’t have a true zero point. With ratio scales, there’s quantitative value because the absence of an attribute can still provide information.
In other words, the distance between 0 and 1 on a ratio scale is the same as the distance between 2 and 3 or between 100 and 101.
A ratio scale measures variables on a scale where the value 0 represents the absence of the variable and division can be used to compare two values on the scale.
2. Select Your Survey Question(s)
or “What’s our customer satisfaction rate?” You should take note of the overarching survey question(s) that you’re trying to solve. Perhaps, it’s “How do respondents rate our brand?” or “What’s our customer satisfaction rate?”
Start by determining what your research goals are. Once you know what you want to learn from your survey, look for questions that will help you achieve those goals.
For example, if you want to learn how likely people are to recommend your brand to others, look for a question that asks that specifically. Segmenting your survey questions in this way will help you get the most relevant data for your purposes.
It is also important to ask both types of questions, those that have a definitive answer and those that do not.
A close-ended survey question does not allow respondents to explain their answers. The respondent can only choose from a pre-determined set of answers, which could include yes or no, multiple-choice, checkboxes, dropdown, or a scale question. It is important to ask a variety of questions to get the best data.
An open-ended survey question is one that allows the respondent to explain their opinion. For example, in an NPS survey, you would ask how likely a customer is to recommend your brand.
After that, you could ask customers to explain their choice. This could be something like “Why or why wouldn’t you recommend our product to your friends/family?”
3. Use Cross-Tabulation to Better Understand Your Target Audience
If you group all of your responses together, it isn’t effective for gaining accurate information. People who aren’t your ideal customers can ruin your data and make your survey results inaccurate.
Instead of segmenting responses using cross-tabulation, you can analyze how your target audience responded to your questions.
Split Up Data by Demographics
Cross-tabulation allows you to see the relationships between variables by comparing two sets of data within one chart. This can reveal specific insights based on your participants’ responses to different questions.
For example, you may be curious about how likely your customers in Boston, MA are to recommend your brand. You can use cross-tabulation to see how many respondents said they were from Boston and said they would recommend your brand.
If we plot several variables on one chart, we can focus survey results on a particular group of responses. This ensures that your data only relates to your target audience.
4. Understand the Statistical Significance of the Data
It’s important to make sure your data is accurate and representative of your target audience.
The data indicates that a large portion of those who would recommend the brand are not in the target age group.
This data is not significant because the people who took the survey do not reflect your ideal consumer.
Random sampling is a good way to get a diverse group of survey responses. The more people you survey, the more accurate the results will be.
p < 0.05,” meaning that if a p-value is less than .05 (5%), then it is statistically significant. So in this case, almost all of the differences are statistically significant.
SPSS will tell you if a data point is statistically significant using a p-value. If the p-value is less than .05 (5%), then it is statistically significant. In this case, almost all of the differences are statistically significant.
The correlation is significant at the 0.05 level (2-tailed) and 0.01 level (2-tailed).
A data point has moderate statistical significance if the p-value is 0.05 or lower. This means that there is less than a 5% chance that the data is incorrect.
If the p-value is lower than 0.01, this means that the results are more likely to be true, as the probability for error is less than 1%.
5. Start With the Numbers
It can be difficult to know where to start when there is a lot of work to do. Sometimes we can feel overwhelmed and frozen in front of a project without knowing what to do next.
You should prioritise quantitative data over qualitative data as it is easier to compare numbers than to analyse long-form responses.
Would you like to know if customers plan on returning to your store?
The respondents are given the option of responding with “yes,” “no,” or “not sure,” along with an opportunity to explain their answers.
Karen, one of your customers, has selected “no” and has written a three-page essay on the service in your store in a tone of what can only be described as ‘seething rage.’
If you began your data analysis with Karen’s manifesto, where would that lead you?
The text is saying that the opinion of the person mentioned could be different from the opinions of the rest of the customer base.
When you examine your data quantitatively, you get an immediate visual representation of it.
We can see from the data that 71% of people who took the survey would return to the store.
When you compare Karen’s feedback to the store’s popularity data, you get a very different idea of how well the store is doing.
The reason you begin with quantitative data is that it is the most essential type of data to collect.
After you’ve looked at the numbers, read the responses to get a better understanding of the data.
6. Contextualize the Data
It’s best to think of data in the context of the situation, similar to how water is to a dolphin. Without the proper context, the data doesn’t full mean anything.
Benchmarking is a method of finding context for your data. It allows you to make sense of numbers and understand what they are telling you.
In other words, benchmarking is a way to compare your company’s performance to other companies in your industry. This can be done by looking at financial reports, customer surveys, or other metrics.
The data from last year’s charity event survey would be used as a baseline for this year’s survey.
The key findings from the initial survey would be used as the starting point for the next one.
You can immediately see if you have improved or gotten worse based on past results, without doing any additional calculations.
The frequency with which you compare benchmark data depends on the type of survey you are conducting. Sometimes it is appropriate to compare data monthly, while in other cases yearly comparisons make more sense.
We conduct weekly meetings at Paperform to go over marketing metrics. This includes reviewing website visitors, blog post readership, and new customer sign-ups.
This allows us to keep tabs on our progress, identify areas requiring improvement, and identify patterns as they emerge.
We may get more visitors or customers one week due to a great blog post or more spending on advertisements.
This allows us to see how we are doing in comparison to last week, month, or year. The goal is to see what trends are emerging and how responses have changed over time (longitudinal analysis).
Don’t worry if this is your first time collecting data. Everyone has to start somewhere. Just use your first survey as your baseline.
This can be used as a benchmark for future comparisons and as a way to filter or compare data.
Survey responses can be sorted with filters or crosstab to find interesting information not apparent in the overall results.
If you want to be able to make use of your survey data, it is important to make sure that the data is specific.
7. Draw Conclusions
A time will come when you have gathered all the quantitative data you need. Once you have all the numbers, you will want to understand why respondents are answering in a certain way.
Why something is the case will sometimes be answered through surveys, and other times it will be up to you to investigate further using qualitative data.
When you read through people’s responses and comments, you can get a better understanding of the story behind the numbers. You shouldn’t do this to make yourself feel good or bad, but to be able to see the bigger picture by putting the data together with what people are saying.
In order to effectively distinguish between correlation and causation, it is necessary to understand the distinction between the two.
- causation is when one thing causes another
- correlation is when two things move together, but one didn’t cause the other.
Having a heater on and wearing tracksuit pants are two things that usually happen when it’scold outside. They’re not directly related, but are both caused by the cold weather.
But there is causation between tracksuit pants and winter.
Sales of tracksuit pants go up as the temperature gets colder. The cold weather is a factor that affects the sale of tracksuit pants.
When analyzing your survey results, be careful not to mistake correlation for causation. Just because two variables move at the same time does not necessarily mean that they are affecting each other.
Although your charity event was successful, your survey respondents did not like the location.
After looking through the qualitative data, you discover that a common complaint was that the event wasn’t near any public transport.
You’ve found a correlation between the location of the event and people’s level of satisfaction. People were less satisfied because the event was not close to public transport.
This is your chance to use the information you have to choose a better location next year.
It’s all about finding the hidden story in your data and using it to make improvements.
There are many ways survey data can benefit your business, such as rebranding your business or evaluating your marketing campaign.
It’s important to take your time reading through long responses, so you can filter out the complainants who are are angry about things you can’t control. Only focus on the feedback you can use to improve your service.
You need to listen to your respondents in order to get useful information from them, rather than wasting time, effort, and money.
8. Present Your Results
The data you have collected is meaningless unless you share it with others. In order for your survey data to have any impact, you need to present it in a way that will show its significance.
Whether you’re presenting to stakeholders or giving your sole business partner a rundown of all you’ve learned, there are a few things you want to achieve:
- Make sure that the findings are clear
- Show areas for improvement
- Provide recommendations
- Get data to the right places (in the right way)
The final point is very important. conducting customer research and not sharing the results with your product team is pointless. Make sure the research reaches the right people so that changes can be made.
Don’t try to include every detail of your survey analysis process when talking to someone. You don’t need to include every number and calculation.