Surveys are, of course, one of the best ways to gather information and opinions from your customers or target audience.
They’ve long proven successful because you can focus on exactly the people you want opinions from, and ask questions about specific aspects of your business and products.
Questions are often quick to answer, and the abundance of online survey tools has made conducting surveys even easier.
The hardest part is analyzing that data.
In this guide, we’ll show you how easy it is to analyze your survey results and how you can visualize them in a striking dashboard.
Firstly, let’s go over what we mean by survey analysis.
Survey analysis is the process of analyzing results collected from customer surveys, such as NPS and customer satisfaction surveys. Your survey analysis, if performed correctly, will deliver actionable insights that you can visualize in all manner of business intelligence (BI) tools.
Before you start analyzing your survey data, it’s important to know what kind of results you need to decide what kind of survey analysis you want to do. Do you need to calculate market share? Rate overall customer satisfaction by percentage? Determine the features of your product your customers like best?
Or are you looking for more nuanced data about why your customers like certain features best or what drives customer acquisition, retention, and churn? Do you want to add new features to your product or find out new use cases?
Knowing your goals will help you decide the types of questions to ask, so that you elicit the right type of data: qualitative or quantitative data. Both can be effective tools for measuring customer satisfaction and gaining insights:
Quantitative data comes from close-ended questions, with a predetermined set of responses: Yes/No, multiple choice, or scaled (for example 1 to 5 stars). Close-ended responses are simply given a number value, so the results can be easily calculated, but they only go as far as comparing whole numbers, values, and percentages.
Net Promoter Score (NPS) surveys are an example of quantitative/close-ended survey analysis results, aiming to answer the question:
From 0 to 10 how likely are you to recommend [our product or service] to a friend or colleague?
Qualitative data comes from open-ended questions – questions that can’t be answered with a pre-set choice of responses. Open-ended questions are worded to elicit new information from the responder, in their own words. Open-ended responses dig into the responder’s feelings and opinions and allow for findings and personal data that the questioner may have never even considered previously.
You need to analyze your data, of course. Quantitative and qualitative survey results need to be analyzed in different ways, and there are many different tools you can use to make this process a lot easier.
First, let’s dive into how to analyze quantitative (close-ended) survey results.
Close-ended/quantitative survey questions and responses are relatively easy to analyze because the results correspond with whole numbers and percentages, so it’s mainly about comparing them against each other.
This is a scale question review, ranking the app from 1 to 5 stars on General Use (the “Stars” column), Value for Money, Features, Customer Support, and Ease of Use:
To begin analyzing survey results, follow these 4 steps:
Close-ended survey results can easily be filtered to show respondents that gave a rating of 3 stars or lower, for example:
Excel, Google Sheets, and other spreadsheet programs have built-in formulas. In Excel, simply highlight the entire column (or group of columns), and choose the correct formula in Excel. Type ‘=average’, ‘=max’, or ‘=min’ in the corresponding field and Excel will calculate the statistics for you.
Cross-tabulation allows you to easily compare one group of respondents with another. This is done in Excel with a pivot table. Focus on specific age groups, occupations, geographic regions, etc.
In this table, we can see a summary of the column "Stars." It counts the number of times each star appears in a review.
Close-ended results will give you a good idea of what is happening. If you’ve collected data correctly, you’ll be able to understand in what region and with what demographics your brand is strongest and what new areas you may be able to expand into.
You could, for example, find out that Pricing is not a particular issue for any of your customers, so it might be about time to raise your rates. Close-ended results, however, don’t allow you to fully understand why your customers feel the way they do, simply that they “do” or they “don’t.”
Open-ended survey responses give you a much more nuanced view into the opinions, feelings, and emotions of your customers, simply because they are responding in their own words and not reacting to leading questions or with presupposed response choices. Although this qualitative data is much more useful, it’s also a bit harder to analyze.
Fortunately, you no longer only have the option of costly, time-consuming, and not entirely accurate hand-annotated human analysis, you now have the power of machine learning on your side.
MonkeyLearn integrations make it easy to gather data from online survey tools, like Google Forms, SurveyMonkey, and Typeform or other tools you already use like Excel, Gmail, Zapier, Zendesk, and more. Or, if you’re a programmer, you can connect your data using MonkeyLearn API.
Once you have your open-ended responses, it’s just a few steps to results and data-driven insights.
Open-ended survey responses and other forms of customer feedback (from social media, online reviews, customer support data, etc.) often contain more than one opinion or statement in a single response – or even within a single sentence.
So our survey responses need to be separated into individual “opinion units” for optimal analysis. Sign up to MonkeyLearn for free to automatically separate thousands of survey responses into opinion units using try this pre-trained opinion unit extractor:
MonkeyLearn offers dozens of text analysis options, but we’ll go through just two that are often the most relevant for survey data analysis: sentiment analysis and topic classification.
Topic classification, also called topic or “aspect” analysis, is a machine learning text analysis technique that can automatically sort open-ended surveys (and all manner of text data) into pre-set categories, subjects, or topics.
Try out this pre-trained survey analyzer that automatically classifies survey results into categories: Ease of Use, Features, Pricing, and Customer Support:
The confidence level is pretty high. However, when you train your own models to your criteria, you can regularly expect total accuracy and even higher AI confidence. With machine learning, the more you train your open-ended survey results model, the smarter it will become.
Sentiment analysis is a text analysis technique that’s perfect for open-ended survey results analysis because it can automatically read responses for the feelings and emotions of the respondent by analyzing text for “opinion polarity” (Positive, Negative, and Neutral).
Try out this pre-trained MonkeyLearn sentiment analyzer to see how it works:
It’s fast and accurate. Imagine running an open-ended survey results analysis with a custom-built sentiment analyzer on thousands, even hundreds of thousands of open-ended responses.
When you join the three techniques above, you get an unmatched, expert machine learning survey analysis technique called aspect-based sentiment analysis.
Aspect-based sentiment analysis automatically classifies each unit by topic or aspect, then performs sentiment analysis. So you end up understanding the sentiment behind each opinion, as well as which aspects of your business or product are performing most positive and most negative.
MonkeyLearn offers a number of other powerful tools to get the most out of your data, like keyword extraction (to automatically find the most used and most important words from any text) or intent classification (which reads emails, support tickets, and more, for the purpose of the writer, for example: Complaint, Product Question, Purchase Question, User Support, etc).
Data visualization tools allow you to show off your survey data results, often in quite striking detail, for immediately recognizable insights.
Excel, Google Sheets, and online survey tools have built-in quantitative data analysis tools, like tables, charts and graphs: line graphs, pie charts, bar graphs, etc. This bar graph shows the distribution of stars by category from our Slack reviews above:
To visualize your open-ended survey data results, you might want to take MonkeyLearn Studio for a spin. MonkeyLearn Studio is an all-in-one survey data gathering, survey analysis results, and survey results visualization tool. Once your analysis is set up, it works automatically, in real time.
The below is a MonkeyLearn Studio analysis of open-ended survey results from Zoom customers:
You can see top keywords by aspect and sentiment in the columns on the left, overall sentiment in the bottom right, word clouds to the left of that – it’s a lot of data but easy to digest.
Try out the MonkeyLearn Studio public dashboard to see what it can do – change by date, category, intent, find the most positive or negative statements, and more.
Whether you’re working with the simple statistics, whole numbers, and averages of close-ended quantitative survey results, or diving headlong into your customers’ free-flowing opinions, feelings, and emotions with open-ended qualitative survey results, you’re guaranteed to uncover useful insights about your customers and the customer journey.
While close-ended survey results may be a bit tougher to work with, they can produce much more valuable information to help your company gain real-time, real-world, data-driven insights. And SaaS tools, like MonkeyLearn make machine learning accessible to everyone, at any level of computer literacy.
Sign up to MonkeyLearn to try the tools for free before you buy; take a look at MonkeyLearn’s pricing page to see your plans and options, or schedule a demo, and we’ll walk you through everything you need to know, from survey execution, to analysis, to survey results visualization.
January 18th, 2021