The Benefits of Online Survey When They “Talk” Less and “Listen” More

In this article, Rasto Ivanic, co-founder and CEO, GroupSolver, shares how an online survey platform can lead to better customer engagement and in-depth insights on what customers want.

Online surveys give businesses valuable insights into their customers’ perceptions, preferences, and buying behaviors. But their inability to easily support open-ended questions has meant researchers spend more time “talking at” respondents than really listening to them. As they search for better alternatives, researchers are finding that a combination of machine learning (ML), natural language processing (NLP), and crowdsourcing techniques accommodates open-ended questions and creates a more conversational, less transactional online survey experience.

Why Surveys Are So Hard of Hearing

Researchers would like to hear the voice of the customer, but traditional online survey platforms have made this listening difficult. Natural language data collected through the traditional text box generates massive amounts of raw, unorganized data that requires substantial time and resources to read, clean, and process. As a result, researchers often gravitate to using closed-ended questions. While this makes the project much easier for the researcher, it severely limits respondents’ ability to “talk back” to the researcher.

It also impacts data quality. Respondents who check box after box of multiple-choice options experience survey fatigue, resulting in “straight-lining” answers (i.e., selecting the same option question after question). Plus, respondents often do not entirely focus on each question.

Learn More: 4 Ways To Use AI-Powered Chatbots To Boost Your Customer Experience

Researchers who understand the pitfalls of closed-ended questions have tried to retain the text boxes and rely on text analytics to deliver the desired insights. But this does not address the problem of respondents who often enter meaningless or gibberish answers or just skip over one or more text boxes altogether, making the researcher forgo any data from them. This dramatically reduces the depth of the resulting data. Moreover, seeing text box after text box can lead to the same survey fatigue observed with long multiple-choice surveys.

These challenges can only be solved by making online surveys smart enough to turn freely expressed thoughts from respondents into useful data for researchers in real-time. New technologies make this possible, creating a win-win for both respondent and researcher.

 Smart New Listening Techniques

The road to better listening through online surveys starts with a smart combination of NLP and ML technologies. Additionally, respondents’ own “crowd intelligence” can be used as part of a new, more interactive process of collecting and validating data, leading to more useful results for the researcher.

First, NLP and ML technologies are used to perform a first-pass clean-up of text answers in real-time automatically. The cleaned-up answers can then be fed back to the survey takers in the form of a mini-survey that allows them to elaborate on their earlier answers. This crowdsourcing process also helps to quantify the original text answers, which previously could only be accomplished using traditional multiple-choice surveys.

Because this new process is interactive and more “gamified” than multiple-choice surveys, it ensures that at least some useful data is acquired from all respondents — even if it is only a read on how they feel about others’ answers. Respondents enjoy the process, making them more attentive and engaged. This gives researchers higher-quality data, the majority of which has been generated directly from the voice of the customer.

The Benefits of Better Listening

 A survey conducted by a large streaming music service provides one example of how these latest listening techniques create new opportunities for researchers and respondents alike. The service’s research team needed to understand customer behavior but faced all the challenges of conventional online surveys. It could only glean minimal information through the traditional close-ended survey approach. But the alternative of using open-ended questions in a typical qualitative survey with 10,000 study participants was simply not feasible. The team knew it could take many days of manual data review, cleaning, and organizing. This would delay critical decision-making, which, along with the additional labor required for data review, would cause research budget overruns.

Learn More: 4 Ways Data Drives Truly Amazing Customer Experience

The latest ML and NLP techniques solved the research team’s problem of capturing and processing text responses from a large number of respondents and then enlisting those same respondents in an engaging and enjoyable real-time data validation and quantification process. The process’s efficiency enabled the team to capture from less than 2,000 respondents the same amount of useful data that was previously going to require 10,000 survey-takers using traditional online techniques. The answers they captured to open-ended questions about why respondents listened to music enabled the team to start describing their customers in more meaningful behavioral and attitudinal terms. These range from the deeply spiritual (“I connect with God through music”) to the practical and behavioral (“Music allows me to focus while I’m working”). Furthermore, they can also use the statistically validated qualitative data in quantitative models, such as net promoter score (NPS), segmentation, or pricing studies, where natural text data can now become categorical variables in quantitative models.

Combining AI and NLP technologies that smartly integrate into online surveys is making lasting inroads into the traditional market research toolkit. Combined with leveraging survey respondents’ crowd intelligence, these technologies make it possible to seamlessly integrate open-ended and choice questions while moving the survey experience from box-checking toward free-flowing natural conversations at speed and scale. This is creating an opportunity for significantly better listening to the customer’s voice and for researchers to integrate unstructured survey data more directly into insights and the statistical models that support them.