Summary: We discuss some of the challenges employers, marketers, and others have when collecting qualitative feedback. We also discuss how the world of surveys might change thanks to new technologies and research approaches.
Employees receive pulse surveys, quarterly surveys, 360-degree feedback surveys… There are many ways to collect feedback, and the frequency of feedback collection is growing. We regularly meet with companies who run surveys on a monthly basis – where employees provide a mix of quantitative answers (e.g., employee net promoter scores or Likert scales) and text-based feedback.
The challenge with more surveying is that response rates decline; survey fatigue means you get less complete responses when employees are asked to elaborate. Most HR and employee listening programs prefer frequency over depth because at least you still get more data and can continuously improve the employee experience. So how do you try to improve both frequency and feedback quality?
Phase NLP is now experimenting with a new type of survey approach. Most surveys are structured the same for every recipient. This makes sense to the extent that questions and answers can be compared, but is a problem for qualitative feedback – some employees provide enough context to address a problem, and others give short answers that are hard to decode.
With our approach, we take a text-based answer from an employee and pass it through our NLP models to score whether the sentence is raising specific themes, and whether it is providing enough data to be actionable.
If the response doesn’t have enough details, the survey generates questions that ask the employee to dive into details of their feedback. For example, if an employee raises dissatisfaction around compensation, the follow-up question becomes, “What is it about your compensation that makes you dissatisfied?”
When the employee’s response is complete and provides details on the issue, then the survey concludes.
The above is only possible with numerous models and the ability to generate natural language questions in real-time, as an employee answers a survey. This is a key part of our product roadmap and approach!