Evaluate Weigh the pros and cons of technologies, products and projects you are considering.

Employee sentiment analysis may trump humans in accuracy

Employee sentiment analysis has the potential to help HR managers examine survey data more quickly and accurately.

Also called opinion mining, employee sentiment analysis uses natural language processing (NLP) and machine learning to analyze huge amounts of data quickly.

'Lighthouse Research and Advisory and cloud HR vendor Ultimate Software investigated whether employee sentiment analysis is better than humans at identifying employee sentiment and emotion in employee survey data, said Ben Eubanks, principal analyst at Lighthouse Research and Advisory, in a video interview at the 2019 HR Technology Conference in Las Vegas.

The researchers surveyed 1,000 North American workers to understand work issues such as stress and relationships with managers, he said. They then shared the results with consulting firms and asked them what the data told them about hot buttons and friction points in the employee experience. Then they ran the same data through Ultimate Software's NLP engine, which has employee sentiment among its capabilities, he said. Another term for employee sentiment analysis is "augmented listening," he said.

"We were able to prove with this research that AI is better than humans at analyzing survey data for those kinds of things," Eubanks said.

Eighty percent of HR managers he surveyed said they would be interested in a tool that auto-analyzes employee survey results, Eubanks said in a conference session.

Applying AI to employee sentiment data has the potential to improve employee engagement and reduce turnover and training and recruiting costs, he said. One company that used employee sentiment analysis reduced survey turnaround by 87% to two days. 

The near-instant analysis is one of the major draws of AI-based technology.

"AI is better than humans at analyzing [employee sentiment], but obviously it can't go forth and solve the problem," he said.

Such timeliness is critical in employee sentiment surveys, Eubanks said. One 20,000-employee company took six months to decide on a survey and get the results.

"The problems those employees had four months ago? They're not even here anymore," he said.

If HR managers can make at least one step of the employee survey process easier, that's a win for everyone since surveys pose a number of challenges.

More than half of HR managers only do employee engagement surveys annually, according to a separate survey by Lighthouse Research and Advisory.

Many HR managers struggle with how to use the data, Eubanks said. Few have time to analyze the unstructured data from open-ended questions, so they tend to use only multiple-choice responses.

Others worry that if they ask employees for feedback, employees will expect them to act on it.

"Once I [as an employee] told you that there's a problem, I expect some sort of response to that, and if I don't get it, that creates a very negative experience," he said.

Measuring employee sentiment can create privacy issues but companies can take steps to ensure they use survey data, AI and NLP applications in a legal manner.

View All Videos