Employee monitoring vs listening: Meta’s AI lawsuit raises the stakes

An employer trying to predict who will perform, leave or give extra effort should combine behavioral data and metadata, structured survey data, and unstructured experience data such as comments, said Granger. “You’re going to get much better prediction if you include all three, than if you just include one or the other, because they tell you different things.”

Large language models (LLMs) make the unstructured layer usable for the first time. “Now with some of these LLMs and tools that we have available, we can really get down to how people feel about particular issues. How strong is their emotion around it? What’s their intention?” he said.

But that potential comes with heightened conditions, Granger continued: Organizations must be transparent about what they do and don’t do with chat and forum data, he said, and use it to improve the employee experience before anything else.

“Assuming that’s true, then that passively collected data is an incredibly rich source of how people really feel about things,” said Granger. But if passively collected data doesn’t first definitively improve conditions for employees, employees will never buy in with trust and may shortly turn their backs, he said.

Why feeling heard pays off on its own

Qualtrics’ own research supports the case for transparent listening, though the company is an experience management software provider and the findings are its own. The company’s 2026 Employee Experience Trends Report, which surveyed 33,831 employees across 24 countries and 30 industries in September and October 2025, found that organizations that increased listening frequency saw employee experience metrics up to four times higher than those that listened less. The same study found that 72% of employees had experienced significant organizational change in the prior 12 months.