Glassdoor reviews as a signal of corporate misconduct

Company reviews posted by employees are an indicator of the cultural health of an organisation.

Damming Glassdoor posts from bad leavers are part of the cut and thrust of managing an employee brand in the social media age.

I’m suspicious of any organisation that has a perfect Glassdoor score. You can smell the corporate overreach of an incentivised employee communication campaign to scrub reviews.

However, new research set to be published in the journal Management Science suggests that Glassdoor posts are a good early warning sign of a sick culture and corporate bad behaviour. You can find a summary on the Harvard Business School website.

Harvard Business School Professor Dennis Campbell and Tilburg University Assistant Professor Ruidi Shang conducted an experiment to investigate the relationship between employee Glassdoor posts and corporate culture.

The research team scrapped Glassdoor and used machine learning to analyse posts left by employees. They identified factors that signaled emerging scandals and corporate misconduct.

Campbell and Shang suggest that this technique could be applied by compliance departments in regulated industries to prevent fraud.

Glassdoor enables employees to post reviews of their employer anonymously effectively providing a safe outlet for whistleblowers.

The researchers created a dataset of reviews from more than 4,000 public companies between 2008 and 2016 and extracted a vocabulary of more than 11,500 words. An algorithm matched words to subsequent corporate violations reported in the media.

The result is a schema that calibrates the language in Glassdoor posts with high and low risk factors. It tracked the corporate misconduct at Wells Fargo between 2009 and 2013.

Campbell and Shang said that they we able to spot corporate woe at least a year of time.

The technique could be applied to any digital platform where employees are discussing their organisational culture.

Applications include employee listening, a signal for compliance departments, and a monitoring and alerting system for regulators.

As I start to think about research methodologies for my PhD I’m inspired by research projects such as this that use publicly sourced data to identify important societal issues.