IBSI Announces Subscription Flexibility !

IBS Journal now starts at £9 per issue, and custom Reports at 40% off
READ MORE

Dan Somers, CEO of Warwick Analytics

Banks that use ‘sentiment analysis’ technology to assess customer responses may be failing to gain genuine insights, an analytics expert has warned.
Dan Somers, CEO with software developer Warwick Analytics, has suggested that sentiment analysis – a form of analytics that picks out the emotions of people from writing or speaking – when used alone can produce misleading results, and is far more effective when deployed in tandem with ML and AI techniques.
“Sentiment analysis has served financial institutions well,” he said. “But the challenges, and opportunities, are becoming more significant. Machine learning promises much in the way of assisting text analytics to make it actionable, and now there are additional AI techniques and methodologies appearing which can uncover valuable hidden customer feedback and intent to turn customer words not just into charts but into actions, increased customer satisfaction and more profit.”
He said sentiment analysis used alone still requires a bank to carry out further processes to establish why customers might be unhappy with a service, potentially involving huge amounts of time and resource: “With machine learning, you simply act on the insight, making it a much better place to invest time and resources,” he added.
Somers said that as the volume, diversity and complexity of customer feedback data in the finance industry grows, there are increasing issues with sentiment analysis. Used on its own, for example, a sentiment analysis algorithm might be unable to understand irony, making a response like ‘Thanks for messing up my bank account, I really enjoyed speaking to a representative for two hours’ take on the opposite meaning to that intended.

by Guy Matthews
imp-loader
preloader