One common approach to measuring customer satisfaction after each call is to simply ask customers to rate it right after the interaction. Sounds straightforward, right? But what happens when customers choose not to participate in the survey?
This is a challenge many businesses face.Take Intuit, for example. Known for popular software like Quickbooks, Mailchimp, and TurboTax, they encountered a frustrating issue—only 10% of their customers were filling out post-call surveys.
This left a significant gap in understanding customer sentiment, especially from those who may have felt misunderstood or dissatisfied.While team managers could review call recordings, analyzing 90% of them isn’t feasible.
Training customer service agents regularly is another option, but it can be costly and often lacks targeted insights.
Determined to find a more effective solution, Intuit began transcribing past phone conversations linked to survey data to train an AI model capable of predicting the sentiment of conversations—good, bad, or neutral.
Once the model was ready, every new call transcript was analyzed, allowing agents to see their performance metrics by the end of each day and make necessary adjustments. The best part? The AI model continuously improves from ongoing feedback and new data, refining its predictions and personalizing ratings for each agent.
Thanks to this innovative approach, Intuit gained unprecedented visibility into their customer service organization, all while significantly reducing costs compared to traditional human supervision.
This is how Intuit automated customer service call ratings and transformed their approach to customer satisfaction! 💡✨