https://support.tethr.com/hc/en-us/categories/201234238-About-Tethr About Tethr
https://support.tethr.com/hc/en-us/sections/202103128-Feature-spotlight Feature spotlight

New feature: content organization, collaboration and filtering

This article describes a new feature that may not be available to all users. If you don't see this feature and want to learn more, please contact your Tethr admin.
Section only

Overview

CSATai is designed to predict a customer satisfaction score on 100% of your customer conversations. Gain a holistic view of the customer service experience–and glean insights to improve your service delivery. By identifying factors that are decreasing customer satisfaction and actions that can be taken to improve it, businesses can maximize customer loyalty, reduce churn, improve future sales, and enhance their reputations in the market.

Contact your Tethr account manager to learn more about getting started with CSATai.

CSATai makes it easy to:

    • Gain a deeper understanding of customer satisfaction
    • Identify the factors impacting the customer experience
    • Rethink the way you’re measuring CSAT

Customizable CSATai dashboards are your home base for:

    • Monitoring satisfaction and tracking trends over time for your entire customer base
    • Recognizing and observing trends among your dissatisfied, neutral, and satisfied interactions 
    • Identifying what and who drives customer satisfaction as it relates to reason, behavior, and team hierarchy

Note: CSATai appears on individual interactions and can be used in reporting (filters, graphing, export).

How it works 

CSATai is a proprietary AI model that measures satisfaction by analyzing the words customers use in context. The model is trained on millions of customer survey results combined with their preceding interaction, either voice or chat. Using machine learning, the models capture the mathematical relationships between words and phrases in the conversation and the survey responses provided by the customer. The models are fine-tuned to ensure equal accuracy for both good and bad survey responses, thereby eliminating some of the affirmation bias present in the survey data. During the tuning process, the model parameters were varied and tested against untrained data to ensure they would generalize effectively to new conversations. While the test data is never used directly in training the model, tuning helps to identify information in the training set that is most important for capturing customer satisfaction. The resulting models are able to predict, with a high degree of accuracy, how a customer would most likely respond to a satisfaction survey.


Articles in this section