top of page

Health Score Ingredients overview

About Health Score Ingredients

See how your organization’s KPI’s uniquely impact health scores

Health Score Ingredients shows you how involve.ai’s machine learning models weigh each type of customer data your organization provides. The weighting you see is unique to your organization, based entirely on which variables have historically most significantly correlated with churn.



In this example, the KPI most impactful to health for this particular customer is Product Usage.

Understand your customers’ data-driven segmentation

Health Score Ingredients also shows how our AI segments your customers. By analyzing all possible variables simultaneously to determine which customers cluster most closely together, these segments allow normalization, the ability to determine accurately whether each metric for a specific customer is low, medium, or high.

The use of multivariate data and correlation analysis to segment customers creates a much more highly accurate comparison than traditional segmentation, which relies on the use of arbitrary attributes or intuition.



How to use Health Score Ingredients

Purpose

Health Score Ingredients is to help you understand and trust the accuracy of involve.ai’s machine learning models. Unlike the involve.ai customer health score or KPI fields, your Health Score Ingredients will not regularly change or update, nor will it provide specific information for any individual account. It is meant to help you understand how our AI functions in the aggregate. We recommend reviewing the Health Score Ingredients as a team on a quarterly basis, or if you add a new set of data or integrations as doing so may impact the weights of each ingredient.

Access

To review Health Score Ingredients, click the HEALTH SCORE INGREDIENTS link at the top of your Customer Health Data dashboard.

59 views

Recent Posts

See All

How we test accuracy and performance

Our AI predicts customer behavior with greater than 90% accuracy. How do we know? We test and measure the performance of our models regularly, in a variety of ways. Model training test accuracy - 94

Comments


bottom of page