Top AI Customer Support Metrics Explained

If you are thinking of implementing Conversational AI for your customer support, it is important to measure the impact and ROI of your initiative otherwise it will turn out to be a technology investment for technology’s sake.

In Part 1 of the series, we attempt to break down the top AI Customer Support Metrics that you should be measuring with a focus on self-service.

AI Customer Support

Auto-Resolution Rate
When deploying a virtual assistant, your North Star metric will be to measure the percentage of inquiries that are automated by AI. This can also be a proxy for First Contact Resolution rate.

Escalation Rate
Not all conversations can be handled by a virtual assistant. It is important to sense when to handoff to a live agent or escalate to a customer support channel such as callback or ticket.

Top Conservation Intents
Breaks down conversation topic drivers to understand the types of inquiries users are engaging with the virtual assistant. This can range from reset passwords to canceling orders.

CSAT
Customer Satisfaction Score is a commonly-used key performance indicator used to track how satisfied customers are with a company’s product or services. It is usually measured in the as a percentage scale: 100% being total customer satisfaction, 0% total customer dissatisfaction.

CES
Customer Effort Score is used to measure how much effort a customer has to put in to get their issue resolved. It can range from a scale of “very easy” to “very difficult”. The more friction there is for the customer to get their issues, the more likely they are to abandon their journey. 

Average Sentiment Score
Machine Learning and AI models can now extract the sentiment from natural language conversations so you can be on top of the overall sentiment score of your customers and users while setting up workflows to react to interactions where customers are frustrated.

Ticket Volume
While it is good to measure how much of your incoming inquiries and requests are handled through a virtual assistant, it is good practice to correlate with the total ticket volume to see if the volume is trending down assuming all other factors are the same.

Average Resolution Time
The median time it takes for the virtual assistant to resolve a request from the time the user sends a request to the time it is resolved. AI Customer Support solutions drastically reduce average resolution time.

Top Knowledge Base Articles
As a supervisor, it is important to understand how effective is your knowledge case in covering self-service requests and so the breakdown of the top requests is an important indicator of what is used frequently.

Total Number of Sessions
Indicates total user traffic that the virtual assistant is receiving on a given day, week, or month. Ideally, you want more user traffic to be diverted to your self-service channels from other channels assuming your users are satisfied with the experience.

Additional Resources