Developing an AI analyst to automate customer support insights

How to use ChatGPT to analyze your support system based on key metrics.

Tags Streamline Icon: https://streamlinehq.com
Wrench Streamline Icon: https://streamlinehq.com
Uses
People Man Graduate Streamline Icon: https://streamlinehq.com
Beginner
Paragraph Bullets Streamline Icon: https://streamlinehq.com
Included in the course:
Implementing AI in customer support

Over the past three tutorials in this course, you’ve learned about the importance of customer support AI assistants, how to set one up, and how to manage its documentation. In this tutorial, you’ll learn how to analyze your support system based on key metrics.

It’s important to understand the metrics you’re tracking and your progress towards them to measure the efficacy of not only your customer support AI assistant workflow but your customer support function writ large.

You’ll need:

Steps:

  1. KPI assessment
  2. Analyze the system based on responses
  3. Improving the system
Products Give Gift 1 Streamline Icon: https://streamlinehq.com

Try before you bite?
This one's on us.

Sign up for a free account to view our free courses

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.

Step 1: KPI assessment

Before we dive into the existing data of our support system, we should understand if we have a comprehensive set of metrics to properly measure our customer support function.

We can use ChatGPT to review our KPIs and provide feedback on our measurement system. We recommend using ChatGPT for this exercise, not your customer support GPT.

Open up a new window with ChatGPT to get started.

Sample prompt:

We currently measure the efficacy of our customer support team via [Insert Metric] and [Insert Metric]. What are some other metrics and KPIs we can track to improve our customer support function?

You can ask ChatGPT to provide instructions on how to measure and implement a specific metric in your customer support workflow.

Customer Effort Score (CES) is something we want to track. How can we best implement a tracking mechanism for this metric?

Step 2: Analyze the system based on responses

Once you have your set of KPIs and you have reportable data, the next step is to start analyzing the data. We recommend using ChatGPT again for this exercise, not your customer support GPT. We’ll start by analyzing the sentiment of our customer support conversations and tickets.

#1. Sentiment analysis

Ask ChatGPT to create a comprehensive sentiment analysis of your support conversations. Make sure to upload a file of your customer support conversations with your prompt.

Attached is a set of customer support conversations with my customer support agents and customers. Can you provide a comprehensive sentiment analysis of the conversations?

From there, you can dive deeper into the sentiment analysis, asking for example threads with the most negative and positive sentiment (for future employee training examples).

Can you provide the ConversationIDs with the most positive sentiment and the ConversationIDs with the most negative sentiment? If there are any common themes among the conversations in these two groups, please provide a brief analysis.

#2. Response times

Once you have an understanding of sentiment, you can explore the response times of your customer support tickets. To start, we’ll have ChatGPT provide the average handle time (AHT) of our support conversations. This is the total time it takes to resolve a customer issue, from initial contact to resolution.

Can you provide the Average Handle Time (AHT) of the customer support conversations? This is the total time it takes to resolve a customer issue, from initial contact to resolution.

We can have ChatGPT plot this data visually over time as well.

Can you create a bar chart that plots the Average Handle Time (AHT) by time of day? Please use minutes for the y-axis and hours for the x-axis.
💡 Tip: Based on your dataset, you can instruct ChatGPT to construct different visualizations based on the day of the week, month, or other parameters. You can also ask ChatGPT to group the data based on what it feels is the most interpretable.

Next, we can explore a first contact resolution metric (FCR) of our support conversations. This is the percentage of customer issues that are resolved on the first interaction without requiring follow-up or escalation.

Can you provide the percentage of customer issues that are resolved on the first interaction without requiring follow-up or escalation?

#3. Customer satisfaction

Finally, you can explore the customer satisfaction (CSAT) and net promoter score (NPS) of your customer support conversations.

CSAT is the percentage of customers who report being satisfied with their support experience, typically measured through post-interaction surveys, and NPS is a measure of customer loyalty and likelihood to recommend your brand based on a single survey question.

Based on the Feedback Scores in the data, what is our CSAT?
Based on the Feedback Scores in the data, what is our NPS?

Step 3: Improving the system

Now that we understand the various metrics within our customer support tickets, we can start working with ChatGPT to improve our system.

Let’s start with customer sentiment.

What are things we can do to improve our customer and agent sentiment scores?

Next, we can explore changes we can make to improve response time.

What are things we can do to improve our response time, average handle time, and first contact resolution metric?

Lastly, we can dig into improving our CSAT and NPS scores.

What are things we can do to improve our CSAT and NPS scores?
💡 Tip: You might notice many of the recommendations to improve your various metrics overlap. These overlaps often are a good place to prioritize process improvement for the highest ROI, as they drive performance improvements across multiple metrics.

Get full access

✔️ All 100+ courses & tutorials in our catalog
✔️ New content added weekly
✔️ Private community access
✔️ No subscription, $150 paid once
✔️ Expense it using this template. Or get a team account.
✔️ 30-day refund policy. No questions asked
Join 3,107 learners from companies like Microsoft, Coca Cola, NBA, Adobe & Google

More tutorials like this

View all

If you scrolled this far, you must be a little interested...

Start learning ->

Join 3,107 professionals already learning