What's my plan?
Add-on AI agents - Advanced

Optimize the performance of your advanced AI agent by using analytics to identify opportunities for improvement and by building smarter dialogues.

This article covers the following topics:
  • Using analytics to improve the AI model
  • Using smarter dialogues to improve user experience

Using analytics to improve the AI model

Use analytics to learn where you can improve AI agent understanding and the deflection rate.

This section covers the following topics:
  • Improving deflection rate
  • Reviewing other key metrics

Improving deflection rate

Any conversation that doesn’t end in escalation is considered deflected. Review conversations that were not deflected and take steps to improve your deflection rate.

Read through conversation logs to see how well your dialogues work, looking for the following:
  • Broken dialogues: A user might break a dialogue by not using provided buttons, or by asking about their order number in the middle of the escalation flow. Consider including free text to guide users through the flow.
  • Misunderstanding instructions: Users generally don’t always read long messages. Consider making the message shorter and easier to follow.
  • Missing information: The user may lack key information to help them through the flow. In that case, think about what a live agent would do and add as much of that information in the dialogue as possible.
Take the following actions to improve deflection:
  • Adjust the default reply: Manage expectations and guide users through flows. See About system replies for advanced AI agents.
  • Use content coverage analysis: Identify potential new use cases that can be automated. See Performing content coverage analysis.
  • Use an API integration: Identify suitable use cases for an API integration to automate more conversations. See Preparing to create a custom integration.
After taking steps to improve your deflection rate, review resolution states as another way to improve your AI agent performance. Filter conversation logs by custom resolution to review escalated and unresolved conversations. Look for trends over time.

Reviewing other key metrics

In the left sidebar of the AI agent, go to Analytics > AI agent analytics for additional key metrics. To optimize the AI agent’s performance, focus on the following:
  • Recognized messages rate vs unrecognized messages rate
  • AI agent-handled rate
  • Escalation rate
  • Custom resolution rate
For more, see Analyzing advanced AI agent performance with the Performance summary dashboard.

Using smarter dialogues to improve user experience

Use the following tools to create smarter dialogues and improve your customers’ experience with the AI agent:
  • Backend integrations: Use backend integrations where possible to fetch data that the advanced AI agent can provide to users.
  • Conditional blocks: Use conditional blocks to jump into another flow based on certain keywords. This provides a more streamlined conversation before escalation if the user has been through the flow before.
  • Escalation templates: Streamline replies by managing the escalation process in one centralized template, rather than in each specific flow. Set operating hours to manage expectations and escalate appropriately.
  • A/B testing: Use A/B testing to optimize dialogue flows with data-driven decisions. See Performing A/B testing for advanced AI agents.
  • Confidence score: (Expression-based AI agents only) Use the native parameter confidence_score in a conditional block to provide a fallback in replies where the AI agent might be less confident.
    • For messaging AI agents, if the score is below 90%, the AI agent can confirm the intent another way. For example, “I want to make sure I understand correctly. You’ve forgotten your password and want to reset it. Is that correct?”
    • For email AI agents, if the AI agent is less confident about a topic, you might omit a reply and only trigger the actions.
Powered by Zendesk