Reviewing Analytics on the Dashboard

Broccoli tracks the most common reasons calls are unbooked. You can view these categories directly in your dashboard. We recommend starting with the categories that show the highest number of unbooked reasons and applying the best practices below.

Based on current data, a booking rate above 70% is where you should be. Anything lower should consider recommendations below.

If Agent Requested is high:

A live agent can be requested when the AI agent has difficulty understanding the customer. This may happen when:

There is significant background noise
The customer speaks unusually fast or slow
The caller is unsure of what the AI can or cannot do

Adjusting AI Responsiveness

If you notice a pattern in how your customers communicate, our team can help fine-tune your agent’s interruption sensitivity and response timing. For example, in regions where customers tend to speak more slowly, we can adjust the AI’s responsiveness to allow more time before it replies.

As a best practice, allow at least 30 days of call activity before making adjustments. This provides enough data to identify overall tendencies and avoid changing settings based on just a few outlier calls.

Common Live Agent Requests

The most common reason customers request a live agent is that they don’t want to talk to a “machine” or are surprised the voice was not a real person. Two effective approaches to manage this are:

Adjust the greeting to mention that an AI is answering. This sets expectations and often encourages callers to speak more clearly.
Use the live transfer feature to route calls directly to a designated number when a live representative is requested. (Avoid transferring to your main number to prevent call loops.)

If Estimate on Call is high:

This challenge is similar to what human CSRs encounter, and you should consider the approaches your own team currently uses to address it. Your AI Agent can be adjusted to follow the same process and help overcome objections.

By default, the agent follows a script found in the FAQ section:

“Q: If the user calls in and wants to get an estimate on call”

You can edit this script to reflect what your CSRs typically say, or adjust it to match the process you want your AI Agent to follow.

If Timing Issue is high:

Unbooked calls related to timing naturally increase during busy seasons when availability is limited. Monitoring seasonal patterns will help set realistic expectations.

If Dispatch Fee Issue is high:

Some customers may object to the dispatch fee. What do your current CSRs do?

One suggestion would be to consider waiving the dispatch fee if the customer disagree's with it. To do this, you can add a prompt to your Dispatch Fee Value Build (Agent Config > AI CSR > Call Flow > Dispatch Fee Value Build):