This article is always guaranteed to help answer the common question of "why did my agent say that?" CloseBot logs are available to everyone. This makes troubleshooting things like this possible.

Learn exactly how the bot decides what to say by hovering the info icon for the message in question within the contact conversation. This hover will show some info about the message and give two available links:
​

See Prompt Log: This link will bring you directly to the log that constructed the message. Sometimes a message is built from multiple actions, we try our best to land you in the most helpful area, with the introduction of the Agent action there will be many steps taken and we display these steps by the name of the tool used

Examine Action: This will bring up your actual build so you can find where the message in question came from.

Didn't behave the way you expected? Agent said something unexpected? You now know what action caused the message and what logs were related. From this you can create a plan for where adjustments can be made

In the conversation view, we also show "Quick Indicators" to describe actions the agent took. This includes booking appointments, solving objectives, resolving conditionals/switch actions, and activating custom scenarios. Use these where possible to help find where your flow maybe had a hiccup:

Here are some of the most common areas people mess up that causes undesired behavior. Check these out to make sure you're not making a no-no update to your objectives.

If your agent is messing up on an objective, start by taking a look at your short description. This is the most important field for your objectives and you should follow the best practices defined HERE.

Any time you ask WHY your agent said something, you'll be directed to this article. This is the exact method support will use if you ask them why your agent said something as well. We even have a t-shirt in our store with the CloseBot Classic article link... that's how popular this article is 🤣

Get familiar with clicking to see the conversation logs... it's not scary!