Engage your customers with Proactive Conversations
Overview
Proactive Conversations enable you to configure your AI agent to reach out at the right moment to assist customers, resolve issues, and enhance the customer experience. Once the customer engages with the conversation, you can also use the context to trigger a relevant Process. Learn more about Processes.
Example Use Cases
- For a frequently occurring question, such as how to check an order status, you can configure a proactive message, such as "Need help checking your order status?" to show up on your 'Contact Us' page.
- If you have a system outage, you can notify customers as soon as they navigate to your webpage. They can then engage with the AI agent to ask for help.
- When a new user logs in for the first time, the AI agent can proactively offer to share more about the services your business offers.
How Proactive Conversations Work
There are two ways to trigger a Proactive Conversation:
- URL Matching – The Proactive Conversation is triggered based on the page URL.
- Embed2 Javascript API – You can programmatically trigger a Proactive Conversation and optionally pass in parameters to personalize the message.
When you trigger a conversation proactively, you may want to disable any greeting message that you have already configured.
Create a Proactive Conversation with a static message
- On the Ada Dashboard, go to Training > Proactives.
- Click New Proactive.
- Give your Proactive a name.
- Choose Use a static message.
- Craft out your message in the given text box.
- If you want to trigger through URL matching, add Trigger conditions for your proactive:
- Click +Add.
- Select the appropriate URL matching option from the drop-down menu. To learn more about URL matching, see the table below.
- Input the full URL or URL component.
- Repeat for additional URLs where you want the proactive to be triggered.
- You can choose whether the proactive should trigger for every page load, or once per session.
Option | Definition | Example |
---|---|---|
Exactly Matches | This option will look for the exact match of your URL. Query parameters will not be part of the matching logic. | Specifying https://www.example.com/checkout in the configuration will match to https://www.example.com/checkout?q=hello and https://www.example.com/checkout/ |
Contains | This option will match URLs that contain your specified string anywhere in the URL. Query parameters containing your values will also be included in the matching. | Specifying example in the configuration will match to https://www.example.com/checkout and https://www.hello.com?q=example |
Ends With | This option will match URLs ending with your specified string. Query parameters will be ignored for this option. | Specifying /checkout in the configuration will match to https://www.example.com/checkout and https://www.hello.com/checkout?q=example |
Regular Expression | This option gives you full control in specifying the matching URLs using the regular expression pattern matching language over the full URL, including query parameters | Specifying product[0-8]+[^9]$ will match any page that includes the string 'product' and any number of any length afterwards, as long as the number 9 never appears. Specifying ^https://acme.com(?!./(promo|pricing)$).*$ will match on all https://acme.com pages except for those ending with /promo or /pricing. |
- Now that you have your Trigger Conditions, you can choose whether the Proactive should start immediately after page load, or if it should show after a delay. You can choose the delay in seconds.
- You can also configure how long the proactive should be displayed, if the customer does not click on it. You can opt to show it indefinitely, or dismiss it after a set time in seconds.
Create a Proactive Conversation with a message template
If you want to personalize the content of your Proactive Conversation, you can use a message template. This allows you to pass in dynamic content when you trigger the proactive through the Embed2 Javascript API.
- On the Ada Dashboard, go to Training > Proactives
- Click New Proactive
- Give your Proactive a name.
- Choose Use a message template.
- In the given text box, you can type out the content of your message, using double curly brackets for placeholders:
{{placeholder_name}}
- As a message with placeholders cannot be triggered through URL matching, the configuration options for URL matching will be disabled. You can trigger this message using your Embed2 Javascript API.
How to configure a Greeting so it doesn't show on a Proactive Conversation
To ensure that your Greeting does not show up in addition to the Proactive message when a customer clicks to chat to the agent, you will need to use the proactive_conversation
metavariable. This metavariable becomes available in the Greeting block builder once you have at least one conversation with a triggered proactive.
- On the Ada Dashboard, go to AI Agent Profile > Greeting. The Greeting page opens.
- Add a conditional to only show the Greeting if the metavariable
proactive_conversation
Is Not Set. You can also customize your Greeting using the same variable. The value of the metavariable will be themessageKey
of the latest proactive delivered to the user.
Limitations
- Ada Chat is currently available only on web or mobile web.
- No conversation in Ada is created unless the user clicks on the widget to open the chat, so unless a user clicks on the widget, a Proactive Conversation will not be visible in the Conversations View.