How to view Conversational agent transcripts

This guide explains how to view Conversational agent transcripts after you open Conversational agent's back office. Filter transcripts and examine interaction patterns to understand how shoppers engage with Conversational agent.

Prerequisites

Permissions

Ensure you have the following permissions:

• Clarity User permission to view transcripts. This permission is necessary to access the Conversational agent application, which includes the transcripts.
• (Optional) Personal Data Viewer permission to view the customer filters. The permission is required to access the customer filter.

Knowledge

If you have the Personal Data Viewer permission, we recommend reviewing the filtering data article.

Access Conversational agent transcripts

1. Navigate to the Conversational agent > Conversations > Transcripts.

2. Review the tabular view showing all shopper interactions. The transcript table displays:

   • Last message received by the shopper
   • Agent name
   • Chat timestamp
   • Events associated with the interaction (purchase, cart update, product view)

3. Find the transcript you want to review. For quality analysis, each transcript shows:

   • Time of the chat

   • Detailed chat record

   • Events

   • Shopper cookie ID and transcript ID

     

4. Examine event attributes. The attributes vary by event type.
   
   Event types in transcripts: Transcripts contain two types of events that help you verify whether shoppers
   purchased items recommended by Conversational agent or understand the context of their interactions:

   • Browse behavior events show how shoppers interact with your website outside of Conversational agent. These events include purchase, cart update, product and category views, and checkout events.
   • Conversational agent events provide detailed information about interactions with Conversational agent beyond the text messages you see in the transcript. For example, when you click on a "trigger clicked" event, you can see:
     • The buttons Conversational agent presents to the shopper.
     • Whether the buttons were AI-generated.
     • The buttons the shopper clicked.

5. Click Customer profile to analyze a customer's profile. A list of all customers with their attributes and event details opens.

Filters

Understand filter types

The filters apply in the following order:

1. Customer filter

With the Personal Data Viewer permission, filter customer profiles by attributes. The system always applies this filter first and then returns all transcripts associated with the matched customers.

Filter customers with multiple transcripts

Optionally filter out customers who have multiple transcripts to focus on single-session interactions. Each time a shopper clicks Clear Chat, the system generates a new transcript ID. This results in multiple transcripts for that shopper.

To remove customers with multiple transcripts, click Apply filter to exclude users with 2 or more transcripts, then click Apply filters.

2. Transcript filter

After identifying the customer profiles, you can further filter transcripts by date range and conversational search.

Customer filters use case examples

Apply customer filters and use funnels to filter customer profiles that align with specific business goals. Here are common scenarios:

Product view after Conversational agent interaction

Business goal

Identify profiles where shoppers viewed a product detail page within 30 minutes of interacting with Conversational agent. This helps you understand whether Conversational agent successfully drives product discovery.

Build filter

1. Add the clarity event and set the action attribute to user_respond. This identifies customers who sent a message to Conversational agent.

2. Add the view_item event. This identifies customers who viewed a product detail page.

3. Set a 30-minute completion window. This ensures you only see customers who viewed the product within 30 minutes of their Conversational agent interaction.

   

Cart abandonment analysis

Business goal

Find profiles where shoppers added items to their cart shortly after interacting with Conversational agent, but this didn’t result in an attributed purchase. This helps you identify potential friction points in the checkout process.

Build filter

Add to cart

1. Add the clarity event and set the action attribute to user_respond. This identifies customers who interacted with Conversational agent.
2. Add the cart_update event, set the action attribute to add, and specify a 30-minute funnel constraint. This finds customers who added items to their cart within 30 minutes of the Conversational agent interaction.

Cart abandonment

1. Add the clarity event and set the action attribute to user_respond.

2. Add the purchase event.

3. Specify the negative condition (not matching) that corresponds to the cart abandonment. This excludes customers who completed a purchase.

4. Specify a 7-day funnel constraint to attribute purchases within this window.

   

Save and reuse filters

You can save and reuse filters as templates for quick access. Bloomreach recommends saving filters you expect to use frequently.

Filter transcripts by date range

To filter only transcripts where conversations occurred within the selected date range:

1. Click the date range selector.

2. Choose your start and end dates.

3. Click Apply filters to view transcripts from that period.

   

Filter transcripts with conversational search

Find transcripts using the conversational context. This filter searches broadly using full-text matching and returns either agent or user messages that contain the defined keywords.

To filter transcripts using keywords:

1. Enter one or more words in the conversational search bar.

   • Use “ ” quotes to match exact phrases and “-” to exclude terms.
   • Search is insensitive to case and accent marks.
   • Special characters work as separators and don’t match exactly.

2. Click Apply filters.

   

Manage transcripts

To ensure GDPR compliance and manage your transcripts:

• Retrieve and permanently delete their Conversational agent transcripts.
• Set transcripts to expire automatically after a retention period.

Contact your Product Engagement Manager to help with the above.

PII in transcripts

Conversational agent doesn't ask customers for personally identifiable information (PII), but PII may be shared by customers via chat and stored in transcripts. To minimize risks associated with storing PII, follow the data storage and expiration guidelines.