Skip to content
By using AWS re:Post, you agree to the AWS re:Post Terms of Use
AWS re:Postre:Post
About re:Post

Bulk Review for Amazon Connect

0

I need to analyze a large volume (5,000+) of calls in Amazon Connect to identify any mentions of a specific keyword(s) and understand common themes or patterns from those specific calls. Rather than listening to each call individually, I'm looking for a way to automatically search through call transcripts for this keyword(s) and then generate a summary or analysis of the main topics and trends from just those filtered calls. This would help identify recurring issues or patterns much more efficiently than our current manual review process.

For example, if 500 calls mention "Device 1," I want to quickly understand what those conversations were typically about, what issues were commonly reported, and any patterns in customer concerns. This should happen all without having to listen to each individual call. This would help us better understand and address customer issues related to this specific topic, while significantly reducing the time spent on manual call review.

Tableau isn't going to help because it can't help search through actual call recordings or analyzing conversation content. It can only visualize data that's already been analyzed and structured. I need the initial work of understanding what's being said in these calls.

Is this functionality currently possible within Amazon Connect, or would it require additional features or tools to accomplish this goal?

  • dmacias
    a month ago

    You already have the call recordings, right? If so, I would not use Amazon Connect for this, Connect can do this, but the calls need to happen and then they are analyzed. You're going to have to roll your own, some ideas:

    • Put everything in a bucket.
    • Amazon transcribe (each recording).
    • Run every transcription via Bedrock or some other analytics tool to look for specific keywords.
Topics
Application IntegrationAnalyticsMachine Learning & AIBusiness Applications
Tags
Amazon ConnectAnalytics
Language
English
asked a month ago78 views
2 Answers
1

You could try the post call analytics solution:

https://aws.amazon.com/blogs/machine-learning/post-call-analytics-for-your-contact-center-with-amazon-language-ai-services/

It allows you to decide which calls to process, has a bulk insertion process and each call is transcribed with transcribe and analyzed by bedrock so you can get insights into each call. Its not exactly what you are looking for but it could be customized to answer prompts regarding the totality of the calls and look for different word patterns. it also has sentiment analysis so that could also help with what you want about what was happening when the word was mentioned etc.

AWS
answered a month ago
AWS
EXPERT
reviewed a month ago
0

Amazon Connect Contact Lens provides the functionality you need to analyze your large volume of calls efficiently. Here's how you can accomplish your goal:

First, Contact Lens offers conversational analytics that automatically transcribes and analyzes customer-agent interactions. You can search through these transcribed recordings using specific keywords, which addresses your need to find all calls mentioning "Device 1" or other specific terms.

To search for conversations containing your keywords:

  1. Log in to Amazon Connect with an account that has "Contact search" and "Contact Lens - conversational analytics" permissions
  2. Navigate to Analytics and optimization > Contact search
  3. Use the search filters to narrow down your results, including the specific keywords you're interested in
  4. You can search by speaker (agent or customer), keywords, sentiment score, and other criteria

For analyzing themes and patterns across these filtered calls, you have a couple of options:

Contact Lens provides post-call analytics that helps identify trends in customer conversations. This allows you to understand common issues without manually reviewing each call.

For more advanced analysis of themes and patterns, you could leverage Amazon Bedrock and Amazon Q in QuickSight. This approach has been implemented by other organizations to dramatically reduce analysis time from weeks to minutes. The solution would:

  1. Generate call metadata (summaries, root causes, next steps) from transcripts
  2. Store this structured data for querying
  3. Allow you to use natural language queries to identify themes and trends across all calls mentioning your keywords

This functionality exists within the AWS ecosystem, though the more advanced theme analysis might require integration of Contact Lens with additional AWS services like Amazon Bedrock and Amazon Q in QuickSight.
Sources
Analyze conversations using conversational analytics in Amazon Connect Contact Lens - Amazon Connect
Review analyzed conversations using Contact Lens - Amazon Connect
Asure’s approach to enhancing their call center experience using generative AI and Amazon Q in QuickSight | AWS Machine Learning Blog
Search conversations analyzed by Contact Lens - Amazon Connect

answered a month ago
  • austinwalks
    a month ago

    Yeah I'm looking for a more "bulk" or "full" analysis of these thousands of calls.

    I like the summarization of each call individually, but I'm looking for a way to get a summarization (with trends, reasons, etc) for thousands of calls at a time. Can bedrock help with that?

You are not logged in. Log in to post an answer.

A good answer clearly answers the question and provides constructive feedback and encourages professional growth in the question asker.

Relevant content