Contact Lens for Amazon Connect (Preview)
Today, AWS announced Contact Lens for Amazon Connect, a set of capabilities for Amazon Connect enabled by machine learning (ML) that gives contact center supervisors and analysts the ability to understand the content, sentiment, and trends of their customer conversations to identify crucial customer feedback and improve customer experience. Amazon Connect is an omnichannel cloud contact center service based on the same technology that powers Amazon’s award-winning customer service. Companies like Intuit, GE Appliances, Capital One, and Dow Jones use Amazon Connect to run their contact centers at lower cost, while easily scaling to thousands of agents.
Contact centers field large volumes of customer conversations every day which results in millions of hours of recorded calls. Companies want to be able get accurate transcripts of these calls, and search across all of them to identify issues, common themes, and opportunities for agent coaching. They can use existing contact center analytics offerings, but these tools are expensive, slow in providing call transcripts, and lack required transcription accuracy. This makes it difficult to quickly detect customer issues and provide actionable performance feedback to their agents. The inability of existing tools to provide real-time analytics also prevents supervisors from identifying and helping frustrated customers on in-progress calls before they hang up. Given these challenges, some companies tackle this themselves by hiring data scientists and programmers to apply machine learning techniques and manage custom applications over time.
Contact Lens for Amazon Connect addresses these challenges and provides customers the ability to easily turn on a set of ML-powered analytics features as part of an out-of-the-box experience that empowers their contact center staff to easily use the power of machine learning with just a few clicks. Getting started with Contact Lens for Amazon Connect is easy. You can configure which calls to analyze in the contact flow settings by checking the Contact Lens for Amazon Connect option in the “Set Recording Behavior” block. Once this is completed, Contact Lens for Amazon Connect will start analyzing your specified calls automatically. You can click here to sign up for the preview.
Now let’s see how Contact Lens for Amazon Connect helps you to better analyze your customer conversations.
1. Enhanced Contact Search:
The contact search page in the Amazon Connect allowed users to search for historical contacts based on criteria such as date range, agent login, phone number, and queue. With Contact Lens for Amazon Connect, all your specified calls are automatically transcribed using state-of-the-art machine learning techniques, fed through a natural language processing engine to extract sentiment, and indexed for search. Thus, contact center supervisors and analysts can now use an out-of-the-box experience on the contact search page that enables them to identify contacts based on specific words and phrases that were mentioned by the customer and/or the agent during the call. Being able to search for contacts based on what was discussed in these calls allows organizations to perform a deep dive into issues that are impacting their customers. For example, an organization realized that their recently launched promo code for one of their products was not working. Being able to search for contacts where their customers mentioned this issue allowed them to understand the severity of the issue and diagnose the exact situation in which the promo code was failing.
In addition to being able to search based on the content of the conversations, organizations want an easy way to understand the quality of their customer interactions. Contact Lens for Amazon Connect analyzes the sentiment of the words being spoken by the customer and generates a score of -5 (most negative) to +5 (most positive) for various portions of the call (start, end, and entire call duration). The Contact Search page makes it easy for organizations to search contacts based on these scores so that they can identify the steps they can take to deliver a better customer care experience. Similarly, users can also search for contacts based on non-talk time (defined as silence plus hold time) that often helps identify new customer issues and agent training gaps. The Amazon Connect contact search page allows users to filter calls based on an absolute (in seconds) or relative (specified as a % of total call time) amount of non-talk time. See below for how the new contact search experience will look.
2. Enhanced Contact Trace Record (CTR) detail page:
The contact search page allows users to identify calls of interest. Users click on the search results to access a Contact Trace Record (CTR) detail page that allows users to see specific details of an individual contact such as contact id, queue name, agent name, phone number, and the associated call recording. With Contact Lens for Amazon Connect, users can view new details related to the call content and various conversation characteristics such as sentiment progression during the call. As shown in the screen shot below, the trendline at the top of the page shows how customer sentiment progressed during the call. Similarly, a breakdown of the customer sentiment and participation by speaker allows contact center supervisors and analysts to get a deeper understanding of the specific customer interaction. The turn by turn transcript and sentiment allows supervisors to quickly review the entire customer interaction and only listen to a specific part of the call recording to capture additional nuances of the discussion. This helps them save valuable time by not having to listen to the entire call recording and only focus on the portions that are important.
3. Automated Contact Categorization:
Contact Search allows users to conduct a deeper analysis for a recently discovered issue or explore historical calls. However, organizations also want to be able to automatically monitor 100% of their ongoing customer interactions for compliance with internal company policies and customer experience issues. For example, an organization may want to track all their contacts where customers mentioned one of their competitors. As another example, organizations want to measure whether their guidelines and best practices for interacting with customers are being followed. With Contact Lens for Amazon Connect, contact center supervisors and analysts can use a new categorization UI in in Amazon Connect to define their custom criteria based on keywords and phrases for organizing all their contacts for further analysis. Once a new category is defined, every new customer call is automatically evaluated against the specified criteria and the relevant categorization label is assigned if the criteria is met. The categorization labels are available as part of all the rich new metadata that is generated for every contact by Contact Lens for Amazon Connect. See below for the new UI for contact categorization.
4. Theme Detection (available in early 2020):
Contact search is an effective tool for doing a deep dive while the categorization feature helps automatically monitor conversations to spot issues that customers already know. However, organizations also want to automatically discover previously unknown issues based on what is going on in their contact centers. Contact Lens for Amazon Connect offers a new theme detection feature that applies machine learning to analyze multiple customer conversations (pre-defined/configured period) and automatically presents a set of themes along with their magnitude that indicate the top reasons for customer outreach directly in Amazon Connect. See below for a preview on how this feature will appear in Amazon Connect.
5. Real-time Supervisor analytics and alerting (available in mid-2020):
Today, supervisors in contact centers lack real-time visibility into how live customer calls are progressing. There may be situations that need their immediate attention. Some examples include a customer asking to speak with a supervisor, the agent repeatedly saying “I don’t know”, or an angry customer who wants to cancel her subscription. In many cases, understanding these issues after the customer hangs up does not help organizations to retain valuable customers and avoid costly churn. In mid-2020, Contact Lens for Amazon Connect will allow supervisors to get a real-time view of live customer interactions, including real-time customer sentiment progression and alerting based on pre-defined words and phrases. See below for a preview on how this feature will appear in Amazon Connect.
6. Open, flexible data:
Contact Lens for Amazon Connect helps organizations address their contact center analytics use cases by using the above features in the Amazon Connect console. However, we realize that customers want to tackle more use cases beyond these and need the underlying data. Contact Lens for Amazon Connect produces an output file in customers’ S3 bucket that contains metadata (such as transcriptions, sentiment, categorization labels, talk speed, and interruptions). Organizations can leverage this data in various existing systems and do not need to invest in new tooling. For example, you can use this data in a Business Intelligence tool, like Amazon QuickSight or Tableau, along with your CRM data to gain insights into customer engagements. Your data science teams can also use this data to create custom machine learning models with Amazon SageMaker.