ILLUMINATE: CallMiner’s New AI-based Search Tool

By Mingren Xiang, Allen Shapiro, Ronald Mueller

One new feature in the Call Miner Eureka Engagement technology platform issued in 2020 is Illuminate. It is an AI-driven search capability that can greatly assist analysts to update and fine tune their search words and logic for setting up categories in their scorecards, for example. There are currently 3 areas in the Call Miner Platform where you can find the Illuminate icon and can use this new function, including the Analyze component. This capability is built directly into the Eureka platform, so it is available to use as soon as you start to use Eureka. The AI-algorithms underlying Illuminate are all pre-trained and ready to go.

In this paper, we briefly explain how to use Illuminate and what is its intended value. We describe the 3 sections of Eureka where Illuminate can currently be used, and how it greatly helps in each of these areas.

The main portion of this short technical paper is to show a real-life example of how it can be used and is helpful in the Analyze component of the Eureka Engagement platform. We show its benefits both from the point of view of accuracy and completeness of analysis and also in terms of the significant performance boost it provides to analysts.

We are now using Illuminate in all of our client implementations of the Call Miner Eureka Engagement platform. In our view, it embodies outstanding, practical and easy-to-use AI-based functionality.[1]

Background

Macrosoft is a technology support partner of Call Miner. In contrast to many of the other partners of Call Miner, our sole focus is to provide technical support and maintenance to Call Miner’s new clients and to maintain and expand support where needed for existing Call Miner clients. We do not sell or recommend other Contact Center technology.

To fulfill our intended mission as responsively as possible, we continue to invest significantly in knowledge building of our staff and in researching and getting hands-on experience with all newly released Call Miner features, so that we can incorporate them quickly into our technology support practice for our clients. Our objective as a Call Miner partner is to provide state-of-art technology services to our Call Miner clients. Given the tremendously fast pace of innovation of the Call Miner system, we believe this to be a critically important responsibility for us as a Call Miner’s technology support partner.

This paper deals with the new AI-based search functionality, Illuminate, that is built-in to the Call Miner Eureka platform. We will be issuing other papers in this same vein relating to other new features of Call Miner on an ongoing basis.

What is Illuminate

In July 2020, CallMiner released Illuminate™, its new AI-driven search feature that makes it quick and easy for organizations to discover, extract and act on insights from voice and text interactions with customers.

Categorizing is the name of the analytics game in Call Miner. The meaning of how customers and contact center agents express themselves is fundamental to CallMiner’s ability to accurately automate analytics. The Illuminate feature within Analyze is intended to speed category customization by uncovering related words and phrases and presenting them to the user for consideration and inclusion in the category customization.

This process can be repeated again and again within Illuminate, persistently fueling a dynamic discovery process that empowers Analyze users to search for what matters most with a single click. Both expected and unanticipated associations for intent, action, and emotion are surfaced with the user-friendly ability to single click to add to a search and return more results.

Based on keywords provided by the user, Illuminate utilizes the pre-built and pre-trained AI algorithms within Eureka to identify other words and phrases mined from the customer’s own interaction data that users might consider including when they analyze and categorize a given customer topic or issue. That is the essence of Illuminate.

One point that needs to be emphasized, this is much more than simply a word association game – that is, to find other words or phrases in the customer’s conversation data that similar to the word used in the search. When used iteratively (as we show below) Illuminate can actually be very helpful in understanding topics and key ideas embedded in the conversations that might not be known or fully appreciated by the analyst. 

This capability is reminiscent of the drill-down capability available within advanced data analytics BI platforms when analyst tries to understand deep down the reason for some sharp change in aggregate data. However, as opposed to this BI example, the Illuminate capability is much more useful and long lasting in the sense that, as you go thru various levels of word searching using Illuminate, you can easily incorporate some or all of the related words presented to you by Illuminate to your category definition. Thus, going forward they will be included in the category definition, to make it a better measure of the topic of interest.

What is Behind Illuminate

Illuminate applies unsupervised machine learning to voice and text-based interactions to comprehensively reveal meaning for users of Eureka Analyze. Out of the box, Illuminate has already been widely trained using a good portion of the voluminous amount of speech and text data available to Call Miner. So it will work well on Day 1, that is pre-training is already in place when you start using the capability.

In addition, as user conversational data begins to build up, the Illuminate feature will update its algorithm training to take into account this user-specific conversational data to fine tune the ML models for even better and more specific performance on your own data.

Once this local training has occurred, Illuminate no longer relies solely on generalized models and generic suggestions, it will include the fine-tuning as a result of the model training based on including the user’s own conversational data.

Download article as PDF

You can now download the entire article as a PDF for free and use for future reference. Click the button and get the PDF version

Illuminate Found in both Search[2] and Category Builder

Currently, the Illuminate icon and functionality can be found and used in both Search and Category Builder in the Eureka Analyze platform, as shown in the chart below. The capability is simple and intuitive to use, so little or no training is needed to begin using it.

However, it is important to practice and to learn to use it well and with a purpose beyond simple wordsmithing. At an advanced level, you can learn so much from your conversational dataset, in an interactive fashion from its feedback at every level of your analysis. It is the ultimate drill-down capability that all of us from the BI-world have longed for so long. As we demonstrate in the example case study in the next section, that is when you see the real magic of Illuminate!

CallMiner Analyze  Components Where Illuminate is FoundUser Screenshot with Illuminate icon highlighted  
  Search        CallMiner Search
    Category Builder        CallMiner Category 

The basic functionality of Illuminate is the same in both Search and Category builder, and we illustrate that functionality in the next section with a case study of using illuminate when doing search query in the Analyze platform

Case Study of Using Illuminate in Analyze Component

We provide a simple case example of using Illuminate to learn more about a specific topic in the user’s conversational dataset. In this example, we are using a publicly available conversational dataset.

  • Step 1. The user starts using Illuminate by inputting a single word search. Let’s start with the word “return”. Illuminate is user friendly as it works with a search phrase as simple as one word. This is particularly useful for non-analyst users such as executives and program managers who do not have advanced search skills. Below are the suggested terms and phrases that illuminate returned after just a few clicks:

Based on the suggestions presented by Illuminate, and without any prior knowledge of this conversation dataset, the user can quickly spot two main topics here: The first being processes related to return authorization with the phrases such as “non-returnable”, “return policy”, “return instruction” and “return authorization”. The second topic is voice mail related to phrases like “return your call” and “leave a message”. Again, suggestions provided by Illuminate are based on the ML model your local conversation dataset. So, these two topics are happening in your customer interactions.

  • Step 2. Now, the user may want to dig deeper into the process of return authorization in the organization. Instead of listening manually through thousands of calls and coming up with search terms related to this topic, Illuminate gives the user control to narrow the list of suggestions, essentially giving the user the ability to focus on a certain topic.

The chart below shows the results Illuminate returns after the user applies the narrow list of 6 terms related to returns, as shown. We see more suggestions around return authorization such as “proof of delivery” and “return department”. The user can create a narrow list of terms ranging from 1 or 2 terms up to any practical limit. And for each such list of narrow terms input by the user, Illuminate will return its list of relevant terms.

  • Step 3. The user can now decide to add some or all of the suggested phrases and terms returned by Illuminate to make the search more focused and robust by simply clicking on each of the terms they want to be added, as shown in the chart below. Illuminate will then populate the search query with the selected phrases. Below is an example of adding 5 terms to the simple search for ‘return’.
  • Step 4 and ongoing. The power of Illuminate does not stop here. Now that we have a very comprehensive search around ‘return authorization’, the user can apply Illuminate again with these new terms, thus continuing to recursively dig into the conversational dataset around this topic with ease. This type of recursive search can continue until the user is satisfied they have asked out all the knowledge available in the conversational dataset on a particular topic or word.

And of course, they can repeat the same process for as many terms or words of interest. So hopefully this makes clear how powerful a tool Illuminate is in allowing a user to mine out all the meaning within the conversational dataset!

Wrap up: Illuminate advantage

Illuminate allows users to pinpoint the exact language and concepts they want to monitor in customer conversations. In addition to saving the language as a search or category, users can create a real-time alert to be triggered in Eureka Alert for in-the-moment tracking and reminders to agents.

Customer interactions across phone, email, and chat represent millions of words each day to analyze. Illuminate provides an easy and effective way to quickly identify trends in your interactions and provide suggestions on what to explore further. Illuminate provides an exceptionally powerful root cause analysis for common themes in your customer engagements. One-click search building also empowers non-analyst users such as executives and program managers to easily conduct searches with meaning and results without requiring advanced search syntax skills.

As Bruce McMahon, Vice President of Product at CallMiner said, “Illuminate shines a bright light on powerful customer insights.  We’re simplifying the process for analyzing customer and brand interactions, and in turn, making real-time agent guidance and strategic business-process transformation faster and more comprehensive. This new AI-driven intelligence expands the ability for our customers to uncover previously-hidden insights and leverage the findings to enhance customer experience, sales, marketing, operations, contact center performance and more.”


Download article as PDF

You can now download the entire article as a PDF for free and use for future reference. Click the button and get the PDF version

[1] We see this feature as a practical example of bringing AI into your enterprise without all the ‘fuss and muss’ of AI ops, that is being built and tested internally to your company using one of the Augmented AI platforms. The algorithms are already set, and have undergone a wide range of training and testing, so it is useful from Day 1 of your use of the Call Miner Platform. Macrosoft is continually seeking practical AI implementations of this type in all our client work.

[2] Illuminate allows users to pinpoint the exact language and concepts they want to monitor in customer conversations. In addition to saving the language as a search or category, users can create a real-time alert to be triggered in Eureka Alert for in the moment tracking and reminders to agents.

Share this:

By Mingren Xiang, Allen Shapiro, Ronald Mueller | November 6th, 2020 | CallMiner

About the Author

Mingren Xiang

Mingren Xiang

Mingren is Macrosoft's technical lead in voice and conversational analytics using the CallMiner suite of utilities. He has a Bachelor of Science from University of Wisconsin-Milwaukee.

Allen Shapiro, Director – CCM Practice

Allen Shapiro

Allen brings more than 25 years of diverse experience in Marketing and Vendor Management to Macrosoft Inc. As the Managing Director of our Customer Communications Management (CCM) practice, Allen leads the Onshore and Off-shore CCM development teams. Additionally, Allen oversees pre-sales activities and is responsible for managing the relationship with our CCM software provider Quadient.

Dr. Ronald Mueller CEO of Macrosoft

Ronald Mueller

Ron is CEO and Founder of Macrosoft, Inc. He heads up all company strategic activities and directs day-to-day work of the Leadership Team at Macrosoft. As Macrosoft’s Chief Scientist, Ron defines and structures Macrosoft’s path forward. Ron's focus on new technologies and products, such as Cloud, Big Data, and AI/ML/WFP. Ron has a Ph.D. in Theoretical Physics from New York University and worked in physics for over a decade at Yale University, The Fusion Energy Institute in Princeton, New Jersey, and at Argonne National Laboratory.

Ron also worked at Bell Laboratories in Murray Hill, New Jersey., where he managed a group on Big Data. Ron's work focused around the early work on neural networks. Ron has a career-long passion in ultra-large-scale data processing and analysis including predictive analytics, data mining, machine learning and deep learning.

Recent Blogs

CallMiner Call Fidelity Case Study
Read Blog
How AI, ML, and NLP contribute to the CCM Revolution
Read Blog
How to Virtualize your VFP Application
Read Blog
Identifying Benefits and Calculating ROI of CallMiner Investment
Read Blog
TOP