We are at the dawn of the age of Artificial Intelligence (AI). AI is being discussed everywhere, by nearly everyone. AI is writing college papers, giving interviews, and is being viewed with both wonder and fear as the eventual replacement for most human professions.
We’re really excited about the possibilities at Authenticx, as we believe that AI is foundational to improving the lives of humans. AI can be a “canary in the coalmine”: surfacing issues that may fall through the cracks at any significant level of scale.
The potential is huge, particularly in healthcare, but AI in and of itself is not a one-size-fits-all solution when it comes to solving business problems.
While AI is starting to take on more creative, subjective roles in healthcare, there are truly transformative issues requiring change that ChatGPT cannot solve for. At its most valuable, AI in healthcare (and really any industry) can be leveraged to uncover and aggregate signals that can give us, humans, clear direction.
Training AI Built for Healthcare
To begin identifying and flagging these signals at scale, the AI must be trained on what to look for.
AI is built on training data: which is data that teaches a computer/AI to recognize certain patterns that exist in the data set. This allows the AI model to identify those patterns on a much larger scale when applied to new, non-training data.
These patterns, or signals, are generally hand selected by humans or trained by data labelers. Many organizations rely on crowdsourcing or offshore labeling teams to help them source the training data they use to train the AI. Understanding what a productive, valuable signal is can be extremely difficult.
Authenticx started as a human-powered operation in 2018: team members with healthcare and social work experience listening to calls and tagging them. What those team members noticed was a similar theme appearing across all conversations.
Patients were running into recurring issues that made them feel stuck in a particular problem:
- “I’ve called back three times…”
- “How is this the first time I am hearing about an outstanding balance…”
- “You are the third person I have been transferred to…”
These problems, and the disruption they caused in the customer journey, mimicked a natural phenomenon known as an “eddy”.
An eddy refers to water flowing downstream that encounters an obstacle which interrupts the journey and creates a swirling effect, essentially resulting in the water being “stuck.”
Customer problems, like the ones outlined above, were experiencing “eddies” in their customer journey. They were issues and disruptions that caused the caller to feel stuck, and ultimately prevented forward progress through an expected journey. Authenticx referred to the presences of obstacles (and their consequences) as The Eddy Effect.
AI Empowers Healthcare Organizations to Dive Deeper into Conversations
In the early days of Authenticx, the insights we discovered proved to be extremely valuable to our customers, but the problem was that the insights surfaced only represented 1% of all conversations happening.
AI became the key driver to help us scale our efforts in replicating what we were learning for significantly more conversations. We started building a deep learning AI model for healthcare that would identify and aggregate portions of conversations where we flagged a customer feeling “stuck”.
We used this training data to build the first iteration of the Eddy Effect algorithm: which has become a core signal for healthcare organizations trying to create better business outcomes and improved customer experiences.
Without doing our initial research and being careful about what use cases we wanted to tackle with AI for healthcare, we would not have identified the right training data to build into our platform. Getting too caught up on KPIs (such as average call length or first call resolution rate) can lead to your organization ignoring the larger problems laying just beneath the surface.
What does average call length tell you about what steps you need to take to fix a call guide?
What does first call resolution rate tell you about how to prevent the call in the first place?
Data without direction cannot help you fundamentally identify transformative opportunities, validate decisions, or take actionable next steps.
The more we dug into the conversational data that our healthcare clients delivered to us, the richer insights we were able to unlock by scaling the Eddy Effect: correlating the presence of ‘Eddies’ to topics of conversation that surfaced major issues.
When Customer Problems Become Business Problems
Our team started to track these types of statements in conjunction with the beginning and ending sentiment on calls, return callers, agent performance, and a host of other factors.
The correlation between the presence of conversation-based obstacles and identification of real operational inefficiencies became apparent:
- Confusing instructions on billing statements were the source of a 20% increase in calls to the billing line for a healthcare provider.
- Failure to follow call guides regarding the collection of personal information was hurting HIPAA compliance rates for nurse triage lines.
- Timely delivery of prescriptions accounted for a significant portion of program adherence failures
The Critical Importance of Listening at Scale
We believe that the best way to identify the root cause of a customer problem is to leverage the conversations found in your contact and support centers.
These conversations are unsolicited, unfiltered, and unrestricted. Unlike survey responses that ask a person to reflect on an experience that happened in the past, or only provide feedback on specific topics put forth by your organization –conversations are organic. They are driven by the customer, and thus highlight the customer journey and firsthand experience of a process or product from their point of view.
Prioritizing these conversations ensures that your organization is not simply collecting data, but uncovering the source of poor experiences, negative outcomes and unnecessary expenses.
How Can Authenticx Automate Capturing Context to Identify Disruption?
The Eddy Effect is the perfect inflection point to signal that customer problems had become business problems. Eddies are almost always indicative of process, technology, or product issues that need to be addressed.
Diving into calls that contained these disruptions allows leaders at healthcare organizations to cut through the millions of conversations they were having with patients or customers and identify significant issues, all while preserving context on what caused the pain.
This allows them to make data-driven decisions on how to strategically invest resources to address and correct the root cause of their biggest disruption, eliminating guesswork and providing a faster path to resolution.
To ensure that all the feedback provided in these unsolicited conversations was captured and considered, identification of the Eddy Effect needed to be executed at scale.
Authenticx models can identify Eddy Effect obstacles because they can capture and categorize the context provided around certain key terms or phrases.
Leveraging Natural Language Understanding (NLU) models and Large Language Learning Models (LLMs), along with Deep Learning, Authenticx is able to analyze complex patterns within datasets and go beyond just flagging key terms.
Training NLU and LLM models to uncover unseen patterns not only improves the overall accuracy of the models, but actually gets you as close as possible to capturing human level insights at scale.
Take Action with Authenticx
Authenticx started as a human-powered enterprise and continues to prioritize the partnership of humans and AI. Nothing can replace listening, but AI can help enhance listening so that your team can focus on taking action. Think of AI as a means to identify signals of disruption or opportunity that can have enormous impact on your business strategy.
The Authenticx platform automates data correlation to point to causation. We identify disruptive conversations, then dive into the topics and themes that drove those obstacles in the first place.
Our platform surfaces connections, for example: the mention of phrases like “billing confusion” and “website portal” to pinpoint the presence of the Eddy Effect. The Eddy Effect Model (and all proprietary Authenticx models) provide clear signals that point to opportunities for improvement across the healthcare industry, and the aggregated data to support transformative action.
Intelligent, immersive, and personalized data-storytelling.
Leverage conversational intelligence to capture all customer interactions in one platform.
Authenticx was founded to analyze and activate customer interaction data at scale. Why? We wanted to reveal transformational opportunities in healthcare. We are on a mission to help humans understand humans. With a combined 100+ years of leadership experience in pharma, payer, and healthcare organizations, we know first-hand the challenges and opportunities that our clients face because we’ve been in your shoes. In 2023, Authenticx was ranked No. 349 on the Inc. 5000 recognized as one of America and Indiana’s fastest-growing private companies.
Want to learn more? Contact us!