2023 was the year of Generative AI (Gen AI). There were significant levels of investment in advancing machine learning and deep learning technologies resulting in ChatGPT, Bard, GitHub Copilot, Dall-E, and Gen AI (artificial intelligence) tools getting mainstream popularity.
These GenAI technologies stand to deliver billions of dollars in productivity, efficiency, and revenue enhancement gains to companies and individuals. But every technological paradigm shift has a dark shadow. In this case, it’s deepfakes.
Previously, deepfakes were relegated to the realm of researchers and academics since the cost of creating those as well as technological burden was too high.
But now with GenAI tools it is possible for anyone to create a voice clone and a realistic chatbot for a monthly subscription of a few dollars. This has resulted in a worrisome rise in deepfakes all over the world.
To illustrate, a KPMG study reveals that the number of deepfakes available online is increasing by 900% annually. Pindrop’s own research, our Deepfake and Voice Clone Consumer Sentiment Report, found out that more than 90% of U.S. consumers have concerns about deepfakes.
From a contact center perspective, deepfakes, especially audio deepfakes such as voice clones, no doubt pose a major security risk. But they have several implications for customer experience (CX). The fundamental problem created by deepfakes is to break all trust in customer interactions.
- How can contact center agents know for sure whether the voice on the phone is a real human being or a clone?
- How can they trust the interaction if they don’t even know who they are interacting with?
- How can they provide a positive experience to customers while keeping malicious actors out?
To answer these questions, we need to understand the stages of customer satisfaction, how deepfakes have compromised the process, and what measures should be in place to improve CX.
Customer Satisfaction
Customer satisfaction is still the goal of contact centers. And the primary way to do that is to answer calls from customers.
Contact centers have come a long way in the last four years after customer calling behavior changed rapidly after the COVID-19 pandemic, work-from-home became the norm, and call volumes surged. Many operations are now better prepared to handle these impacts. Some have also enhanced security measures to combat fraud, both in the contact center and in the IVR.
These are important steps to close the gap between surviving and thriving. But to fully cross the chasm, contact centers should aim for a true balance between the intuitive, seamless experiences they provide to customers and the enhanced security measures implemented to protect them.
While revamping the entire CX system is beyond the scope of our conversation here, we can consider ways to improve the process and make it more secure.
But in order to do so, it’s vital that we identify the areas of influence during a phone call that the current system ignores. We will refer to them as the “micro-experiences” that contribute to the customer’s overall impression.
There are three distinguishable categories of micro-experience: Pre-call/On Hold, Authentication, and Post-Auth. By dissecting these micro-experiences, we’ll better understand where there may be room for improvement.
1. Pre-call/On Hold
The first micro-experience is the hold phase, where the caller is awaiting a response from an agent or IVR. Hold times can be significantly impacted by the ability of customers to solve their own issues with an IVR, the number of IVR options they are given, and the number of agents available.
If any of these factors are limited, wait times can be extended across the board. However, the ability to automatically identify a caller in the IVR can help to facilitate self-service and open up IVR menu options, both of which can reduce the burden on live agents.
But to do this with confidence, the contact center needs to know that there is a real human at the other end of the line and not an AI-generated chatbot, a recording, or a voice clone. The way to regain that confidence is to perform a liveness check on the call, much like a website performs a Captcha assessment.
A liveness score that passes above a certain threshold can inform the contact center that (a) there is a real human voice on the call and not a synthetic voice and (b) the caller can be allowed to access self-service IVR menu options without exposing the organization to fraud risks.
The liveness check can be bolstered by utilizing other tools such as caller ID validation, spoof risk assessment, and call origination analysis to bolster the confidence in the authenticity of the caller.
If automatic identification is a way to improve the On Hold portion of the CX, then Authentication (the next micro-experience) becomes vitally important.
2. Authentication
Authentication is the process by which a business confirms the identity of an individual. The question here is “are we talking to the right human?” To answer this question, contact centers rely on a frustrating process of asking customers to spell out basic information that the business should already have: like their names, addresses, or account numbers.
To the customer, this process can understandably feel like a waste of time. Making matters worse, authentication can also include answering knowledge-based answers to questions or providing one-time passcodes (KBAs and OTPs).
It’s no guarantee that customers will remember their own answers, and more frequently, criminals do. In fact, according to a sample case study in our 2023 Voice Intelligence and Security Report, up to 80% of fraudsters were able to pass KBAs while genuine customers struggled.
...the contact center needs to know that there is a real human at the other end of the line...
Removing active authentication processes, like KBAs, in favor of passive ones, like phone number validation, can provide a boost to CX in ways that do not compromise security.
However, the modern threat landscape is expanding with new threats like synthetic voice, IVR trawling, data mining, and social engineering that require more comprehensive solutions. In fact, a Forrester study we commissioned found that 76% of fraudsters use the IVR to deploy brute force attacks to extract account data: and use the information for fraudulent activity on other channels.
The ability to detect and respond to these tactics in real-time can be the key to striking the experience-security balance. For example, deploying voice authentication allows a contact center to detect a customer by analyzing the unique characteristics of their voice, while also detecting deepfakes.
In combination with liveness detection, voice authentication can provide a secure method to authenticate callers and improve CX. We have demonstrated, by way of our blog, how bolstering voice biometrics with liveness detection can supercharge a contact center’s ability to detect synthetic voices: even ones that are based on new AI models.
Other behavioral analysis tools can be used to measure incoming call data and verify a caller based on their device, location, keypresses, and calling patterns. These subtle data clues are extremely difficult for fraudsters to identify and replicate.
Each micro-experience intertwines to form the total experience for a customer.
Regardless of the combination, the layering of tools and technology is essential to stay ahead of the curve. Combining techniques can also add improvement and efficiency throughout the CX by shortening hold time, expanding the options for self-service, reducing friction with agents, and resolving issues faster.
3. Post-Auth
Once authentication is complete, the interaction largely consists of getting down to the business of why the customer called in the first place. The interaction can start and end in IVR or can involve an agent.
Businesses should consider tracking and analyzing the common interactions from their customer base, and the unique preferences of their customers, to facilitate faster, more seamless interactions.
For example, an intuitive IVR system could play automated messages upfront during known service outages, prompt specific callers if they have an open support ticket, or automatically connect a caller to an agent if they always press 0.
The last micro-experience is the resolution. Was the caller’s issue resolved to their satisfaction?
While it’s difficult to be prescriptive about how to better resolve issues, it’s important to note that the resolution can strongly influence the experience overall. Tracking and analyzing first call resolution (FCR) metrics can help bring clarity to how your operation is performing and suggest when it might be helpful to train (or retrain) agents.
Bringing It All Together
Each micro-experience intertwines to form the total experience for a customer. Since traditional systems may not provide a holistic view, it’s important that businesses diagnose their operation during each micro-experience to identify imbalance while ensuring they keep malicious actors out.
Designing a system that caters to each micro-experience - and getting more granular from a security standpoint - can inform areas in need of technology investment, process review, and/or staffing changes. Doing so can empower a contact center and its agents in ways that keep customers happy and reward their loyalty, without compromising security.
Often, the key to striking the right balance is investment in passive authentication methods integrated with liveness detection that add automation whenever possible. Along with employing flexible and personalized IVR to accommodate a broad set of needs and the ability to detect bad actors when and where they try to game the system.