Fall 2017

How Digital Employees Will Change Your Quality World

The Rise of the Digital Employee

What is a “digital employee”? Robots and artificial intelligence devices are all around us. Think of robots in manufacturing, iRobot for floor care, Siri/Alexa/Cortana on your personal devices, and even self-driving cars. How does that add up to digital employees (DE) in a contact center?

Any kind of digital device that can interact with a customer instead of, or as a supplement to a human employee can be considered a digital employee. Interactive voice response (IVR) systems that have been in call centers for years complete transactions for more and more customers, or at least gather some information, saving the human agent from spending time on the simpler tasks. Web-based offerings including order handling, answers to common questions, and general support.  As an example, a huge percentage of today’s banking takes place at ATMs, via IVR, and on the web, while the number of tellers has shrunk dramatically. A digital employee is one that interacts in natural language and has responsibility for serving a digital customer interaction channel. And it can do so in multiple languages. Program it once and use it to handle many transactions.

As the possible interactions become more sophisticated and artificial intelligence brings more possibilities to the table, the human agents will take fewer contacts (and probably the ones that are longer and more complex). Calls may start with humans and then be transferred to the digital employee for simple tasks, or may start in the systems and then transfer to the human when the complexity exceeds the systems’ capabilities. For example, a request to a human agent for a password reset may be transferred to the digital employee for completion of the transaction. On the other hand, when the digital employee cannot understand the request or needs assistance, the call can be transferred to the human employee. Some interactions will be completed in one channel and some will require multiples.

Gartner projects that by 2020, up to 85 percent of interactions with an enterprise will be completed without a human agent. While the millennial generation is tech savvy, there are also an estimated 73 million Generation Z consumers (born between 1995 and 2012) who are used to interacting with the web and voice in their phones/computers so the adoption of more self-service options seems inevitable. These customers want to do business via their channel of choice, and many will choose several channels over time. Mobility, 24-hour access, and ease of use are expected.

The capital investment and ongoing maintenance of these digital systems generally have a strong return on investment compared to the ongoing cost of human agents. They remove the mundane, repetitive task from the human agents, scale without limits, work for free (after implementation is completed), and do it 24-hours a day, 365-days per year. And there is no need to involve the human resources department to work effectively.

The digital employees can handle problem solving, contextual, and informational tasks but organizations will still need human employees for empathy, humor, intuition, and creativity. Dealing with processing mass interactions is best left to the digital employees. As you look at your interactions, you will likely find that the largest quantities are in the simple processes the digital employee can handle, but many centers have more human resources than digital deployed to handle them. Over the next 10 years, the distribution of the work will push more and more to the digital employees. According to Forrester, “Over the next decade, 25% of all jobs will be transformed in terms of responsibilities as a result of increased automation.” This means change for the human agent expectations as well. They will need to be more highly skilled and educated, better at building relationships and selling, and able to work in an omni-channel environment. It will also be time-consuming and more expensive to replace them. While the need for human employees will continue to rise (but at a slower pace), the rise in digital system deployments is expected to rise rapidly.

A significant process change in the workforce management area is also suggested. Forecasting and scheduling the human agents, taking into account the availability and capabilities of the digital agents, could be challenging. Predicting the mix of simple and complex calls and the need to transfer between digital and human employees will require keeping a close eye on the changing nature of caller needs and the capabilities of the digital systems.

Impact on Quality Assurance

What does this mean for quality assurance? There is the continuing need to measure quality for humans to support coaching and development. But it is also important to measure quality on the digital employees to identify the need for reprogramming or additional functions. Standards will need to be developed to measure the effectiveness and efficiency of the digital employees. Measuring these against customer satisfaction will also be necessary to ensure the greatest level of completion in the systems. When standards are in place and measuring is done, calibration is also important, just as it is for human interactions.

How is the Digital Employee QA Process Different?

  • Transaction Recording
    • No screen recording for chat (i.e., no agent screen)
    • Record end-to-end conversations for voice, chat, SMS, etc.
    • DE systems need to enable speaker separation and new metadata types
  • Sampling
    • Sample less because there is no variability between Digital Employees – only variability in stimuli and how the DE responds
    • Sample more because the technology is new and needs to be proven
  • Evaluating
    • Very different evaluation criteria (e.g., ability to understand customer, similarity to human speech, narrowly focused questions)
    • Omit human characteristics such as empathy, humor
    • Add elemental versions of human agent evaluation
  • Calibration and Scoring
    • New evaluation criteria and new standards for DE will require new calibration standards
    • Balanced score cards will be composites of new evaluation criteria
  • Coaching
    • Recommendations for improvements in DE interactions
    • New types (non-human) types of problems will be typical
    • Changes made via artificial intelligence (AI) professional services


    Role of Analytics

    Automated engagement analytics will play an important role in the quality assurance for the omni-channel center. You can’t stroll the floor and hear digital employee conversations. Analytics of this new technology is essential for getting it right in your center. Human agents and digital employees will grow in non-voice channels. Automated omni-channel engagement analytics will become increasingly necessary.

    With 100% call recording, analysis of the interactions becomes rule-based to spot interactions that need human quality analyst review. The digital employee can be programmed to search for tags such as key words (i.e., overcharged, dissatisfied, talk to manager, ridiculous, third time calling, etc.) Also tracking the number of transactions transferred into and out of the digital agents will reveal opportunities for improvement.


    Artificial intelligence and digital employees will become more prevalent. It is a powerful but relatively new technology. The recommended approach is to start with informational tasks as these are the simplest. Also deploy the system to serve internal employees to minimize the risks before deploying it for customer use.

    Aspect offers a product called Mila that can be demonstrated for the company employee interactions such as checking schedules, vacation requests, bulletin board, understanding performance improvements needed, etc. It can work with desktop, text, or phone to interact. Using an internal group to test your artificial intelligence strategy may be a good way to explore the possibilities in your business before introducing externally.

    Christina Cowell is Director, Product Management at Aspect. She presented a session on this topic at the 2017 QATC Annual Conference. She can be reached at christina.cowell@aspect.com.

Human Agent Digital Employees
Higher complexity transactions require more sophisticated metrics/KPIs New performance standards for digital employees will be related to language and apparent intellect plus skill New performance standards for digital employees will be related to language and apparent intellect plus skill
The more complex…the more refined the measure required Standard metrics will be much different than human employees: confusion rate, misunderstanding rate, conversation steps, machine learning accuracy, success rate plus traditional AHT, FCR, ASA, etc.
Need to understand effect of more complex transactions on standard metrics/KPIs such as AHT, FCR, ASA, adherence, occupancy Performance metrics may vary significantly by channel
Most metrics/KPIs have developed over many years
New standards are required in non-voice channels