How to Manage a Virtual Bot Agent

In our previous blog in The Times They Are a-Changing: Talent in the Contact Center Series, we discussed how virtual agents, or bots, are transforming the contact center workplace, pushing live human agents to become strategic troubleshooting experts and managers to become agent coaches. While these changes are significant, they are largely beneficial to agents and managers alike, who are now more engaged in the business.

But what about the bots? How do you manage these “virtual” agents? They are never late, do not tire and are programmed to continually improve. But that doesn’t mean they don’t need supervision, or perhaps evaluation, to ensure they are gaining the right experiences and continuing to deliver positive customer care. Who’s responsible to oversee the bots?

Of course, the IT department, data scientists from the research team and even playbook writers play a critical role in managing and assessing AI and bot technology, but it’s the contact center managers and ‘fellow’ human agents who have deep expertise in the customer experience. The contact center managers play a critical role in managing the bots, helping them to learn, improve and meet the changing needs of the contact center. 

The best way to optimize bots for the contact center is to think of them as full-fledged agents. Contact centers can get the best ROI from virtual agents by applying the same basic processes as managing human agents – with some key differences, of course. 

Hiring: AI technologies are evolving fast. With many new players emerging— each with different strengths—managers need to ensure that their organizations stay abreast of these technology changes, figure out which ones work best for which issues, and ensure that their underlying technology platform allows for the swift integration and management of multiple AI technologies. 

Contact center managers are on the front line with their agents and understand what is needed to maintain and improve the customer experience. They should be a key part of the bot evaluation and selection process to essentially help ‘hire’ the best virtual agents.  

Training: Bots learn through AI algorithms and assimilate data to improve. Yet even the most expertly designed algorithm can’t help a bot understand what behavior is proper or not and which outcomes are good without real-life customer experiences to draw on. Managers need to team with the data scientist to ensure bots are getting the right training based on experience in interacting and learning from customers. For training, data scientists will probably take the lead, but managers should emphasize the desired behaviors they want the bots to develop. 

Evaluation: Managers need to track performance and set improvement goals for bots, much as they do for live human agents. The metrics and training techniques, however, are vastly different for bots. Contact center managers should work with the data scientists to implement a program that can measure results and give insight into each bot’s performance to help the bot evolve in a positive way.  

Coaching: In the previous blog, we discussed how contact center managers are shifting from a taskmaster/boss function to a coach/mentoring role. This puts a renewed emphasis on helping live human agents continually improve. The same is true for bots, although the techniques to ‘coach’ them are fairly different. Managers will review evaluation results to assess where a bot can improve. But coaching a virtual agent is more about adjusting training data, modifying bot rules, and perhaps tweaking conversational design, as opposed to positive reinforcement and team-building exercises.

Firing: Just because bots are programmed to continually improve doesn’t mean that new technologies and advances won’t produce even better bots that can deliver faster, more efficient service. Just as contact center managers have to make tough decisions regarding live human agents based on performance evaluations, they should also consider when a new bot is worth the investment and when an old bot is ready to retire. Managers will need to work closely with data scientists to understand how a new technology will integrate and interwork with existing AI technologies. 

As with the other management processes above, ‘firing’ a bot is a shared responsibility that will require managers to collaborate with data scientists. Contact centers might consider a phased approach that keeps the old bots in service while the new bots ramp up and gain the experience they need to learn and improve. 

Managing virtual agents is a fairly new experience. Managers will need to rely on data scientists and the research teams. While it might be tempting to simply let the research team manage the virtual agents, the customer experience is ultimately the contact center manager’s responsibility. The contact center would miss a huge opportunity to optimize the customer experience by not giving managers a central role in key bot decisions. Managers should be central in hiring, training, evaluating, coaching, and even firing virtual agents. Contact centers are providing a blend of virtual and live agent services. This requires a blended management style that includes both bots and human agents. Ultimately, this leads to better service and a richer customer experience.

Check out the previous installment of this series: “The Times They Are a-Changing: Talent in the Contact Center Part 6: Management or the next one, Organizing the Contact Center to Drive the Digital Experience for Customer-Centricity | The Times They Are a-Changing: Talent in the Contact Center Part 8