What should you consider when designing virtual agents?

[davidkwock]

Many enterprises today are exploring virtual agent technology as a way to provide a better user experience to their consumers while reducing the cost of providing service.   Designing these solutions often involves the initial vision of replacing manual human activities with cognitive virtual agents that act consistently while providing the end users with quick accurate solutions.   In this article we will explore what the customer journey looks like when adopting virtual agent technology and some best practices when designing cognitive virtual agent solutions.

The Cognitive Virtual Agent Customer Journey

The customer journey to getting business value from cognitive virtual agent technology can be summarized in 5 steps:

1. Select Content - Select the type of request a virtual agent can handle well for your enterprise that drives the most business value.   For example, if 30% of your call volume is around the topic of password reset or the status of existing tickets, these might be good candidates for a virtual agent to handle.
2. Positioning - Ensure your customers can easily access your virtual agent.   For example if 90% of your request come from phone calls you need to make the virtual agent accessible via the phone.   It might be easier to make the virtual agent available via a chat window, but since the majority of your customers use the phone as their system of engagement it will be important to provide a way for them to get to your virtual agent via the phone.
3. Success Metrics - Define metrics for a successful deployment of your virtual agent.   This might include metrics of how many calls a virtual agent handles vis a human agent (request deflection) or how many times a virtual agent routes a request to the right fulfiller.
4. Implement - give your virtual agent a personality and identity like a human employee.   This includes clearly defining the virtual agents role and scope.   What do you expect your new virtual employee to do and who do you want the new virtual employee to service?   More importantly what do you not want the virtual agent to handle?
5. Measure Success - finally it is very important to measure success after deployment of a virtual agent to ensure the business outcome and value the virtual agent is providing is understood and can be refined.

Virtual Agent Best Practices and Lessons Learned:

Below is a summary of some Best Practices and lessons learned I have seen in the field when implementing cognitive virtual agent technology.

• Clearly define and communicate the use cases and audience the virtual agent serve
• Research what request are most common in your enterprise and what are good fits for a cognitive virtual agent to address
• Attach an automated workflow when possible so that the cognitive virtual agent so that virtual agent can solve problems vs just having long conversations with customers.   The better the automated workflow the better the adoption the virtual agent will experience.   At the minimum you always want to move your customer closer to the right answer not be another step in the customers quest for the fulfillment of the request.   A great example of this is if you are going to leverage a virtual agent to help customers get the status of request, you will want the virtual agent to be able to route the request to the right fulfiller and give a time to fulfillment to the customer vs just reporting that the status of the request is "waiting to be fulfilled".
• Leverage search capabilities to enable the virtual agent to give relevant answers from existing knowledge bases or previously solved requests.