The use of virtual agents in business, for both employee and customer use, has accelerated in the past few years and has overcome the trials and pilots to deliver real measurable business value. As the need for smarter, more capable, complex virtual agents increases, a virtual agent operations industry is rapidly emerging to help companies navigate the difficult post-launch roadmap needed to be successful.
Clients considering outsourcing their virtual agent operations are looking to vendors with successful deployments, strong references and powerful services and tools. But, is the industry delivering?
What is Virtual Agent operations?
Virtual agent operations is the work that begins after a virtual agent has launched. Unlike traditional software, NLP-based applications require continuous “tuning” to ensure they evolve and learn new things. Virtual agents can always get smarter and operations teams have the expertise, experience, and state-of-the-art tools to deliver the continuous evolution your business and customers deserve.
They know what it takes to create and sustain a successful virtual agent program, and use various AI management, customer journey management and insight-gathering tools to keep your virtual agents working the way they should.
Companies need to invest in Virtual Agent operations teams to be successful
Yes, anyone can build a chatbot in hours using the many builder applications or websites. However, that approach is only good for automating the very simplest of questions.
Business-grade virtual agents require more time and investment to develop a high-quality and high-value level of service that delivers what customers need. Companies need to invest time and money with their in-house and external teams to operate virtual agents based on strong analytics and customer insights, or they really should not bother at all.
This is especially true for high-volume and complex use cases including airlines, banks and healthcare, where chatbots can handle between 50% to 90% of typical use-cases. Research firm Gartner estimated that 25% of digital workers will use a virtual agent daily by 2021, and with the rapid growth, that number could easily be higher.
The value of outsourcing the more difficult or tedious aspects of Virtual Agent operations
Operations teams need access to niche skills and tools.
An analytics-based approach starts with studying every single message between a customer and a bot. Given that there are typically millions of messages this obviously requires specialized analytics tooling. Building your bots around a strong knowledge base of real-world data also adds greater value over teams building a bot from a blank piece of paper.
In operations, teams measure each chatbot on many dimensions to understand the performance across many areas. Analysis starts with the core contact reason and then dive into specific metrics such as volume, containment, sentiment and handle times by topic. Operations teams are also paying close attention to lesser-known examples including false positive rates, repetition rates, undertrained intents and confused intents.
These will show the true value the chatbot provides and highlights statistical and functional areas where it can be improved to deliver greater service to customers. Understanding those metrics and the resulting statistics is one thing but any business looking to build a bot will gain more from an experienced team who have seen plenty of bots in action and know what the numbers actually mean.
They can help deliver continuous improvement as a value-added feature of chatbots, filtering and delivering fresh or improved use cases, channels and transactional capabilities. Taking an analytic-based approach in identifying the greatest value, especially for firms migrating from live customer support to chatbot-based approaches as part of a strong partnership.
As an industry, Virtual Agent operations is maturing
As the rapid rise in enterprise and business class virtual assistant deployments continues, so does the field of supporting operations. Virtual agents are finally reaching enterprise-grade, however, they can always be improved and the training cost of these systems can often be prohibitive.
Underlying the process is the need to build a data-driven roadmap of all candidates for automation across the business. This should highlight the path to success with AI and explicitly state what is needed for each use case and what the success goals are.
With a strong roadmap, operations team and an overall virtual agent plan, firms will be able to develop and deliver strong projects that deliver amazing results early on and become a key part of the business, freeing up live agents and other workers for high-priority and higher-value work. If you have any concerns about your virtual agent operations, give Wysdom a call.