Artificial intelligence (AI) and automation continue to change the business landscape. Nowhere is this more apparent than in customer service. In fact, Forrester calls both out as major contributors to the three “megatrends” they see taking shape in 2020, calling both “a long-term investment … to better support your customers and your agents.”
Taking a closer look at these trends, the first is how the use of AI and automation make it possible to more readily scale customer service operations. The most obvious is in having AI and automation handle repetitive and simple inquiries. Another example is how AI and automation will power self-service technologies such as chatbots, knowledge management, communities, and automated solutions. This all comes in addition to tasks like prioritizing, categorizing, and routing cases.
With more and more of the high-volume and repetitious work being addressed by self-service and automation, more complex issues and those requiring empathy are directed to agents. For this reason, the second trend identifies the importance of desktop technologies that utilize AI. This provides agents with timely, in-context answers and guidance to help reduce resolution time. Likewise, with help “riding alongside,” new agents can be on-boarded and become productive faster.
With AI and automation moving into the agent’s territory, the third trend sees changes coming to customer service culture. While the need for basic-skilled agents will lessen, the need for higher-level roles like superagents, chatbot designers, and other AI-related positions will grow. FUD–fear, uncertainty, and doubt–will also likely occur.
Acting on these trends means a brighter future for customers and new opportunities for agents; but if poorly executed, they create the potential for spectacular failure. To smooth the transition, consider these three pre-adoption steps to make success possible.
Ready the data
While AI holds great promise of speeding work along, the truth is that it’s really not all that smart or skilled–at least to begin with. AI (better termed as machine learning in this context) requires teaching the machine how to behave in very narrow and structured circumstances. Just as humans learn, AI requires three things: credible, correct data; a good-sized volume of data to analyze; and time to build a model from it all. Without all three, an AI initiative will fail.
For example, if AI will work with cases–prioritization, categorization, and routing–prior case data should be validated for accuracy before starting the learning process. Along those lines, any means by which human-entered case data can be standardized (e.g. training errant agents in proper case entry, use of pick lists, etc.) will help minimize future case data review. Experts suggest at least six months of data is required to build good models, and AI must be constantly learning from the work of human counterparts as well as their own mistakes to succeed.
Clean up processes
In addition to aligning how data is collected, look at how it moves around the organization. This is important to automation for a few reasons.
First, a process might not be nearly as efficient as expected. Processes have a tendency to evolve and become more complex over time. One of the benefits of automation is the speed by which it can execute. It will operate much more quickly if its actions are reduced to the minimum necessary. Use the move to automation to evaluate if the process end-to-end still makes sense.
In preparing to automate a process, this is also the opportunity to update its documentation. All series of tasks–automated or manual–should clearly state its intent and the steps it takes. Moving to automation doesn’t remove this requirement; in fact, it makes it even more important.
And the bonus? Since it’s a form of technology, automation may fail. If the process has been simplified and documented, tasks can quickly transition (temporarily) back to manual until automation is available again.
Address human concerns
It’s in most people’s nature to fear and resist change. AI and automation are not just a fundamental change coming to customer service, they are technologies many don’t understand, further fueling that apprehension. In some cases, that anxiety even leads to illness.
Alleviate these challenges by getting ahead of employees’ concerns. Involve employees early in the planning process. Adopting AI and automation is a journey and not a destination, so share the roadmap. Transparency and discussion help prevent insecurity from taking hold.
Having addressed the emotional side of the equation, the truth is not every employee in a role affected by AI and automation will have the skills and knowledge to transition into new positions. Provide a route into those new jobs with training–internally and externally–that will help them build the necessary expertise. By starting these discussions and taking these actions well in advance of AI and automation entering the scene, agents have more time to up-level their skills for what’s needed.
Succeed through preparation
They may be megatrends, but that doesn’t mean bringing AI and automation into customer service is a simple undertaking and a quick win. Though they offer to handle the mundane, repetitive work and make it possible to solve customer issues faster, much is needed to realize that value.
Ensure a successful outcome by doing the pre-work. Amass accurate and sound data by cleaning-up what exists and the adjust (if needed) the process of generating new data that machine learning will consume. Evaluate, rearchitect, and document processes headed for automation. Respond to employees’ emotional needs while also setting them up for the new opportunities ahead. Taking these issues into account will smooth the addition and strengthen the success of AI and automation in customer service.