Virtual Assistants Can Be Your Agents’ Best Friends

In last month’s article, I provided a list of game-changers that can transform a lackluster self-service experience into a stellar one. There is a great deal of buzz in the contact center air about “bots” and artificial intelligence (AI) and their potential to make agent-supported contacts go away. However, amid all this excitement, our “Future of the Contact Center Survey” revealed a healthy dose of skepticism. Moreover, some early bot adopters have reported less than stellar results, leading to hand-offs of frustrated callers to agents!

For many centers, agent-handled contacts still dominate­—and will for the foreseeable future. Contact centers face the daunting challenge of labor markets that are not conducive to hiring and retaining great agents, while fielding increasingly complex contacts. Our Future of the CC Survey put Robotic Process Automation (RPA, aka Agent Assistance) as the second “top technology” for contact centers of the future. So, let’s focus on how technology can help agents succeed in handling contacts efficiently and effectively. Whether you call them agent-facing bots, virtual assistants or RPA, the goal is the same: Help agents perform tasks and serve customers properly and quickly.

Identify Your Opportunities

Technology-supported process design proves most effective when it targets “pain points” for which an automated solution provides a healing balm. A few typical examples leap to mind.

As you move from ideas to action, pursue process redesign with all the possibilities the new tools offer.

Agent training generally takes too long and overwhelms trainees with the complexity and diversity of customer interactions. New-hires may be so intimidated they believe they’ll never achieve mastery. And, let’s face it, no one wants to feel inept in a customer-facing job with the concurrent pressure to act swiftly. So, ask yourself: Where does training get bogged down? What makes it take longer than you’d like? What overwhelms agents? What can we do to relieve the pressure?

If training does not prove an agent’s undoing, time to proficiency just might. It takes months for an agent to feel comfortable and confident and perform to expected standards. Complexity and diversity of contacts play a major role here. But it’s also fueled by the difficulty in routing the “just right” contacts to the newbies. Fortunately, the same tools that make training easier could also pay dividends in accelerating agent proficiency and confidence.

Some tasks are tedious—the common, routine ones that still get to agents but could be handled by self-service if only the customer would use it. Think about how those repetitive, simple tasks could be made so much faster and easier for an agent if the system worked on their behalf.

Other tasks are complex, challenging and less common. The agent must spend more time trying to figure out the right process steps to follow and where to get information. Failing that, they end up escalating to a more seasoned resource, which isn’t good for anyone! Think about which elements of those more difficult contacts could be streamlined so that the agent could spend their time on problem-solving skills and a personal touch.

Sales opportunities often get missed by an overwhelmed agent or an increasingly impatient customer. A harried agent can easily skip an important step that might prevent a subsequent contact or ensure compliance. If you take the frustration out of the equation, how can you make it easy and convenient to complete all those “next steps?”

Focus on Transforming the Agent Experience

As you move from ideas to action, pursue process redesign with all the possibilities the new tools offer. Consider all channels—voice, email, chat, text, etc.—with an eye toward the end-to-end process that completes each transaction. (In other words, think about the full ecosystem, not just the contact center!) Some companies start in the back office and then move to the contact center. Non-real-time tasks are good “test beds” that advance learning while shielding customers and agents from bumps.

Define how tools could work on behalf of your agents, customers and company. Handle the simple stuff with automated process steps. Provide pathways through the more difficult stuff (or more likely, parts of it) with faster/better access to the right information and guidance on what to do. Prompt for the additional input and consider the context to figure out whether to do something or escalate.

Seek RPA and/or business process management (BPM) tools to support your high-level design. Identify the various contact types and channels, and consider the integration points—KM, CRM, core systems. Recognize that there are differences between enterprise-grade (often called iBPM) initiatives and solutions that are a bit more manageable. You don’t want to “boil the ocean,” but you also don’t want to simply “microwave a teacup of sea water.” The latter is often the refuge of expedience that fails to deliver any material value—e.g., “screen scrapes” of legacy solutions that only fix clunky desktop tasks.

Determine if you are going to make AI part of your process design—now or in the future. You may not need this technology for the simple stuff, but it can provide great value there or for complex interactions. Keep in mind, AI needs enough volume to make the investment worthwhile. (See past articles in Tech Line/Tech Corner.)

Determine if your process redesign and associated technology will tie into your self-service initiatives. Consider the interfaces, intelligence engines, and knowledge and data sources that agents tap and how they might be leveraged for self-service interactions. Even though the buzz in bots seems to focus on self-service, assess if these initiatives are likely to drive significant volumes away from agents. Invest time, resources and budget in agent-handled contacts in proportion to the workload that they handle. Then work backward to see if that investment can give a boost to self-service.

Define phases. Start simple, focusing attention on the easiest, biggest hitters that will deliver high value and let you learn quickly. Consider a proof of concept before you get in too deep. Build on what you learn from implementing the PoC and initial phase(s).

Find the Right Solution and Partner

Exciting ideas paired with great design and supporting technology will only be as good as the implementation and support that goes with them. Give due consideration to the resources and professional services that you’ll need to develop and optimize your process automation tasks. While vendors often lead the charge in the development phase, both IT and operations will need to play a role to ensure the solution meets your internal requirements and can be sustained long-term. Likewise, plan for vendor support post-implementation to optimize the solution if you lack the subject-matter experts or internal resources to do it yourself. And keep in mind services may go beyond design and implementation; they may include using the vendor’s tools to help uncover process improvement opportunities!

Bots and Virtual Assistants for Customers and Agents

As you wade through possibilities for technology-enabled process redesign, the industry lingo might trip you up. Some people think of “bots” and “virtual assistants” as being exclusive to self-service, but these terms can apply to either a self-service activity or an agent-assisted interaction. The distinction you want to make is between “customer-facing” and “agent-facing.”

In discussions with vendors, you’ll want to get to the nitty-gritty of how the following three elements function:

• The user interface (text or voice)—whether for agents or customers
• The intelligence engine—i.e., where the business rules reside and/or the AI operates
• Data and knowledge sources—i.e., where the engine finds context, information and answers

Whether a solution supports customer-facing or agent-facing applications (or both!), these three elements must align. They should generate the same answers or information, with the knowledge and data sources providing a “single source of truth.” And they both should benefit from targeted continuous improvement efforts.

Find vendors that fit your requirements. Review analyst reports from Gartner, Forrester, Ovum and the like. (They are all writing about this hot area!) See what your existing vendors offer. Explore a variety of sources, including:

• Vendors that target RPA (e.g., Jacada)
• Vendors that sell performance management tools (e.g., NICE)
• Vendors that sell CRM and BPM (e.g., Pegasystems)

Consider offerings for “attended” and “unattended” process automation. Attended solutions work with agents; unattended solutions are fully automated (e.g., an event can trigger them). Some vendors offer both self-service and assisted service, so assess their broader offerings as appropriate for your vision.

Build a Quantitative and Qualitative Business Case

As you are working through the design process, making decisions on vendor partners, and sharpening your pencil to capture the design, implementation and support costs, be sure to identify all the benefits that will sweeten the bottom line and garner approval from your customers and agents. Factors to consider include:

• Shorter training time and time to proficiency
• Reduced turnover, especially early attrition tied to the “O-factor” (Overwhelmed!)
• Shorter handle time
• Higher first-contact resolution
• Improved employee engagement by eliminating mundane transactions and simplifying complex ones
• Better customer experience resulting in higher loyalty and increased value, typically captured in customer sat and NPS score
• Improved access to information and data that boosts both assisted service and self-service

You need the efficiency and the agents and customers need the improved experience. A bot that helps your agents (not replaces them) may be just the answer!

Build on a Good KM

Too often, centers seeking to improve the agent’s world focus on a basic knowledge base—via SharePoint, an internal portal or some sort of Wiki. While that has value, it doesn’t go far enough. A good Wiki/KM provides superior search capabilities, a strong mechanism to add information and answers (and weed out bad ones), and guides agents toward answers. I’m a big fan so I’ve written several articles and white papers on KM. And now I’ll add one more reason to pursue it.

A bot/VA/RPA leverages KM and does more. It provides workflows and guidance for process steps. It can prompt the agent to do the next thing as well as trigger other departments to attend to their responsibilities. A bot/VA/RPA can help with routine or complex processes. And it supports compliance, too. Align your KM efforts with your automation efforts so everyone succeeds.