Artificial Intelligence and the “Creepy Factor”

By Andrew Pine posted 09-17-2019 05:36 PM


Yesterday @Allen Lee, @Mike Gathright, and I hosted the #ArtificialIntelligence Interest Group session at Customer Response Summit, which yielded some surprising insights into how brands and business partners are leveraging AI to smooth the customer experience – often by focusing on the human agents who support them.

A quick show of hands revealed that the vast majority of represented companies are somewhere in the middle of their AI journey. About 10% consider themselves early adopters and 5% are still thinking about dipping their toe in the water – meaning 85% are either in the process of AI implementation or are already building on the learnings of AI solutions now in play.

Our session led off with a brief presentation in which Allen shared the framework Symantec is using to apply AI technologies across the spectrum of customer and agent experience, from low-touch automation to high-touch agent assist. We then broke out into table discussion as attendees shared their personal experiences and challenges around AI. Here are some key trends and takeaways, plus a few creative use cases some companies have applied (with varying rates of success). Are you trying any of these within your own organization?

Think Problems, Not Solutions

Smart application of AI technology entirely depends on the problem you’re trying to solve. First identify what you’re trying to accomplish, then decide which data points you need to collect in order to create the desired outcome.

“We’re a global company and we don’t want to provide 24/7 live support – especially for simple questions. A self-service approach enables us to provide easy answers any time of day.”

“Deflection from voice is huge for us. We’re looking primarily to chat/SMS support to reduce call volumes.”

“We’re looking for call deflection and next issue avoidance, but we also need automation that enables agents to step in at the right point of contact.”

“In the healthcare space we do a ton of voice because the focus is on agent accuracy in potentially life or death situations. Process automation is great for certain steps but it’s hard to recover if you get off on the wrong track.”

Start small by automating the most repetitive customer support functions before leveling up to a more robust solution.

“The back office is a good place to look for opportunities for RPA (Robotic Process Automation).”

Managing the “Creepy Factor”

When does intent prediction start to feel invasive? Avoid the “creepy factor” by applying emotional intelligence. In the same way that people don’t necessarily share everything they know about another person during the course of conversation, so can your AI be selective in its messaging.

“It’s all about customer perception. In our experience, the customer doesn’t necessarily mind when we refer to personal data points – provided that it is relevant to their situation. They want their problem solved.”

Also consider the current regulatory environment when deciding what and how to collect customer data.

“Assume that all the data you collect on your customers will one day end up in your customers’ hands.”

“Machine learning can play a big part in privacy protection and fraud prevention.”

System Integration

Knowing what a customer wants doesn’t mean you can do anything about it. You need good system integration to make sure you can act on the data you collect. Don’t forget the backend aspect to AI solutions.

“Most AI technologies are optimized for contact center and web. Figuring out how to apply AI learnings to the retail environment is an opportunity. How do we collect retail data and deliver it to a human agent when they need it?”

Employee Engagement

AI can be a key driver of employee engagement because it frees agents up to work on more complex, potentially rewarding tasks. It can also yield valuable data to improve quality and customer satisfaction.

  • Chatbots response rates can be tied to agent approval (NPS, CSAT, etc.) to help refine the accuracy of responses.
  • Use speech analytics to QA all transactions and perform call driver analysis.
  • Capture where the customer has been so you can predict NBA (Next Best Action).
  • Machine learning can pool typical responses to come up with the most likely answer to a question. Agents then approve or reject the answers to fine-tune the suggested responses.
  • Humana routes calls based on personality type, matching customers with live agents who share similar traits like talkativeness, assertiveness, etc.
  • Nordstrom focuses on agent empowerment, making the agent the “hero” with real-time injection if an AI experience isn’t yielding positive results.


Are You a CX Leader at a Consumer Brand?

Join our online community for more great discussion, and to take advantage of a members-only special offer of 50% off registration for our next Customer Response Summit in Hollywood, FL, March 2-4, 2020.