AI in Customer Service: The Triangular Model of 'Experience-First' Implementation

Customer service, even when powered?by technology, isn’t?about?technology; it’s about customers and about service. Likewise, the customer experience, even when informed by technology, isn’t?about?technology; it’s about customers and the experience.?

I find myself making these points repeatedly because, as a customer service and customer experience consultant, I don't want to see my clients?chase promising new technology and channels of engagement without the nuanced implementation and integration necessary.?In customer service and support,?technological implementation?needs to be?undertaken with a deft touch to ensure that you don’t end up, after you’ve spent a ton of money and two tons of time, with something that alienates rather than engages customers.

Consider the issue of bringing chatbots into the customer support mix.?The customer-hostile way to go about implementation is with an either/or mentality, thinking of any new interaction as being either a chatbot conversation?or?an agent conversation, which goes with the unfortunate model of “escalation”: let the customer languish in the limited care of the chatbot; then, only if that doesn’t work out, condescend to permitting them to get “escalated” to a human agent.

I prefer a triangular visualization, as follows:

That conversation might start with a chatbot, or it might start with an AI-powered dynamic search bar. But AI can't do everything; nor will it be able to in the near future.?That’s where the agent comes in—not with the bot “handing it over” to the agent, but with the bot continuing to assist the agent, in one of two ways:

? The AI-powered bot can then continue to support the customer conversation and if the customer asks a question the bot knows the answer to–the trackable status of a shipment, for example–the bot can automatically insert those tracking results into the conversation and free the agent to focus on more important things or work with more customers simultaneously.

? Similarly, the bot, interacting behind the scenes with the agent, can be making recommendations such as, "Based on what they're talking about, I think the best approach for them is X." Then, the agent can either insert that recommendation verbatim or reword it in a more empathic manner. (Hat tip to Ryan Lester.)

Consider this hypothetical conversation between the customer of a financial institution and an AI-powered bot and a human agent.

Customer:?"I want to roll over my IRA"

Bot:?I can help you with that. Here's the link where you can fill in the info for the institution and account you'd like to roll it over to.

Customer:?Thanks. But I see that the form you linked me to is demanding the address of my [external] institution. I only have their account #, not their physical address.

Bot:?Let me get you to an agent who can help you with this.

Customer:?Oy vey.

Agent, stepping in:?Unlike the bot, I actually speak a bit of Yiddish––and, I can help you with this. The address is a formality but you can't leave it blank.?Let's see what we have as far as addresses for your institution.

Bot to Agent (invisible to customer):??I find their address in Orlando at XX.

Agent:??We have found an address in Orlando that may work for you.?We would always recommend checking with your institution, but the address we have found is an official on-record address for the external institution so it should work.

Customer:?I retract my “Oy vey.” Thanks for the help!

Ta-da! A happy customer, an empowered agent.

________________

Micah Solomon is a customer service and customer experience consultant, speaker, trainer, and training designer. Micah was recently named “the World’s #1 Customer Service Turnaround Expert” by?Inc.?Magazine. Reach him directly at [email protected], visit his?website, or check out his new bestseller:?Ignore Your Customers (and They’ll Go Away)?(HarperCollins Leadership).