Before You Automate People's Jobs, Here Are Some Things to Consider

Industry wisdom holds that CX automation squares away the “boring,” “repetitive” parts of an agent’s job, thereby “freeing them to do more high-value work.”

This cryptic claim is regurgitated by influencers left and right. And yet, no one actually elaborates on what that “high-value work” is, nor do they mention the drastic changes to workforce structure that would have to be made by management: altering workflows, hierarchical structures and recruitment practices, retraining, departmental reshuffling, or even layoffs.

While it’s tempting to focus on the possibility that AI-powered bots will steal customer service jobs, I think it’s premature. For that to happen, companies have to unanimously commit to major capital investments and overhauling their workforce, supplanting low-skilled labor with expensive, top-tier engineers to run and maintain the bots.

But companies are investing in technologies that automate a significant part of an agent’s job. What matters now is how you would feel if your job were automated, assuming you can keep it. What are the implications for agent engagement when companies invest in these time-saving technologies?

Most solution providers for CX automation insist their product is designed to work alongside human agents to reduce human error and make their jobs more engaging, but I wonder if that’s an unsubstantiated marketing claim designed to squelch general unease around Humans vs. Bots: Who Does it Better?

Yes, the job is repetitive. The alternative is…thumb-twiddling?

According to research by Digital Genius, one in three contact center leaders say 40 percent of tickets are repetitive questions, while 20 percent are “highly repetitive, yet easy to resolve.” Can you eliminate 40 percent of an agent’s job without eliminating 40 percent of your agents? Sometimes, the answer is yes, but not all companies are able to offer “high-value” work - which, let’s not forget, means something different to everybody.

A customer service rep at Amazon or Jet likely spends most of their shift fielding the same few queries: “Where’s my package?” and “How do I return an item/get a refund?” Assuming a conversational IVR is able to answer those questions, and a virtual attendant automates the refund on the back-end by pre-filling forms and copy-pasting data, what would the agent do instead?

The NICE virtual assistant known as NEVA (NICE Employee Virtual Attendant) listens to customer calls and surfaces suggested responses, information from the knowledge base, and links to next steps. While that’s standard fare amongst AI-powered agent assists, NEVA takes it a step further. Once the agent makes a selection, NEVA takes care of the rest, pre-filling forms and even filing requests in the back-end by deploying robotic process automation. Essentially, the agent just has to point and click - and talk on the phone, of course.

At the keynote in Orlando announcing the NEVA launch, Oded Karev, VP of Robotics Process Automation at NICE, demonstrates an inbound call to a hypothetical cable company with a customer who wants to upgrade to the latest sports channel package.

While listening, NEVA automatically flashes several callouts identifying the customer and his previous purchases, signs him up on the backend by auto-populating a series of forms based on existing CRM data on that customer, adds the additional channels to his account, and then generates an email summarizing the call that is automatically sent to the customer.

Meanwhile, Karev, who portrayed the demo agent, only had to explain to the customer what the virtual assistant was doing. “Awesome, consider it done,” says Karev, not taking his eyes off his computer screen, watching NEVA work her magic. “The channels will be activated on your TV set and you will be able to see them in 20 minutes. You will see the charges on next month’s bill. Enjoy!”

Granted, it was a staged demo; but it’s easy to see how a real agent using this technology could attend to a customer’s needs while relying entirely on NEVA’s suggestions instead of having to make decisions themselves. What if by automating the “boring” and “repetitive” parts of our jobs we inadvertently make the entire job boring and repetitive?

Automating CX makes perfect sense when it’s clear what agents can and should be doing instead. Nonprofit health care provider Banner Health introduced an IVR that enabled callers to reach patient rooms directly instead of having to go through an operator, which shaved call volume by 20-25 percent.

Cippy Seidler, director of the Consumer Care Center, says she was able to upgrade job responsibilities and pay scales for staff on the switchboard team, appointing them to a caregiver-type role to help customers with “bigger concerns,” like providing updates on a sick family member.

“Self-service can work in your favor to drive employee engagement,” says Seidler, who is slated to speak at CCW Australia in Queensland this February. On the flipside, automating without having a clear plan or purpose can leave you with confused agents who aren’t sure what to do with themselves.

Making voices sound perfect - but human - over the phone

There’s one technology that frees up call center agents from having to speak on the phone; instead, they push buttons to play back human-sounding voice recordings. The Yodel switchboard automates the voice channel through an interactive soundboard delivered via avatar, which enables agents to pre-record a script spoken by a “model agent” and manually program playback during a live call.

Basically, it’s an analog IVR designed to make the customer think they’re talking to a live agent. While its primary audience is outsourcers with offshore operations, its stated selling points are to reduce human error, the costs of hiring highly trained agents, and the likelihood that a customer will be affronted by an agent's “accent.”

If, like me, you arrogantly assume your creative job can never be automated because it’s too...cerebral, seeing the Google Duplex in action will blow your mind. A feature of the Google Assistant, the AI is designed to make phone calls on behalf of the user. Say you want to book a table at a restaurant; the Google Assistant can call on your behalf.

At its developer conference in Mountain View, CA, last year, Google CEO Sundar Pichai showed off the new technology, playing a recorded sample call to a salon to book a customer’s hair appointment. You can almost hear the audience holding its breath to see if the receptionist on the other end of the line will pause and say, “Wait, am I talking to a robot?”

But the Google Assistant sounds infallibly human, even interjecting “uh” and “hmm…” and other speech disfluencies, like pausing to think. It’s like listening to an actor read a script.

This same technology is being used in a series of AI solutions for contact centers within its Dialog Flow Enterprise Edition. The service aims to help customers build natural-sounding chatbots for the contact center, as well as phone-based conversational agents. With the Dialog Flow Phone Gateway, you can assign a phone number to your agent and begin taking calls in under a minute, “all without infrastructure,” because everything is stored on the cloud.

While one of the biggest time sinks when implementing new technology is building a knowledge base, mapping intents, and connecting it with pre-existing systems, Google’s Dialog Flow Knowledge Connectors can understand unstructured documents like FAQs or knowledge base articles to build automated responses sourced from internal document collections.

We’re seeing conversational IVR that’s able to recognize human speech, but it’s still used mostly for routing purposes. Google’s latest iteration of conversational IVR can answer questions, and it’s being used by companies like TicketMaster to handle thousands of routine customer queries.

“Today, a lot of  our customer calls are ‘general’ questions about specific events or venues that have to be handled by a live agent because the details vary by event,” said Tariq El-Khatib, product manager at TicketMaster. “Now, with knowledge connectors, we can begin to explore answering these questions within the IVR, giving our customers easy, 24/7 access to that information without constantly having to build new intents.”

There’s the good, the bad and the ugly

Yes, companies routinely outsource all of their contact center needs to solution providers, but this technology is being marketed mostly to in-house contact centers and touted as an aid for agents to achieve better job satisfaction and performance.

If companies really are committed to their agents’ success, they should do significant workforce planning before implementing the technology to make sure the agent does end up doing high-value work instead of growing disengaged or ultimately laid off. Influencers should also be careful about making unsubstantiated claims about shiny new technology just because it’s so heavily buzzed about, when their comments can have implications on people’s livelihoods.

Sure, certain technologies absolutely warrant layoffs simply because they do the job better and more cheaply than humans can. What's important, however, is not to invest in new technologies with the promise of making people's jobs more engaging and valuable unless there is latitude in existing company workflows to actually do that - and that's a hard ask.