Bots vs Humans: 4 Ways to Balance Automation and Live Agents

The bots versus humans debate rages still – and not only at the behest of workforce automation detractors.

Organizations concerned with leveraging CX automation for a better customer experience are navigating the gray area of when to delegate tasks to bots and when the human touch is crucial.

Instead of getting caught up in the one-dimensional rhetoric bots are for repetitive queries, humans for complex ones, “it’s far more compelling to look at where chatbots are actually more valuable than live agents, not just areas where they’re good enough to get the job done,” Brian Cantor, principal analyst at CCW Digital, said in a recent webinar with Teleperformance, ‘Unleashing RPA as a High-Tech, High-Touch Solution.’

To make a judgment call, it’s best to consider whether the situation involves a transaction or a conversation. More often than not, transactions are better handled by bots and self-service channels because they involve data. Conversely, conversations that are consultative or situation-based need the human touch.

Here are some lenses to help you make the right distinctions.

1. Is it a Q&A-type conversation, or a discovery process?

Conversations based on FAQs are best handled by bots. In fact, making a customer wait on hold in a phone queue to get an answer to an elementary question is an abomination in this day and age. However, some customer queries are more discovery-based, usually when the customer needs an informed recommendation from someone who understands their needs.

“An example might be, let’s say you’re asking about a gym membership cancellation policy as opposed to diving deep into a fitness plan with your trainer,” explains Cantor.

Talking to a trainer is consultative and needs-based, and the customer expects to exploit the knowledge of a seasoned professional, whereas the gym membership query is a matter of regurgitating information from a knowledge base and making any needed adjustments on the backend – something a bot can do in a flash.

Read more: Special Report - Increasing Humanity with Chatbots

A second factor is the type of experience you seek to provide. For example, retailers like North Face have introduced AI-powered personal shopper chatbots on their websites to help customers make purchasing decisions by asking questions about their needs and making automated recommendations. It’s all well and good for a relatively low-stakes item like a ski jacket, but you wouldn’t dream of buying a $20,000 Rolex watch without picking the brains of a sales representative at a showroom.

2. Are accuracy and speed more important than relationship-building?

When it comes to matters like compliance issues, due dates, payment updates - or frankly, anything numbers-oriented - it’s best to hand the conversation off to a bot. “These are areas where there’s a clear right and wrong and you want to make sure the data is accurate for the customer,” said Cantor.  

Bots are wired directly to your CRM, accounting software, sales database and more, so there’s no go-between that might misinterpret or misrepresent the information. Human agents, on the other hand are prone to inconsistencies, human error and the lens of personal interpretation.

“You may also want to prioritize speed over relationship-building,” added Cantor. Say your flight is delayed and you need to rebook for the next leg of your trip. Most people would prefer to select from a timetable of available flight times, compare fare changes, and receive immediate confirmation regarding the new flight.

A human agent, on the other hand, would have to read aloud the flight times over the phone and the customer wouldn’t be able to see a visual comparison of times and fares, which makes for a stressful (and time-consuming) back-and-forth.

3. Are you executing a transaction or having a conversation?

Similar to the above example, any situation where a customer wants to execute a transaction involving a back-end process, a bot is usually a superior conduit, even when speed isn’t top-of-mind. For instance, say you’ve ordered a pizza and, after hanging up the phone, decide to add a side order of garlic knots. You call the restaurant back and the person has to take your name and address, look up your order number, add the item to your order, alert the kitchen and recalculate your bill. That’s a lot of work for the restaurant worker and a lot of dead air on the phone call with the customer.

If you could instead pull up an app and message the bot about your additional order request, the bot can recalculate the price (or automatically charge your credit card) and alert the restaurant with an app notification.

“The simple reality is that machine-driven tools are far better than humans are at repetitive tasks, things like data entry, reporting, scheduling and call logging,” said Cantor.

4. Are you identifying customers based on account information or their personality?

A customer calling with an insurance claim probably isn’t interested in schmoozing. Rather, they want a fast and frictionless determination of whether or not they’re entitled to an insurance payout - and if so, how much and when they’ll receive it. On the other hand, an art dealer who calls a gallery wants to be recognized as a trusted partner, because it’s a relationship-based interaction.

“If you want to focus on specific aspects of a customer’s account history, such as payments, a bot can crunch that data faster than a human can,” advised Cantor. Our research shows that 80 percent of organizations say their agents have to use multiple systems to process a support issue, so it would take much longer for a human to process an insurance payout than a bot.

Bots are also useful for augmenting self-service channels because they can “gather more context about customers and pass that context onto the agent, which result in stronger interactions overall and a better omnichannel experience,” Cantor added.

And finally, an important consideration for RoI…

Beyond using chatbots to boost your NPS and customer satisfaction scores, you want to make sure your business is getting a return on automation, says Cantor.

“You want to look at things like self-service and digital containment rate. Were issues being fully resolved in these channels or are customers still screaming to escalate to a live agent?”

Successful chatbot deployment should result in reduced call volume and cost per call, because customers should then be able to resolve issues and have their questions answered on digital channels.

“If bots are handling simple tasks,” says Cantor, “that should change the types of calls you’re getting to your live agents in the call center, and that will have an impact on volume and costs.”