Robots can be scary.
They sometimes feels like a barrier to real customer service. Such as when you find yourself yelling “Human! Human! Human!” into the phone, only to hear an annoyingly pleasant robotic voice respond, “I’m sorry, I don’t understand.”
Other times, automation can create service failures. A bot once tried to send me to the wrong airport when it re-routed me after a weather delay. Another bot hilariously joined a Tweet chat and quickly became confused.
Of course, some automation is inevitable. We use the Starbucks app to skip the ordering line. A self-service portal allows us to troubleshoot our devices without calling support. Automation even sent this blog post to you if you subscribe via email.
Will automation eliminate people?
I recently attended Zendesk’s user conference, Relate Live, where I learned about one bot that’s actually helping people be more human.
The Zendesk Answer Bot
Customers often prefer self-service.
This is especially true for easier issues, such as resetting a password. A challenge occurs when customers can’t get the answer they want via self-service so they grudgingly contact the company for live support.
Zendesk’s Answer Bot tries to save customers from that extra step. We can see how it works with this example from MailChimp, a marketing automation service that uses Zendesk to power its customer support.
Let’s say I want to use MailChimp to automatically send new Inside Customer Service blog posts to email subscribers. I search the MailChimp support site but can’t figure out how to do it, so I decide to send an email.
Once I hit send, Answer Bot jumps in and scans my email. It helpfully suggests a couple of articles based on what I wrote:
Ah ha! The second article is exactly what I’m looking for.
Now I can cancel that support ticket. No need to wait for a live person since Answer Bot already solved my problem. I can just click on the article and walk through the how-to steps.
So how does Answer Bot help humans be more human?
It was a presentation by Brian Crumpley from Dollar Shave Club at Relate Live that helped me see Answer Bot’s true potential.
Crumpley shared an analysis of Dollar Shave Club email interactions. His data revealed that 40 percent could have been handled via self-service. Even worse, these interactions cost a little more and satisfied customers a little less than transactions that truly needed the human touch.
The company wants its agents to have great, personalized interactions with members. It’s tough to do that with more transactional issues. The customer generally wants a quick answer and to be on their way.
High volumes also make personalization difficult. If there’s no budget to add extra staff, agents find themselves racing through contacts just to keep up.
Enter Answer Bot.
Dollar Shave Club implemented Answer Bot to help deflect some of those self-serviceable contacts. After six months, Answer Bot was handling nearly 5,000 contacts per month that would otherwise have gone to an agent.
Those contact deflections gave Dollar Shave Club some extra capacity without adding staff. Here’s how Crumpley was able to use it:
- Expanded live chat availability
- Created a knowledge base task force to further improve self-service
- Debuted a Customer Insights magazine
Best of all, agents now had bandwidth to provide a bit more human service to customers who really needed it.
Forrester predicts that robots and artificial intelligence (AI), collectively referred to as “bots,” will replace 7 percent of US jobs by 2025.
I’m hoping for a different trend.
History tells us the proliferation of automated teller machines actually led to an increase in bank tellers. The teller role just became more sophisticated since they were called upon to handle fewer simple transactions.
I noted this is a 2016 post:
- Robots = good at simple transactions
- Humans = good at complicated transactions
Other customer service roles can follow a similar path. This should free up humans to be more human when customers really need it.
Hi Jeff, interesting insight. I wholeheartedly agree that bots (AI in general) need to assist in order to enable humans to work more on what they are real good at: The complex things. My next column article will actually touch this topic, too. In the light of your two examples I have a little question, though: Given the still limited generalization abilities of AI, where do you see the current boundary? How about bot swarms (maybe even coordinated by a bot)?
@Thomas, it’s a good question. I don’t have the technical foundation to answer it (my focus is on employee performance). What do you think?
Hi Jeff, well, being a techy at heart I think that bot swarms are currently the way to go. But they still are fairly limited in their abilities. I like what companies like agent.ai are doing but am not sure how sustainable this approach is.
On the other hand with Face ID we now get AI/machine learning down to the devices, so maybe the network becomes the ‘brains’?