By Jim Grey (about)
I enjoy the Signal v. Noise blog from Basecamp. Recently they published a post about the human touch in customer service. Read it here. Its thesis, in short: automating any part of support is akin to saying the company doesn’t care.
I am charmed by the work ethos Signal v. Noise espouses: work at a sustainable pace on things that are interesting to you, and get plenty of rest. It’s very appealing. But sometimes I think they take the contrarian position as if it’s their brand, and this is one of those times.
There’s nothing wrong with a good robot providing routine customer service. As long as the experience is good, and can seamlessly hand off to a human when things go off script, it’s a win.
That’s because so much of customer service work can be rote and repetitive. One of my career stops was at a company called MOBI, which provided mobile-device management services to large companies. One customer was a global manufacturer that had something like 10,000 mobile phones deployed to its executives. MOBI helped them manage cost by keeping each phone on the least-expensive plan for its typical usage, and also provided a help desk with the promise of answering a request in 30 seconds and staying on the case until it was solved. (Try getting that from your carrier.)
But it was too expensive for MOBI to hire a bunch of new customer-service reps every time they took on a new customer. It cut the profit margin too thin.
Moreover, a large percentage of the customer-service work was simply boring and repetitive. After you’ve done a handful of password resets, equipment orders, and plan changes, you can do them in your sleep.
MOBI has devised a number of robots (“Mobots,” they call them) to take away the drudgery. Read about them here.
If a MOBI customer user jumps on a chat, Mobot Audrey answers. She can handle many of the common, repetitive tasks. The minute the customer asks something she can’t handle, she seamlessly transfers the chat to a human, who handles the case from there.
I had lunch a few weeks ago with MOBI’s CTO, my former boss, who told me that the customer-service team was apprehensive at first, but is happy now because they are working on things that tap into their deep knowledge of the product, and that require them to solve problems creatively. As a result, they are more engaged with their work.
This resonates with me from another perspective. I was on the team that built the call-center software that Medicare customer-service reps use nationwide. I led the test team. Our contract required us to do a “full” regression test on every release — meaning we had to run thousands of written test cases. It took almost 24 person-weeks to execute those tests! This was about 15 years ago when automation tools weren’t great. But I hired a couple engineers to automate those tests anyway, and they finished after a few months. It took just 0.6 person-weeks to execute those tests forever after. My functional testers were free to spend more time performing targeted, complex, and exploratory tests, using their experience to find critical defects the automation couldn’t find. It was much more interesting and meaningful work for them.
But back to customer-service robots: we all remember how bad the early ones were. That Medicare customer-service center had an early voice-response robot for callers, and it was awful. It didn’t understand callers right most of the time, it took a long time to navigate, and once navigated the caller found that the robot couldn’t help them and they needed to speak to a human anyway. Callers hated it.
But the technologies have begun to mature. Some of them claim to be able to learn, even, although I’m skeptical that it’s true, full-on AI or ML. At least it’s possible to create a good experience now. MOBI’s Mobots have done it; their user testing bears it out: most users have no idea that they are interacting with a Mobot.
So why wouldn’t a company like MOBI manage support costs and increase employee happiness by automating away the boring, repetitive tasks? I’m for it.