If Customer Service That Can't Solve Your Problem Is a Paper Tiger, What Does Good "New-Era" Customer Service Look Like?

It's all about communicating more efficiently — and once robots step in, the humans' job gets more complicated.

"For Mandarin, press 1. For English, press 2…" You sit patiently through the whole recording, waiting for the line "To speak to an agent, press 0." When it finally comes, you press 0 without a second thought.

Almost everyone knows this scene, and it raises a question about what consumers really want from customer service. Most of the time, people don't want to talk to customer service — they want their problem solved. Whether they have a question, a complaint or something else, wrestling with a machine can waste time, and a few plain words with a real person often get things done faster. Yet when a consumer is absorbed in an internet product, whether an app or a website, they are probably not keen to pick up the phone. In other words, if the product's built-in robot service did its job, no one would choose to wait on hold listening to Für Elise just to press "0."

As services grow more fine-grained (and many needs are created — package tracking, for one. Before that service existed, can you imagine checking where your order had got to, or wondering why it had sat in one place for so long?), the customer service trade can no longer be sorted neatly into "pre-sales" and "after-sales." A more accurate way to put it: whatever meets the user's need for information first and fastest is the form customer service should take.

We put this to Wang Hai, director of Alibaba's customer service division. He sketched the whole history of customer service at the company, which may also be the history of the industry: "We launched a PC customer service system and a help center. Then the mobile era arrived, and users started moving to mobile channels. Open the Taobao app and the customer service entry is easy to find." And: "Alibaba realized that service is part of traffic, and that over the next ten years, experience will be a key factor across all of e-commerce."

When Trip.com Group expanded its call center in Nantong in 2011, it probably didn't expect the industry to change so fast. Only four years on, the world's largest travel call center — a grand 80,000-square-meter complex with 12,000 service seats — has added something new: WeChat groups, each run by a "micro tour leader."

Here's how it works. When a group of travelers buys a Maldives package, the micro tour leader texts them an invitation to join a WeChat group. There, they can ask about anything, from Trip.com Group's products to things to do at the destination. Tong Lin, a frontline employee who answers questions as a micro tour leader, told QDaily that he runs about 20 groups. In peak season (Chinese New Year) each group has more than a hundred people; the rest of the year, 30 or 40.

Trip.com Group has fewer than 100 micro tour leaders in total. Yu Xiaojiang, COO of the company's travel division, says phone agents still make up 70% of its customer service, with the other 20% to 30% working in other formats — though the line between them is blurring. "(We're) trying to train all-round customer service staff. The market and our agents both need educating."

Across the industry, what companies spend, the products they build and the ways they talk to consumers are all changing. Efficiency is still the goal, but customer service now carries more of the brand — which is why companies want "robot customer service to feel like a flesh-and-blood person." Whether they use machines or people, companies want consumers to feel their personality while getting a problem solved. If "details decide success or failure," this may be the detail that matters most.

We broke "new-era customer service" down into several features and the reasons behind them. As you'll see, these changes are coming faster and faster.

Most of the Time, Robots Can Handle Efficiency

Intelligent customer service robots do for online service what voice menus did for phone lines. They usually take the form of an interactive help center, so that users who can't find the FAQ don't tie up human agents. Big companies with their own engineering teams tend to build their own robots: Taobao has AliMe, and JD.com has what it calls the JIMI intelligent service system. According to Zhao Pengyi, JIMI took JD.com three years of iteration and an 80-person R&D team to build, and at peak times it handles as much as all of JD.com's human agents combined.

Zhao joined JD.com in 2011. He started as a frontline agent, has worked his way up over eight years, and now directs the Chengdu branch of JD.com's national customer service center. "In 2010 our phone answer rate was a real struggle. We doubled our staff every month, but user demand grew tenfold every month." In those wild early days of online customer service, big companies badly needed new ways to close the gap between demand for service and their capacity to supply it. Pairing a customer service robot with a help center, so users could "look after themselves," became one of the main ways to divert the load.

Take the extreme peak of Double 11, when one agent may face 30 customers at once — far more than a merchant's own staff can handle. Here the robot's value is simple: a merchant fills its shop information into a standard template and gets a customer service robot of its own.

Robots matter in a deeper way, too, one tied to the familiar idea of "Internet Plus." Suppose a passenger hails an Uber, something goes wrong, and after the trip is canceled the app still charges them. There's no customer service number to call; the right way to dispute the charge is to open "Help" in the app, choose "incorrect fare" and submit. Before long, a refund email arrives. Uber's official explanation: "Passengers can complete most disputes and questions in the app. Only the last step — the refund — needs one of our staff."

By breaking every service scenario into its own category, companies go a step further in simulating how a problem gets solved. On Dianping, there's one button to locate a restaurant and another to chase a late order. Many e-commerce sites put special effort into their review systems; you can even quickly put a question to a stranger who has already bought the product, without the old routine of adding them as a friend.

Meanwhile, the Humans' Job Is Getting More Complicated

Wang Lijie, a customer service supervisor at Momo, told QDaily: "You have to keep learning to keep up. If I took a long vacation, I couldn't keep pace with two version updates a month." Tong Lin, the micro tour leader, puts it this way: "Solving travelers' questions about what to do at a destination is surely worth a bit more than being an online agent who issues invoices." He covers Singapore for Trip.com Group. Questions guests can look up online no longer reach him — in other words, what he needs to know is more obscure, and there is more of it.

China has more than 10 million customer service workers, and how society and companies see the job is changing. A "typing speed of 60+ words per minute" may be only the baseline. As JD.com's Zhao Pengyi puts it: "The agent's role has moved from chasing your order, processing your refund and finding out why it's late, toward becoming a service expert and a product expert."

As machines take over the simple tasks year after year, the questions that remain are harder. "What size should I get?" "Are these Aksu apples from Xinjiang really sweet?" Questions like these now go to online agents. Answering them can also turn into sales: what agents are really offering is pre-sales guidance — explaining products, steering purchases and easing shoppers' distrust of buying online.

Staff at Apple's Genius Bar also take turns answering phones at the call center, so that front-line and back-end technical skills stay in step. People rate AppleCare highly because its agents really know the product. Frontline agents understand what users need better than anyone, and they quickly get exposure to the company's entire operation. So when outside hiring can't meet a company's needs, customer service can supply its junior managers. JD.com, for instance, hires several hundred "service stars" each year, and after a little more than a year they can rise into management in its operations system.

It isn't only that agents now handle harder problems from consumers. Because customer service sits closer to product development, the department itself has started to matter. In a media interview last September, Dai Shan, Alibaba's chief customer service officer, described a new feature in the group's internal app called "Nine O'Clock Radio" — in essence, a way to flatten a huge organization. When agents spot a pattern across many customers, they can press a button called "Ring the Bell" to share the case on Nine O'Clock Radio, where colleagues in the business units can see it. If a suggestion is important enough, it may end up as something the product team fixes. "Our thousands of service staff don't just solve problems. What they learn can be fed to every department as content," says Wang Hai.

In Social Customer Service, Language May Matter Most

"JD.com Mutual Help" is a typical example of social customer service: it uses some users to meet the needs of others. The product, still in testing, focuses on the mother-and-baby category. JD.com sets up public discussion topics (about a particular infant formula, say) and cultivates expert shoppers. According to figures from Zhao Pengyi, a phone agent can serve only one customer at a time, and an online agent six to eight at most, but a single topic reaches more than 3,000 users on average. "Some post actively, others just read quietly. Purely in service terms, it's very effective."

The language customer service uses (often the language a robot has been set up to use) has one main purpose: to bring the company closer to the customer. A Weibo influencer called "Zhou Xiaoshuai's Private Kitchen" has a million followers and sells crayfish as his signature dish. Yet his verified Weibo label reads "well-known comedy and humor personality," not "crayfish seller." Scroll through his feed and most of it is chat logs with customers — mostly funny exchanges and jokes, each reply signed off with an "annoyed" emoji. That is how he markets himself.

In late January 2016, the flagship store of a bathroom-fixtures brand went viral after fielding questions from 438 Chinese students studying abroad. Students had thrown a problem from the Mathematical Contest in Modeling, an American college competition, about a bathtub filling with water at the store's customer service. As more and more netizens joined in to tease the shop, its auto-reply became: "Dear, standard bathtubs run from 1.5 to 1.8 meters long and are 2 centimeters thick. Going by the students' answers, the heat transfer coefficient is about 0.19, and the temperature drops by about 5 degrees an hour… So many students are asking that replies may be a little slow."

Playful customer service language still shows up mainly on social media; once the number of users gets large, companies tend to play it safe. Why do these gimmicks sell? In the end, people love the convenience of online shopping, but they also miss the warmth of dealing with someone face to face — and on the internet, that warmth is always in short supply.

Big Data Should Help Build the Solution

If you've ever called SF Express or McDonald's, you'll know they pretend to know you, keeping a record of things like your address and your usual order.

In fact, big data is best used to track users' more complex behavior. JD.com's user-profiling system, for example, sets priorities from a user's activity log. Say you've already tapped the button to chase an order. If the system then gets a call from your registered number, it first assumes the call is probably about that order, then checks the order time to see whether it's late. If it's on schedule, the call goes to an online service robot or a regular agent; if it's overdue, a senior agent who specializes in difficult cases gets ready to take it.

Then there are the credit records a platform builds up. In customer service, credit accumulated on a trading platform can reduce the risks of shopping online. Wang Hai, director of Alibaba's customer service division, described a mechanism that has been running for nearly two years: "If your credit rating is high enough, the money is back in your account the moment you start a return." Where an instant refund applies, a good credit rating lets customer service skip steps: pay first, sort out the rest later. The same goes for merchants. When a merchant with good credit gets a malicious complaint, the platform asks the consumer for more evidence.

Customer Service Doesn't Always Want to Solve Your Problem

We'll end with a classic story.

In 2006, Joe Nocera, a New York Times columnist, wrote: "Customer support is the orphan of the consumer electronics industry." He had a point. For a time, companies wanted you to leave them alone once you'd bought their product, so they handled problems as cheaply as they could. No American consumer would have been surprised to hear an after-sales operator with an Indian accent answer, "Hello, sir."

But Nocera had another insight.

His iPod had broken, and he was determined to get it fixed.

After a great deal of effort, he found the customer service number — and received the sincere advice to "buy a new one." He kept at it and eventually found he could fill out a form and mail the device back for repair. Apple's price: US$250, plus tax.

In other words, Apple had no interest in repairing it at all. Nocera explained why by quoting the investor Andy Kessler: "A phone call costs a company 75 cents a minute — US$45 an hour." Then he added: "Because a music player costs so much less than a computer, Apple has an even stronger incentive to keep people from calling. One long phone call can wipe out the profit on a music player. It makes no economic sense to repair one, even under warranty. Apple simply sends you a new one."

The story appears in his book Good Guys and Bad Guys, which looks at the hidden economics of corporate customer service.