Li Wenxing's Tragedy Points to What's Wrong With How Internet Companies Grow

He probably won't be the last one either.

More than two weeks ago, 23-year-old Li Wenxing was found dead in Tianjin.

Once again, a user was deceived by false information on an internet service and lost his life for it. Like Wei Zexi, Li Wenxing was not the first person taken in by this kind of information — but it took his death to draw enough attention to force the company involved, and the whole industry, to change.

When this keeps happening, and no one knows which industry it will hit next, it stops being one company's problem.

Parts of the case are still unclear. What is fairly certain, according to an investigation by the education-industry outlet Jiemodui, is this: on May 15, Li Wenxing submitted his résumé on Boss Zhipin. Of the twenty résumés he sent out, only one drew a reply — from a "Xue Tingting of Beijing Kelan."

On May 20, Li Wenxing left Beijing for Tianjin to start the job. After that his behavior on WeChat turned strange, and he borrowed money from friends. On July 8 he made his last phone call to his mother, telling her: "Whoever calls asking for money, don't give them any."

On July 14, Li Wenxing's body was found in Jinghai District, Tianjin. The Jinghai District publicity department later announced that he had drowned. His relatives, though, told reporters that the autopsy found "no food at all in his stomach."

Two people have since told reporters that they, like Li Wenxing, were lured into the same pyramid scheme — but neither knows what he went through, or how he died.

The real cause of death of this young man, who graduated from Northeastern University last year, is still under investigation.

But all of it is bound up with one fact: on Boss Zhipin, Li Wenxing was contacted by someone posing as a recruiter for "Beijing Kelan." The "Xue Tingting of Kelan" who swapped contact details with him there, and told him to come to Tianjin for an interview, did not exist.

Once the case was reported, media and public attention converged on Boss Zhipin.

A loophole let Boss Zhipin vouch for a fraudster

Boss Zhipin was founded in 2013, and its app appeared in the Apple App Store in July 2014. Its most recent funding announcement was a Series C round completed last September, totaling US$28 million, from China Growth Capital, Banyan Capital, Ceyuan Ventures, MSA Capital, Capital Today and Shunwei Capital.

Figures from August 2016 showed Boss Zhipin had 10.65 million registered job seekers and 1.81 million registered "workplace bosses."

By Boss Zhipin's own marketing, the platform's edge lies in "direct recruiting." CEO Zhao Peng coined a word for it, "Direcruit" — hiring stripped of the written tests and interview rounds of the traditional process, with the job seeker talking straight to the boss. That led job seekers to assume that once they sent a résumé through Boss Zhipin, the person on the other end of the chat spoke for a real company.

On Boss Zhipin, a company hiring in the ordinary way goes through some verification: a corporate email address, a business license check.

But ever since the platform launched, Boss Zhipin has kept a rule: if a recruiter posts a single opening and nobody reports it, no verification is required at all. Whatever company and job title they claim goes straight up on the site. And because accounts are free, an impersonator loses nothing by being reported — registering again and posting again costs nothing.

To a job seeker, the impersonator looks exactly like someone from the company he says he works for.

On how responsibility should be divided here, Ju Qinyi, a lawyer at Beijing Weiheng (Shanghai) Law Firm, told QDaily: "Since the cause of Li Wenxing's death has not been determined, set criminal liability aside for now. On civil liability alone, Boss Zhipin, as an information platform, failed in its duty of reasonable care. Its vetting of the qualifications of companies on the employer side had obvious defects. That conduct was at fault and constitutes facilitating a tort, and the company should bear the corresponding civil liability for damages."

Boss Zhipin knew about the loophole all along — but growth was too tempting

Since last year, Zhihu users had been posting, one after another, that they'd been scammed in much the way Li Wenxing was: fake recruiters from software-outsourcing firms luring applicants to interviews in Tianjin.

By that point 18 hours had passed since Jiemodui published its report, six days had passed since Boss Zhipin learned of Li Wenxing's death, and at least six months had passed since users first said online they had been scammed through Boss Zhipin.

On August 2, Jiemodui's report on Li Wenxing's death drew attention. But only that night, when several outlets tested the platform by posting fake recruiter listings — a "Meituan Director of Soy Sauce," or a "Boss Zhipin HR president" whose name was given as "pyramid-scheme boss" — did it become clear that a listing still needed no review whatsoever to go live. The fakes pulled in several résumés within the hour.

In an official statement on the afternoon of August 2, Boss Zhipin's CEO said that after Jiemodui's inquiry on July 28, the company contacted the police at once, met a representative of the family the next day, and was cooperating with the investigation at all times.

But it was not until the night of August 2, after the article ran, that Boss Zhipin changed its review system overnight and announced it would add facial recognition, ID checks and the like.

If the review system could be rebuilt in a single night, why, in the better part of a year since the loophole was exposed, had the company done nothing but ban accounts?

The reason isn't hard to guess: user growth.

When internet companies chase growth, they follow a widely accepted funnel: users find the product, users try it, users stay, users bring in revenue, users recommend it to others.

At every stage, fewer people are left.

The loophole sat at the stage where business users start using the product. To get people using something as fast as possible, internet companies usually cut the cost of the first time: the first food-delivery order is free, the first ride-hailing trip is free, an O2O house-call manicure costs 9.9 yuan. At Boss Zhipin, it meant not checking a company's details before its first job listing went up.

Getting from app promotion to money out of business users takes Boss Zhipin through many steps: a company hears about the product, registers, posts a listing, uses the product to talk with job seekers, becomes a paying customer. Add a verification step at the front of that chain and users bought with real marketing money drop out.

That is why internet companies that put growth first, offered a choice between the fiddly work of completing a company's paperwork and fast registration and use, usually pick the latter.

For a startup, growth in users and revenue is one of the main things investors judge it by.

Raising money takes numbers. Risk, by comparison, is uncertain. A startup has to be bold enough just to live long enough to meet that risk.

In the three-plus years since it launched, Boss Zhipin has raised five rounds totaling tens of millions of US dollars. In an interview with Tencent Tech, Zhao Peng said: "Our investors have always paid close attention to our review system. But out of trust in us, they believed our strategy would work."

The whole industry puts growth before risk, and only a disaster brings change

About a year ago, the death of Wei Zexi shocked the Chinese internet in the same way. He believed a medical advertisement on Baidu, missed his best window for treatment, and died.

Starting in 2008, a scandal over Baidu's medical advertising broke every year or two, and every one of them caused an uproar. But Baidu's revenue growth over more than a decade rested on advertising, and medical ads were the biggest single piece of that business.

It took Wei Zexi's death, and the pressure that came afterward from the government, the press and users, to force Baidu to change.

Launch first, fix it when something goes wrong. For twenty years, countless companies have chased growth down this same path.

In May 2016, a woman teaching in Shenzhen was robbed and killed by the driver of a Didi carpool ride; the driver's license plate was forged.

A few days later, Didi banned more than 8,000 driver accounts in Shenzhen. Inside the company, staff spent the night combing the system for "safety loopholes," and soon after Didi rolled out driver facial recognition, a one-tap emergency alert and other features.

It always takes a death to change the system. The pioneers of the sharing economy were no different.

Airbnb looks today like a reputable company, with a fairly complete set of protections and safeguards. But every one of those insurance measures was built only after something terrible happened.

From its first day, Silicon Valley investors had doubts about Airbnb's safety.

"Stay in a stranger's home? Are you crazy?" That was Paul Graham's reaction the first time he heard about Airbnb — Graham being the founder of Y Combinator, Silicon Valley's most influential incubator — and it is still what many people who have never used Airbnb think.

But what Airbnb learned at Y Combinator wasn't how to make the platform safer or what safeguards to add. It was to send people out to photograph listings and run email marketing, and to grow 800% in a year.

In 2011, an Airbnb host known as EJ came home to find her apartment stripped bare. Getting nowhere with her complaint and her claim, she wrote about it on her blog, and the story kept spreading.

Graham, the investor and Y Combinator founder, advised Airbnb at the time: "People want blood. Don't fight it — admit responsibility, and people will let it go."

Airbnb CEO Brian Chesky took the advice. He apologized publicly and, for the first time, set up property insurance (on the advice of another heavyweight investor, Marc Andreessen, the coverage was raised from $5,000 to $50,000).

By 2011, Airbnb had been running for three years with no insurance for hosts or guests. But just as Graham predicted, people forgave Airbnb — for the apology and the fix, and for the convenience.

Four years after that claim, another guest was crushed to death by a tree falling in the yard of the place they were staying. Until then, the person checking whether an Airbnb listing was safe was the guest checking in.

Accident, payout, improvement, accident again. Through a string of them, Airbnb built the fairly solid protection system it has today.

Uber, under even heavier pressure to grow, hasn't stopped producing scandals. The latest, this week, is a car it leased to a driver in Singapore catching fire. The car was on Honda's recall list. Uber knew, and didn't tell the driver to stop work and bring it back.

Had Airbnb built in safety measures from the start, some people might still be alive. It might also have been held back by cost and by slower growth, and never have gotten this big. Now valued at $31 billion, Airbnb is preparing to go public.

These companies grew up in the end, and made life easier for hundreds of millions of people. For the unlucky few who ran into trouble, the tuition was steep.

Alibaba and Facebook, each now worth more than $400 billion, have never had a fatal incident — but in their early years both reined themselves in only after landing in serious trouble.

Taobao opened in 2003, and it was not until Alibaba's 2014 IPO that it began dealing seriously with counterfeit goods.

How much of the gross merchandise volume from that decade and more came from fakes, Alibaba itself probably doesn't know. In the IPO prospectus, though, the letters GMV sat in the most prominent spot on the page.

Before Facebook went public, founder Mark Zuckerberg had apologized in public at least 25 times, ten of them over leaks of user privacy. Luckily for Facebook, back when it was still fragile, none of those leaks led to anything worse.

Today Zuckerberg is fifth on the global wealth list, and Facebook's offices carry giant bold letters on the wall: "BE BOLD."

Fixing problems after they happen is the norm — but internet speed magnifies them

In the book The Black Swan: How to Deal With the Unknowable Future, author Tom Peters takes up a question: if the FAA had required commercial aircraft to install reinforced cockpit doors beforehand, could 9/11 have been prevented? The answer is no. The proposal would never have won support; it would have met resistance from the airlines over cost, weight and the rest.

Resisting temptation is not easy, and part of that is human nature. We thank the people who solve problems, not the people who prevent them.

That a problem hasn't happened doesn't mean it isn't there. But set certain growth against occasional risk, and choosing growth is the more common choice.

In traditional industries, held back by the pace of growth itself, companies have more time to get their systems right.

Internet companies run on a different clock. Only three years passed between Boss Zhipin's launch and this case.

Its loophole made no difference in 2015. In 2016 it produced a few frauds that nobody paid attention to. By the middle of 2017, a life was gone.

Today's new internet companies — one-on-one recruiting, ride-hailing, food delivery — sit closer to daily life, and can do irreversible damage more easily than the earlier business of selling goods or running a social network.

At the same time, a system built entirely around numbers pushes every decision-maker inside these companies, one choice after another, toward growth rather than safety.

From Wei Zexi to Li Wenxing, when things like this happen it is hard to say which particular person inside the company should answer for it.

In the end, nobody does.

After this, Boss Zhipin may never fully recover, and the recruiting industry may rein itself in a little.

But as long as the chase for numbers stays the same, the next avoidable tragedy is only a matter of time.

Cover image from Visualhunt and Stocksnap