Zhihu Is Finally Getting Anxious

But isn't anxiety exactly what a startup should feel?

On Monday afternoon, Zhi Hu 3.0 — Zhihu's paid voice Q&A product — went live. Users can pay to put a question to someone else; every voice answer can be played by anyone, and the money from those plays is split evenly between the person who asked and the person who answered.

Zhihu's previous new product, Zhihu Live, was less than a month old. The button in the bottom-right corner of Zhihu's WeChat official account had already been switched from the Zhihu Live column to an entry point for Zhi Hu.

Seven days earlier, Wang Sicong had made 130,000 yuan in a single day on Fenda, a similar app, answering questions from the likes of Muzimei. Without that, this would probably have been hailed as a breakthrough for paid knowledge.

Less than half an hour after Zhi Hu 3.0 launched, a Baidu search for “Zhi Hu” returned a Fenda ad at the top of the results: “Fenda brings together the biggest names from Zhihu, Zhi Hu and Zhihu Live.” For at least half an hour after that, the same ad read: “Fenda was plagiarized by Zhi Hu.”

At the same time, searching for “Fenda” in Zhihu's mobile app and on its website returned no related questions or answers. The posts had not been deleted; they simply did not show up in search.

We asked both the Fenda and Zhihu teams about it. Zhihu's answer was that “there may have been a bug.” A few hours later the search results were back.

Fenda's official reply to QDaily: “We didn't expect Zhihu to take the plagiarism route too.”

The truth is probably more complicated than what either company is saying in public.

Over the past two months, the companies behind these two products have both been shipping fast, and it is hard to say plainly who copied whom, or who borrowed from whom.

How the two companies iterated

April 1

Zhi Hu 1.0 launches. Users can share a Zhi Hu message to social networks with the key information blurred out; unlocking the blurred part costs money

April 16

The Zhi Hu product manager sets out the design thinking behind Zhi Hu 2.0

May 14

Zhihu Live launches. Over one hour, the host answers the audience's questions online with voice, text and images; the audience has to buy a ticket in advance to get in

May 14

Zhi Hu 2.0 goes into closed testing on stage at the Salt Club conference. Users can pay to ask questions, answers are still text, and other users can pay to read the answer to any question

May 15

Fenda launches. The most prominent category on its “Find People” page is “The Zhihu big names you've asked questions of”

May 16

Zhihu holds its first Live

Afternoon, June 6

Zhi Hu 3.0 launches

Evening, June 6

Guokr, Fenda's parent company, releases “Workplace Salon,” a new product from its MOOC Academy. Built on WeChat, it lets one instructor teach many users at once with voice, text and images

In how they are run, Zhi Hu and Fenda do differ. Both charge one yuan to listen to someone else's answer, but Fenda sells it as “eavesdropping,” Zhihu as “learning.”

The resemblance, though, is plain to everyone. Zhi Hu 3.0 looks far too much like the rival that got there first.

Since it was founded in 2010, Zhihu has seen plenty of competitors come and go, from the small Q&A communities of its own generation to Baidu's Zhidao and Tieba. But it has never before collided with a rival product this directly.

Zhihu Starts to Get Anxious

Last Wednesday we met Zhihu founder Zhou Yuan at Zhihu's office. Compared with the interview he gave us a year ago, Zhou Yuan looked a good deal more tired.

That evening he was tracking the progress of two projects that might ship that night (Zhi Hu was not one of them) while also watching Kai-Fu Lee's live Q&A on Zhihu Live on his phone.

Live is part of the “knowledge marketplace” project, which Zhou Yuan runs himself and which has been under discussion since July last year. One host shares what they know through voice, images and text. Listeners have to sign up ahead of time and pay to take part.

Everything around Live follows a playbook familiar from Zhihu's own first years: start by testing with a small group of users, with the Zhihu team guiding how the thing is used.

Live is not open to everyone. The users who see the sign-up button are picked, and hosts are invited one by one. Whether a user sees the button depends on how much they have interacted with the host and with the relevant topics, plus a degree of screening by the operations team and some manual additions.

Before each Live, the host publishes a warm-up piece in their column introducing the topic and the areas they know well.

Afterwards the host writes a second piece summing the session up.

For now every Live host is invited directly by the Live operations team, and the topics — “How to find a good restaurant,” “Three things about art that stump us,” “For those of you planning to get pregnant,” “The operating system of starting a company,” “The ins and outs of investment banking,” “Why nobody cares about your business plan” — cover every corner of the Zhihu community.

Each Live tests a different kind of effect, setting an example for wider use later on.

Zhihu's operations team watches every Live from start to finish.

Halfway through that day's Live, Zhou Yuan got a call from the operations team. They had noticed that Kai-Fu Lee kept answering specific questions from the people in the room and had not got round to the basic questions the outline said he would cover, so they called Zhou Yuan directly and asked him to step in. After hanging up, he quickly placed another call to the team.

By the time the interview ended at 10:30 p.m., more than twenty engineers were still at their desks, and seven more people were meeting in a small conference room. Counting the marketing and PR teams, close to 40 people were working late that night — nearly a fifth of everyone at Zhihu.

A scene like that is entirely normal in the offices of Baidu or Toutiao. At Zhihu it is not. An operations staffer who interned at Zhihu two years ago told QDaily: “Working at Zhihu back then was pretty comfortable. Unless you had a case that really wasn't finished, people left around six or seven. Nobody worked late much.”

It isn't only the late nights. A year ago, Zhou Yuan did not intervene in the product at all; he saw himself as the company's internal communicator.

“I think there's a saying that gets it right: a CEO is better off not running specific projects,” he said when he explained his management logic to QDaily then. “Say you're an engineer and someone from marketing walks over to you. You'll certainly think: this person wants me to build some website, some traffic funnel, something that hurts the user experience. Now I show up and say ‘do it,’ and you'll think, the founder himself came over, so even a hundred ads would be fine.”

Raise the same subject this year and his thinking has shifted: “A lot of the really important things at a company can only move forward if you allocate resources more clearly, so you can't hold back for fear of breaking something.”

He said: “The speed at which a tech company's environment changes basically doesn't allow it to form a loose, democratic structure, because the company doesn't control resources — it drives speed. A situation like that still has to rely on more centralized judgment.”

Where a year ago he cared more about how people inside the company talked to each other, Zhou Yuan now pays close attention to specific product decisions.

As we were leaving the office, he hurried back into the engineers' area. “I should still go take a look,” he said.

Besides Live and Zhi Hu 3.0, the off-app product that launched yesterday, Zhihu also rolled out tipping for columns last week, so readers can support the Zhihu columns they like. The feature may reach Zhihu's Q&A product in time.

Zhou Yuan also mentioned three new products in development that will probably launch soon, each one meant to help users who supply valuable information get paid for it.

In 2014 and 2015, by comparison, the only thing directly connected to that goal was Zhihu's publishing business.

In recent months, Zhihu has started to get anxious.

The Anxiety Is Necessary

Over the past two years the Zhihu community has grown steadily. By the Baidu Index alone, Zhihu has already passed Douban, Baidu Zhidao and even Baidu Tieba, and trails only Weibo — and Weibo itself keeps tilting further toward entertainment.

Zhihu has become one of the country's largest public platforms. After a public event or a piece of news breaks, Zhihu is one of the main places it gets discussed, and the media quote it heavily.

But against that steady growth of the community, Zhihu's new projects have all been slow.

The publishing project has been through several rounds of discussion and still has not turned into a product at any scale.

And the reading products built on Zhihu's brand and content — Zhihu Daily and Dudu Daily — have not worked out as hoped either.

“One starting point for Dudu Daily was that some fairly active editors were already pushing content from outside Zhihu onto Zhihu Daily. We thought content from outside Zhihu could come in this way too, through users' own judgment. That premise holds up,” Zhou Yuan told QDaily. “But to be honest, it will always need a bigger user base before the mechanism really works.”

Clearly Dudu Daily has not found a big enough user base. Even carrying the mantle of Zhihu Daily's successor, and even after Apple's editors featured it on the App Store home page twice, Dudu Daily's download ranking has hovered around 100th in the News category — 70 places below Zhihu Daily.

Just as Zhihu's new products outside the community were all stalling, Guokr built Zaihang, a product for sharing real experience face to face. Users pay to meet an “expert” in person and hear what that expert has actually lived through.

Zaihang has no community behind it and no way to grow big names of its own, yet it has become the channel through which many people who became big names on Zhihu turn their knowledge and experience into money.

When we asked Zhou Yuan why Zhihu — better suited to this, and with more to gain from it — had let Guokr get there first, his answer was: “We think there's an order to these things. The way I see it, building a community that can grow big names, or building a system that lets talent circulate through a knowledge society, is important, and it really does have to be built by someone. It can't appear out of thin air.”

He added: “Take someone like Wang Sicong. You can always find a way to reach him; people like that are simply there. But how does a knowledgeable, more ordinary person get recognized by others on a platform through their own effort, and become known? That has to be built on purpose. We've spent a lot of time on it, and it fits the reason we built Zhihu in the first place. I don't see it as a question of sooner or later. It's something we have to do.”

There is nothing wrong with building the community, or the system, first. Zhihu's community really has matured, and it has grown many people who draw an audience in their own field. Because that community exists, products like Zaihang and Fenda can hardly truly “compete” with Zhihu.

But a problem has been exposed all the same: judging by its recent launches, paying the people who supply information is exactly what Zhihu wants to do too.

Zaihang, a small team, did faster what Zhihu should have done, even if it is unlikely to really threaten Zhihu. Set side by side, though, Zhihu looks slow.

Wandoujia's Collapse Is a Warning

More than the arrival of Zaihang and Fenda, the thing Zhihu should read as a warning is Wandoujia's collapse.

Many of Zhihu's management practices came from Wandoujia.

Zhou Yuan says he once went specifically to ask Wandoujia founder Wang Junyu for advice, and brought Wandoujia's internal management practices into Zhihu.

Someone who joined Zhihu from Wandoujia told QDaily that the two companies were identical, from management style to the tools they used, so settling into the work at Zhihu was almost seamless.

Zhou Yuan's idea a year ago that “a CEO shouldn't run specific products” came from Wang Junyu too.

But Wandoujia seemed to stop working almost overnight.

The company that had just moved into a brand-new office in 2014 has now shut half that office and let most of its staff go. Kaiyan, the fastest-growing new project Wandoujia produced, has been spun out on its own, has taken money from Matrix Partners China, and has the right to buy back Wandoujia's stake later.

A little over a year ago this company — which came out of Sinovation Ventures alongside Zhihu — was the brightest of the startups of its generation.

Between 2010 and 2014, Wandoujia grew quickly from an “iTunes for Android” into China's most important Android app store. After a funding round in 2014 it was valued at as much as US$900 million.

Wang Junyu recently wrote several posts on his personal blog examining what went wrong with him and with the company. From 2012 to 2014, he wrote, he was “in danger without knowing it,” and even once management saw the danger, the team as a whole could not make itself feel any urgency.

One post, on the state of a company and its CEO, borrows a concept from the founder of a16z, one of Silicon Valley's most important early-stage venture firms —

Ben Horowitz — and his distinction between the Peacetime CEO and the Wartime CEO.

A peacetime CEO focuses on painting the big picture and lets the team make the detailed decisions itself; a wartime CEO pays back every slight, even if it is only an ant blocking his way.

Ben Horowitz

At the end of the post he says that in today's competitive environment there should be no such thing as a peacetime CEO: “If I could do it again, there would be no peacetime at all.”

That seems to be exactly what Zhou Yuan is now pushing at Zhihu — the founder back on the front line driving the product, hitting back at competitors directly, in product and in operations.

Zhou Yuan has become a wartime CEO. The question is whether all of Zhihu can switch that fast.

Cover image from Fast Company, Anostu