Sogou Is Finally Going Public: 9 Charts on the Rise and Fall of a Generation of Old-School Internet Companies

They got their start with tools for personal computers, then gradually fell behind as the industry shifted to mobile.

Sogou, founded 13 years ago, is finally going public.

Last Friday, Sogou filed its prospectus with the US Securities and Exchange Commission, planning to raise $600 million in a New York Stock Exchange listing at a valuation of roughly $5 billion — about one-fifth of what Weibo is worth today.

It is a milestone for Sogou's founder, Wang Xiaochuan. He once said he would not get married until the company went public.

Going public is not the end; more entrepreneurs than anyone can count have said so. And for Sogou it really isn't the end. To get here, Sogou has told a story much like Baidu's: search advertising plus artificial intelligence.

By 2016, rising traffic costs had pushed Sogou's profit down by more than 40% — and that was with 38% of its search traffic supplied by its investor, Tencent.

Still, this is a symbolic moment, and not only for Wang Xiaochuan, who has been talking about an IPO for years. It is a moment for a whole generation of unusual Chinese internet companies.

360, Xunlei, Baofeng, Sogou. By the time they were founded, the portals had already gone public and the big three — Baidu, Tencent, Alibaba — had taken shape.

At the tail end of the PC era, they used software to win tens of millions of users, then ran straight into the painful shift from the computer to the phone. They missed the Nasdaq dot-com bubble, and the venture-capital boom arrived too late for them. They stood out with products that worked far better than anything else, then chased profit until their names became bywords for bundled software.

Sogou is the last of them to go public.

In the 2000s, as the internet spread, utility software became the door to it

At the start of 2000, the three big portals — Sina, Sohu, and NetEase — were China's largest internet companies, worth about $1 billion combined. Personal computers and broadband were spreading fast across the country, and for most people just getting online, a portal was the first place they went for information.

Robin Li came back to China in 2000 and founded Baidu. Its first business was supplying search to the big portals, charging a fee based on search volume. A year later Baidu began building its own standalone search engine and selling paid search rankings.

The portals soon realized they had raised a competitor in the online advertising market, and they cut Baidu off. All three started search projects of their own, but only Sohu's Sogou is still around.

Baidu went public in 2005. Its stock rose 300% on the first day, making it worth more than the three portals put together. After Google later pulled out of China, Baidu had the search market to itself.

Sogou survived because it had one product nobody else had: an input method.

Wang Xiaochuan has said that Sogou Search was in a weak position by then, and that Baidu's successful IPO had drained whatever enthusiasm Sogou's employees had left. In Charles Zhang's words, the plane had begun to wobble and was about to hit the ground.

That was when they thought of building an input method. Sogou already had search, so it knew what people liked to type into the search box, and that data could be reused as the input method's dictionary. The idea came out of Wang Xiaochuan's technical background: in high school he won a gold medal at an informatics olympiad, and after finishing a master's in computer science at Tsinghua University he joined Sohu, where Charles Zhang told him to put together Sogou.

Once the input method existed, Sogou moved to grow its market fast by working with Tomato Garden, China's largest distributor of pirated Windows.

Most people never got into the habit of memorizing web addresses. Today you tap an app on your phone; a decade and more ago, the easiest way to reach a new site was to remember one search engine, type a few characters into the box, and let it take you there.

Baidu controlled the search gateway. Sogou gradually took over the Chinese input method market — it was on nearly every year's list of software you had to install on a new machine.

Its dictionary was big, it remembered which characters you used most, and it suggested the next word, which made it far nicer to use than the Chinese input method that came with Windows XP. Google China's own pinyin input method was even caught using Sogou's dictionary, and later removed it — a backhanded measure of how good Sogou's technology was.

Sogou was one of a set of programs — Xunlei, Baofeng Media Player, 360 and others — that fixed the same problem: the software built into Windows was too hard to use.

Baofeng Media Player played a lot of video formats and was simple to operate, and once rivals matched it on features, getting there first was advantage enough. Whatever the file was — rmvb, mp4, any other container — "just download Baofeng and it'll play" became common knowledge.

Zhou Hongyi's Qihoo 360 was the last of them to arrive. In his early years he was busy building browser plug-ins of every kind, and a lot of rogue software forced its way onto users' computers without their knowledge, and often against their wishes.

After passing through 3721, Yahoo China, IDG and several other companies, Zhou Hongyi took over Qihoo in 2006 and a year later released a beta of 360 Antivirus. By making it free forever, he quickly seized the antivirus market that Kaspersky, Rising, and Symantec had held.

Sogou, Baofeng, Xunlei, 360 — at the time, each of them made the best product in its category.

But the best tool can't charge you directly, so they all took up rogue software

Feng Xin, the founder of Baofeng Technology, started his career in sales at Kingsoft, Lei Jun's company.

He quit in 2005 to start two companies of his own. One, Kure, made a video player; the other made browser plug-ins and earned 1 million yuan in two or three months. After raising more than $3 million in total from Cai Wensheng and IDG, Feng Xin bought Baofeng Media Player in 2007 and built Baofeng Technology around it.

Herosoft Super Decoder was Baofeng's rival, and it had reached the player market before Feng Xin's company did. In early user reviews, Baofeng had a plain interface and few features, but it was light on system resources and played a lot of formats. Super Decoder had more features and a fancier interface, and it ate resources.

The other word users attached to Baofeng was "universal," because it played almost any audio or video file. For people who did not want to fiddle with software, Baofeng met the need that mattered, and it felt like a real leap forward.

One episode shows how many machines Baofeng was on. On the night of May 19, 2009, China Telecom's broadband network failed nationwide. Officials announced the cause the next day: Baofeng had sent too many requests to China Telecom's DNS servers.

But being the best tool did not mean the company could take money from users directly. A video player cannot sell advertising against traffic the way a portal or a search engine can. So when Baofeng set out to make money, it traded away the experience for ad revenue: it packed the product with pop-ups that slowed the software down, and when users made another player their default, Baofeng quietly changed the file associations back to itself.

Working against what users wanted turned out to be the one thing all of these tools had in common. Advertising, for instance, was Xunlei's main source of revenue at first, and it covered every service and program it owned with pop-ups and page ads — the Xunlei client itself, Gougou Search, Xunlei Kankan, web games, its security assistant and the rest.

Sogou's version was to install the Sogou browser by default along with the input method. It did something else, too: when users typed certain terms into another search engine, Sogou's input method had often already worked out what they meant, but instead of offering it among the candidate words it put the suggestion in a drop-down bar — and clicking it sent the user to Sogou Search.

This was Wang Xiaochuan's "three-stage rocket": the input method carries the browser, the browser carries search, and search is where the money gets made. Every stage pushed users toward the next Sogou product.

Given the choice between growth and remaining a good tool, all of these companies took growth and gave up the user experience.

In the mobile wave, most of the tool companies struggled to make the turn; 360 got the furthest

The world's biggest internet companies today each solved the biggest problem in their field: Google offers the best search engine, and Amazon and Alibaba are the largest retail platforms in the world and in China.

The early internet worked the same way. Between 2000 and 2005, as personal computers and broadband spread across China, the number of internet users grew from about 22 million to 120 million and the number of computers online jumped from 9 million to 50 million. Over those five years the portals first met users who were curious about an internet they did not know, and then search and social gradually took over their attention.

Then, in the gap before mobile arrived, the companies making computer tools caught the enormous opportunity that Windows' poor user experience had created. They succeeded because they were tied so tightly to the personal computer, and 360 grew big enough to take on Tencent directly in a fight over security software.

The "3Q War" between Tencent and 360 broke out in 2010. It started that February, when Tencent launched QQ Doctor, a security product almost identical in function to 360 Safe Guard. 360 hit back with QQ Bodyguard, which modified the QQ software to block its ads, among other things.

On the night of November 3, 2010, Tencent said it would stop running QQ on any computer with 360's security software installed, forcing users to pick one or the other — using the hold it had built up over years on people's social relationships to cut off compatibility with 360. The two companies' software worked together again after the Ministry of Industry and Information Technology stepped in.

The commercial fight stopped for a while and moved to the courts. Each side sued the other for unfair competition and for violating antitrust law, three cases in all. Tencent won its two unfair-competition suits against 360 and collected more than 5 million yuan in damages. In October 2014, China's Supreme People's Court rejected 360's damages claim in its own unfair-competition suit against Tencent, and the 3Q War formally ended.

But the year of the 3Q War was also the year smartphones took off in China. The iPhone 4 sold well there; Samsung and HTC seized most of the mid-to-high-end Android market within the year; and Xiaomi led a wave of Chinese brands into the 1,000-yuan segment. Together the handset makers pushed Android past 80% of China's smartphone market.

Users' attention moved from the computer to the phone and never moved back, and countless internet companies found new life on mobile.

In the single year from 2011 to 2012, Android phone sales in China went from 50 million to 200 million. But Google had already left the country, so this fast-growing Android market had no official app store.

In 2012, 360 launched Mobile Assistant. Then Baidu and Tencent pushed hard into the same market. In 2013 Baidu paid $1.9 billion for 91 Wireless, knocking 360 — until then the app store leader — into second place. A year later, squeezed into third by Tencent's app store, 360 started making phones of its own.

Of all the PC software companies in the world that moved to internet services, Tencent is the only one that held on to its monopoly in its original field. 360 stayed close to users after its own transition, but it never had an app that keeps people looking at their phones for 90 minutes a day the way Tencent's does.

Even after surviving the road to an IPO, what came next was not easy

In the early years of China's internet, download software mattered a great deal: it carried all sorts of material from the network onto tens of millions of personal computers. Most Chinese households were still on 512kbps to 2Mbps connections back then, unlike today, when most people have been upgraded to 50Mbps or 100Mbps.

Xunlei caught the spread of broadband in China, and its faster downloads made it one of the first things people installed.

Its technical advantage was pulling from other servers on the internet. If you were downloading a piece of software from site A and the same file also sat on site B, Xunlei could download from B as well, automatically matching you with whichever server was fastest. What you saw as a user was simply a quicker download.

FlashGet, the market leader at the time, reportedly stopped updating for a year because its founder was lost in World of Warcraft, and Xunlei took its share quickly. By August 2006, three-year-old Xunlei announced it had passed 100 million cumulative users and set a goal of going public in 2008.

In 2008, Xunlei made $11.86 million from advertising, 70% of its revenue. A year later it launched a membership system and formally began charging: paying members got faster downloads and no ads. With that, Xunlei finally had a direct business model, and subscription revenue gradually replaced advertising as its most important source of income.

But the failed IPO of 2011 hit Xunlei hard.

"A lot of capital was interested in Xunlei at the time, but very few were willing to buy in," Cheng Hao said.

Zou Shenglong, another co-founder, said Xunlei began considering an IPO at the end of 2010. In February 2011 its pre-roadshow valuation came back at $1.5-2 billion, in line with what the company expected, so it formally began planning a Nasdaq listing.

Then came a run of events that shook the capital markets' faith in Chinese concept stocks. Longtop Financial Technologies blew up in an accounting scandal, and Jack Ma set off a crisis of trust by privately moving Alipay's equity out of Alibaba and into a company of his own. Xunlei had its own problems: film and television copyright lawsuits, and investors who did not understand its business — the capital markets took it for an online video company.

Its underwriters asked whether it should put the listing off. In the end, so much came to a head at once that Xunlei withdrew its application. It gave up on listing because the market had turned cold on Chinese concept stocks, and that decision became a landmark in the broader chill over those listings.

After 2011, in order to get public eventually, Xunlei began clearing up copyright problems and filtering pornography, and some of the users who cared most about download speed drifted to competitors and substitutes — QQ Xuanfeng, cloud storage, streaming services. Three years later, with its old problems patched up, Xunlei made it onto Nasdaq.

Today, memberships that speed up downloads are still Xunlei's main source of revenue. But the business story it wants to tell now is cloud computing and short video on phones.

Baofeng Technology and 360 did not have an easy time getting listed either.

Baofeng also began thinking about an IPO in 2010, and it chose to list at home. Its investors gave three reasons: Baofeng's users were in China, so Chinese investors understood it and Americans did not; the ChiNext board had just opened and was giving listed companies very high valuations; and Baofeng was profitable, which met the bar for a domestic listing.

Having settled on a domestic listing, Baofeng rearranged its equity structure to satisfy the China Securities Regulatory Commission. Then, after it filed, the regulator suspended all IPOs in 2012. Baofeng could only wait — and keep its profit growing every year.

Baofeng Technology's profit fell in 2013. Under the old rules that would have ruled it out. But the requirement for consecutive profit growth was dropped at the end of 2014, which let Baofeng finally list on ChiNext in 2015.

Zhou Hongyi's 360 listed in the United States and then took itself private. He explained why in an internal email to employees: "360's current market value of $8 billion does not fully reflect the company's worth."

That $8 billion was about 60% of 360's peak value, so it was indeed a good deal lower. But by then the internet had moved to phones, and besides giants like Tencent and Baidu competing with 360 in app stores, the handset makers were running app stores and security centers of their own. On the phone, 360's business model really was in trouble.

Today, they are telling new stories

Each of these companies now has a new story.

"Nobody doubts that Sogou will survive anymore. The only question left is whether Sogou can disrupt search, and whether it can lead major innovation in artificial intelligence," Wang Xiaochuan wrote in an internal letter.

That sounds a great deal like Baidu's story, except for the scale.

Baidu has spent recent years promoting what it spends on artificial intelligence, O2O, and autonomous driving. The stories did not change how the market saw it, and its stock fell sharply during a crisis in advertising growth. Only when ad revenue came back and cash flow steadied did the market begin to accept the story.

Sogou is far behind Baidu both in the number of advertisers it has and in how much each of them spends.

Baofeng Media Player still gets updates, but the company is already talking about becoming an internet television company. Like LeEco, it has used a string of unlisted affiliates to invest in virtual reality and film and television rights, and to sell wealth-management products and small loans. Another ecosystem story.

Xunlei's story is the fashionable one: cloud computing and short video on phones. Its stock trades at a third of its IPO price.

Less than a week after Sogou filed its prospectus, Qudian — a campus installment-lending company just three years old — listed on Nasdaq. It raised half again as much as Sogou plans to, and it is now worth 30 times what Xunlei is.