Baidu in the past year

Published: 2017-01-03

Preface

This article is the first in the business commentary section of my personal blog. Why choose this topic? It's simple: as my first business critique, I needed a wide-ranging subject with abundant material and strong topical interest, so I naturally thought of BAT — a major IP that occupies tech media headlines for nearly half the year. I didn't start with Tencent because I plan to read Wu Xiaobo's new book "The Tencent Story" before writing. I didn't write about Alibaba because although I've read many stories, my depth of understanding isn't sufficient yet. I chose Baidu because the past year left a bitter taste — speaking about Baidu feels like an unresolved lump in the throat that must be expelled.

Baidu in the past year

The name “Baidu” comes from a line by the Southern Song poet Xin Qiji, written eight centuries ago: 'Searching for him thousands and hundreds of times among the crowd.' The line describes the poet’s persistent pursuit of an ideal.

In December 2015, Baidu announced a major corporate restructuring and formed a new Financial Services Business Group, appointing Baidu Vice President Zhu Guang as its general manager. Zhu Guang joined Baidu in 2008 and had been responsible for Baidu’s mass-market, public relations, and government relations teams. Deeply immersed in China’s internet ecosystem, he understood its unwritten rules and had overseen several public relations cases regarded as textbook examples. After this adjustment, the PR team was handed to Baidu President and GM of Emerging Businesses Zhang Yaqin. Zhang, with a corporate and technical background and a gentle, scholarly demeanor, had just joined Baidu from his prior role as Microsoft’s Senior Vice President and Chairman of the Asia-Pacific R&D Group. At this time Baidu also reaffirmed its strategic shift from connecting people with information to connecting people with services, aiming to be a service-oriented company centered on AI and big data, with finance, O2O, mobile social, and social-commerce as core businesses. Tieba (Baidu Post Bar), as the only social product within Baidu’s product framework, presented infinite possibilities for realizing that strategy starting from the “people” side. This gave new hope to those watching Baidu — which had lost some of its halo in recent years — and in 2016 a “new Baidu” seemed about to emerge.

In January 2016, as the New Year bells had barely stopped ringing, users on platforms like Zhihu and WeChat began reporting that Baidu Tieba had sold operational rights for several disease-related forums — including one for hemophilia — to so-called “experts” and questionable private hospitals or third parties. These forums contained fraudulent and misleading medical information that seriously misled patients. Baidu was swept into media and public criticism, with CCTV airing a hard-hitting special entitled “Ask Baidu, How Many Consciences?” On January 12 Baidu announced it would suspend commercial cooperation across all disease-related Tieba forums and only open them to authoritative public welfare organizations; the hemophilia forum would introduce the NGO “Home of Hemophilia” as its new moderator. However, the subsequent rectifications did not meet public expectations: moderators of forums like Diabetes Tieba remained of unknown background; a forum dedicated to dubbing was bought by an online game company and overnight transformed into game-related discussion, effectively ruined. Many high-quality forums related to finance, education, and regions also saw original moderator teams removed and replaced by parachuted personnel.

In May 2016, four months after the hemophilia Tieba incident, Baidu again became the target of harsh media criticism. A 21-year-old student named Wei Zexi tragically died from synovial sarcoma. He had been treated at Beijing’s Second Armed Police Hospital, but the department had been contracted to a Putian-affiliated private hospital. That hospital used a treatment it knew to be ineffective, charging Wei nearly 200,000 yuan before he passed away. The treatment choice was driven by promotional information obtained through Baidu search’s paid ranking results.

In July 2016, just two months after Baidu CEO Robin Li was summoned by the Cyberspace Administration, The Beijing News reported that searching keywords like “New Lisboa” after 10 p.m. often returned gambling websites among the top promoted results. These gambling sites, which had illegally used various companies’ business credentials, would appear late at night and be taken down before 9 a.m. the next morning, restoring natural search results. Baidu responded that those sites opened accounts in April and that the gambling content was the result of businesses privately and irregularly changing content late at night. But The Beijing News reported that people had discovered and reported the issue as early as December 2015, and seven months of complaints produced no result.

In the months that followed, incidents kept surfacing: pornographic content on Baidu Cloud, fake meal-delivery scandals, two Cyberspace Administration penalties against Baidu, and agency representatives visiting to demand explanations. In short, 2016 was a hectic year for Baidu and one under intense scrutiny. The Wei Zexi case in May epitomized this, its breadth and duration unprecedented — Baidu surely ranked high on hot-topic lists that year.

Losing its standing in the ecosystem

For a period, China’s rapid economic growth fueled the internet’s rise, producing a fiercely competitive, fast-growing industry that eventually crystallized into a tripartite dominance: Baidu, Alibaba, Tencent — popularly called “BAT.” This shaped the current order and rules: the Matthew effect and the jungle law. But over time, changes within BAT and waves of new startups have begun to shake that old BAT configuration.

An employee at Baidu once joked that BAT had gradually become TAB, and then ATM — with “B” dropped from the lineup — reflecting external doubts about Baidu. Market dynamics show this shift: Tencent, once trailing, is now favored by capital with a market cap around $231.6 billion; its diversified early bets are scaling into profitable businesses, especially in gaming and social, making it the industry leader and a magnet for investors. Alibaba, while less dazzling than before, remains steady and resilient with a market cap roughly comparable to Tencent at about $224.7 billion. Baidu, once China’s top internet company, peaked on November 28, 2014 with a stock price of $251.99 and a market value of $86.8 billion. Two years later its market cap fell to roughly $58.4 billion — about one quarter of Tencent’s at the same time. What caused Baidu’s loss of stature? I’ll explore the root causes through Baidu’s turbulent 2016.

The dispute over the profit model

As a mature domestic internet leader, Baidu’s revenue model had become clear and stable, but a comparison of BAT revenue composition exposes a structural flaw. Compared with Tencent and Alibaba, Baidu relies excessively on online marketing. For a company of Baidu’s scale, dependence on a single revenue source increases operational risk. Baidu’s reliance proved damaging: the paid search model faced intense public criticism, threatening this major revenue stream while long-term investments in big data and AI remained cash-negative. That left Baidu in a difficult transitional state — historically low and confused.

Baidu’s online marketing revenue primarily stems from its paid search and the rebranded but essentially unchanged Phoenix Nest system. Google, also search-born, likewise earns a large share from advertising. Yet Google’s search hasn’t drawn domestic scorn; instead it’s often praised in comparisons with Baidu. Here’s a closer look at differences in the Baidu and Google monetization models.

For both Baidu and Google, ad revenue is the main income source. They aren’t fundamentally hardware or software companies — they are ad companies. However, their methods of selling ads differ significantly.

On Baidu, search results pages usually show two parts: the left side is organic search results and the right has sponsored recommendations. The left-sponsored results are a domestically pioneered performance-based promotion model, charging per visit. In short, advertisers bid on keywords: higher bids may yield higher placement for that keyword, driving clicks and better ad results. The right side is the “Baidu Hot Zone,” also keyword-bid based but charged as fixed-duration fees (e.g., annually) with different rates for different positions.

Google’s main ad model, Google AdWords, sells keywords to the highest bidders and places ads on the right of search results ordered by price; advertisers are charged only when their ads are clicked. The left-side organic rankings are still determined by metrics like visits, popularity, and PageRank, so ads do not interfere with natural results, preserving search fairness and impartiality.

Google leverages strong data analysis to deliver more precise ad targeting and thus higher revenue. Data analysis requires raw material like user behavior data. For example, Google Maps collects massive user data that reveals who lives in an area, what they consume, and their spending power — invaluable for advertisers seeking better targeting. Advertising pioneer John Wanamaker once said, “Half the money I spend on advertising is wasted; the trouble is I don’t know which half.” Precise targeting addresses that pain point, and Google solves it, which explains its scale. That’s why Google provides many free services: more users provide more data, enabling precise predictions of tastes and needs, which drives effective ad targeting and more clicks, translating into ad spend from advertisers. In turn, Google provides timely, relevant product and service information to users and high ad efficiency to advertisers, completing a beneficial three-way loop. Baidu also invests continually in big data, but it lost moral high ground due to the paid-search bidding model.

If you want a more detailed, in-depth understanding of Google’s ad system, you can read the Zhihu discussion“How does Google make money?”Sim Chen’s answer there cites portions of Wu Jun’s book The Wave at the Crest in a clear and thorough way — recommended reading. Google’s polished, top-tier monetization model wasn’t perfected by Larry Page and Sergey Brin at the company’s founding; much of its refinement was driven by external regulation, which I’ll discuss in the final section. Now let’s examine Tieba’s monetization model.

Baidu set a strategic goal in 2010 to reach 100 billion yuan in revenue within ten years. With online marketing growth plateauing, Baidu sought to diversify revenue. Tieba — Baidu’s only social product — needed to monetize after years of operation. Baidu’s search monetization relied on agency sales, so Tieba naturally followed a similar agency approach. Commercializing Tieba inevitably created conflicts between original forum moderators and the appointed agencies, and worse, Baidu failed to adequately regulate post-commercialized forums, which led to the Tieba incidents.

Over years of development, Tieba achieved proud results: more than 1 billion registered users, over 300 million monthly active users, and 19 million topic bars. Tieba’s daily active users and visits surpass the combined totals of Tianya, Zhihu, Douban, and Renren. It aggregates many precise, segmented user groups, so many high-traffic forums inherently possess marketing value.

Since 2010 Baidu has pursued its 10-year, 100 billion-yuan revenue target. With slow online-marketing growth, Baidu tried to adjust its single revenue structure and find other profitable avenues. Tieba, as Baidu’s only social product with a solid user base and strong stickiness — and with Robin’s long-held affection for Tieba and hopes for social commercialization — needed to monetize after years of operation.

Tieba monetization was imminent. In 2014 Baidu proposed a commercialization strategy, and many companies began taking over relevant forums, including NetEase Games, iQiyi, Tmall, and JD. But commercialization moved slowly. Like search, which monetized through agency sales, Tieba adopted a similar agency model. In 2015 Baidu piloted a “Tieba Partner System” intended for shared management by commercial institutions and community users. Previously, forum moderators were user-appointed: Baidu assessed applicants’ activity and influence before granting moderator rights. A forum’s quality was closely tied to its moderators. After the partner system, moderator appointments, ad displays, and management authority were outsourced to agents. The problem lay in this blunt outsourcing: agencies favored commercial interests, eroding communities painstakingly built by original moderators. Consequently, original moderators and users understandably rose up in protest. Thus, Tieba’s core problem became a conflict of interest between original moderators and appointed agencies; an over-hasty commercialization strategy lacked corresponding regulatory operations, causing the 2016 Tieba incidents. The ensuing management chaos and trust loss will make future commercialization far more difficult.

Executive turmoil

I briefly introduced Baidu’s December 2015 restructuring earlier. The most notable changes were Zhu Guang leading the new financial services group and Zhang Yaqin taking over PR and government relations — meaning the team’s primary leader changed for the first time since 2008. Soon after came 2016’s series of crisis cases. Shortly after the Tieba incident, Robin Li said in an internal communication: “We’ve done poorly in operations, product, and market PR — especially this PR response. It wasn’t timely, resulting in a major crisis; public opinion ran away with things and our voice was drowned out.” Clearly the leadership chose to let the PR department take the blame for this mess. Zhang Yaqin, inexperienced in PR handling and less than a month on the job, was overwhelmed by a sudden, cliff-like PR crisis. During the peak days of the Tieba incident, the PR office’s lights burned late every night as staff worked overtime trying every measure to mitigate the negative impact.

Below is an organizational chart after Baidu’s major restructuring:

By 2016 Baidu had dismissed or taken over 30 employees in a year, including an astonishing number and level of senior executives that shocked the industry. In April 2016, E-staff Vice President Wang Zhan was fired for harming company interests. In early November, another E-staff member and once-promising “Baidu prince,” Vice President Li Mingyuan, resigned amid allegations that his involvement in an acquisition, his management-defined business scope, and personal external investments may have harmed company interests. Coupled with prior internal corruption cases, the public perception became that Baidu product lines touching money — Tieba, games, enterprise sales, group-buying, food delivery — had become breeding grounds for corruption.

Quoting a report in Computer World, internally Li Mingyuan was seen by many as a casualty of factional infighting with Xiang Hailong — more bluntly, a result of a failed clash between Robin Li’s faction and his wife Ma Dongmin’s. An internal employee said even Wang Zhan didn’t have economic issues raised in internal emails; internal notices were distributed only to director level. Li Mingyuan’s case, superficially lenient, actually stamped him with an indelible exile. Such internal power struggles had become commonplace across Baidu. “When even employees nervously guess about internal infighting and it becomes the norm, the company is not far from death,” a Baidu employee said sadly.

In September 2016 Baidu announced Zhang Yaqin’s reassignment away from PR and government relations, replaced by former 1Haodian CEO Wang Lu. In the same month, Baidu strategic adviser He Haiwen (Helen) quietly left. A year of quiet-then-sudden major personnel changes invites speculation about hidden intrigues. When the opportunity for rent-seeking grows unchecked, corruption and infighting naturally emerge. Are these issues solely the fault of these professional managers? For Baidu — a company deeply marked by Robin Li — one cannot avoid attributing some responsibility to this central figure.

Clinging to the throne

Twenty years ago Robin Li wrote in Silicon Valley Business Warfare: “Don’t cling to the company you founded — this is a lesson learned the hard way. Not everyone who conquers a domain is suited to rule it forever.” Twenty years later, he must confront his own warning.

Robin Li was born in Yangquan, Shanxi, and is one of the Shanxi merchants his province proudly celebrates. As the only son among five children, he was treated as the family’s protected child and received extra attention. Through effort he entered prestigious schools like Peking University, went to the U.S., and within a few years became a nationally recognized entrepreneur. That upbringing and repeated success cultivated pride: he often appears arrogant, stubborn, and headstrong — once decided, he pursues things single-mindedly. That personality produced the distorted paid-search profit model, which he defended amid criticism because he considered it something he had forged in China, his “child.” His self-centeredness made product strategy unpredictable: he could cut products abruptly; employees wrote to him seeking explanations but received none.

For example, in mid-2015 Baidu announced a 20-billion-yuan investment to build Nuomi (Baidu’s O2O platform). The news was interpreted by media as Baidu pivoting to O2O to connect LBS and payment services. Yet in April 2016 Robin Li announced a shift of Baidu’s business focus toward autonomous driving and AI; whether O2O would stay or go remained unclear. In May’s Wei Zexi storm, a post titled “Former Baidu Employee: Why Baidu Can’t Let Go of Medical Information” hit tech headlines. The author pointed to Robin Li as primarily responsible: “Factory director Robin Li now finds it hard to hear criticism or outside voices; his first reaction is often ‘I haven’t done anything wrong — they’re smearing me!’” That turned Baidu’s “wolf culture” into a flattery culture: after negative news, exec chat groups sounded defensive rather than reflective, sometimes declaring loyalty to reassure the “factory director.”

Baidu became almost indistinguishable from Robin Li himself. Over the years public figures like Interactive Encyclopedia CEO Pan Haidong, writer Han Han, Robin’s Peking University classmates, and Zhou Hongyi have sent letters or publicly criticized him, but he remained unmoved — never responding, never arguing — as if living in his own world, like a spoiled child. Of the seven founders at Baidu’s inception, since 2010 only Robin Li remained. Later founding veterans and professional managers who stayed beside him had joined as early as 2004. That left him isolated and made it difficult to find long-time confidants to share his inner thoughts.

A company that grows from a startup needs, besides an outward-facing leader, reliable insiders to handle internal affairs. For example, Pony Ma has Zhang Zhidong and Zhang Xiaolong; Jack Ma has Joe Tsai, Peng Lei, and an inner circle. Baidu, although it still has people like Zhu Guang, Zhang Yaqin, and Xiang Hailong, lacks a clear second-in-command; everything is under Robin Li’s control. Baidu now looks like Robin Li and his tens of thousands of employees. In a company that large, one person inevitably overlooks things and needs senior employees who treat the company like their home, attending to detail. Because early co-founders left, entrepreneurial culture dissipated; people under Robin Li now focus on targets, KPIs, and bonuses, not the dirty, tiring “housework.” With cultural and value collapse, KPIs and data became the most direct communication tools between Robin Li and employees. The fall of cultural values and beliefs may be the fundamental reason behind these problems.

Encouragingly, Robin Li’s statements over the year indicate some rethinking. After the January hemophilia Tieba incident, a screenshot purporting to quote Robin Li circulated online saying: “Baidu’s values are good, noble even, and I believe no commercial model is perfect; Baidu’s model is fine.” But after the May Wei Zexi incident and the Cyberspace Administration meeting, Robin Li wrote internally: “If we lose user support and lose our commitment to values, Baidu would truly be bankrupt in 30 days!” He began to recognize that “the pursuit of short-term KPIs has squeezed and warped Baidu’s values, driving it further from users.”

Failures in political and business environment

Stepping beyond Baidu, in the Wei Zexi tragedy we must also condemn the fraudulent Putian hospitals and the heavily credentialed Armed Police Hospital. Together they reveal disorder in the medical field and raise the question: where is regulation? In China this problem isn’t limited to healthcare: food safety, securities, real estate, and other traditional industries have had regulatory failures. The burgeoning internet sector is even more riddled with gaps; many legal provisions are blank. In a market without effective oversight, the cost of wrongdoing becomes negligible and businesses will tilt toward profit over ethics. Cases like Didi’s ride-hailing murders, fake goods on Taobao, fake tickets on Ctrip, fraudulent listings on 58.com, and food-delivery scandals reveal that nearly two decades of rapid internet growth also exposed a side willing to pursue profit without scruple.

Can a company’s behavior be restrained merely by its own ethics, social responsibility, self-policing, or the conscience of its executives? Historically and globally, no. A developed market economy relies on a sound legal system where violations are punished — that is best practice. Google has erred too: in August 2011 Google reached a legal settlement with the U.S. Department of Justice over illegal pharmaceutical ads, paying $500 million to avoid prosecution. Google reports that in 2015 it blocked 10,000 sites selling counterfeit goods, closed 18,000 accounts, and blacklisted 30,000 weight-loss product sites. Baidu’s 17-year history has included IP infringement, paid search, and pornography-related crises, but responses were mostly superficial fixes, never addressing root causes. If Google had grown up in China, could it still be the global leader it is today? It might become a second Baidu.

Low legal costs for violations, and penalties that vary with company nature or scale, are the root cause of many corporate sins in China. The core of a market economy is fair competition, and the core of fair competition is the rule of law — so a market economy is fundamentally a rule-of-law economy. Only reasonable regulatory mechanisms can create a healthy business environment.

A final personal view, perhaps narrow, for reference: Wei Hanfeng, former editor-in-chief of Bloomberg Businessweek China, said, “Undoubtedly, a company’s fundamental purpose is to maximize commercial interests.” If businesses pursuing profit is natural and acceptable, then a ruling party pursuing political interests is likewise unsurprising. We are a socialist country with one-party rule; theoretically the party represents the most fundamental interests of the people, which aligns with its political goals. But this is an idealized model, like perfect competition in economics — something approached but never achieved. In Western systems the ruling party’s opponent is an opposition party and market order is regulated purely by law; in China the visible hand of power is stronger. Even with a well-designed regulatory system, purity and autonomy may be lacking. Google’s 2010 exit from China, which left Baidu unchecked, is an instructive example.

Conclusion

Above I analyzed Baidu’s problems from four perspectives. The most immediate consequence of these issues is user experience. Below are two screenshots I casually captured on my phone that show Baidu’s proudest, most lucrative product — mobile search — which ties directly to the company’s survival foundations.

I imagine a scene: a foreign friend hears about a great Chinese high-tech company doing work like Google’s, types the domain into a browser and presses Enter, then sees a homepage that couldn’t possibly be associated with a high-tech firm. They would say: “Are you kidding me?”

Finally, a joke by Lin Mo, an old-school commentator, feels especially apt now; I’ve adapted it:

2016 was a year the entire web cooked Baidu. Stir-fried it, steamed it, braised it, mixed it into a mess — venting frustrations. But tomorrow we still go to work and keep using it.