“You recognize that the world has changed, but you know that to adapt and change your company’s culture you need to defeat a layer of inertia. It’s like setting broken bones, painful and hard.” —Wang Xiaochuan
For Sogou, November 9, 2017, was a very good day – the company had made it onto the New York Stock Exchange. Wang Xiaochuan, who had waited 14 years for that moment, was the one who set Sogou on a new course, the one that had led them to Wall Street and a massive change of fortunes.
In 2003, when Wang Xiaochuan founded Sogou, he hoped search would boost his might against internet forerunner Baidu, established in 2000. But Sogou drifted far from its original goal, so much so that in 2010 and again in 2013, they were nearly bought out by Qihoo 360. But now, not only has Sogou become the most popular Chinese input method on the market, their market share in search had risen dramatically. According to CNZZ data, in April 2013, Sogou search had only 9.15% market share. By the end of March 2018, iResearch puts Sogou’s mobile search market share at 18.2%, making it the second largest in China.
How did Wang Xiaochuan tough it out during this long fallow period? Where will Sogou position itself in the new AI environment? Where will Sogou make its breakthrough? We interviewed Wang Xiaochuan to find out.
Growing against the wind
According to iResearch data, Sogou’s star product, its pinyin input method, is universally used in China, with a market share of 99% in 2018. Its daily active users rank it as the third largest mobile app in China. Sogou’s mobile search market share reached 18.2% by the end of March 2018, second in China.
Search has taken far longer to mature than the pinyin input method, but Wang Xiaochuan never gave up. Even in 2010 and 2013, when it was rumored that Qihoo 360 made two failed acquisition attempts, Wang Xiaochuan came back and, saved by Alibaba and Tencent, he was able to keep his baby. Even after Tencent became Sogou’s largest shareholder, Wang Xiaochuan still maintains ultimate control of Sogou.
Over the course of Sogou’s development, Wang Xiaochuan has dealt with dramatic turnarounds many times, so he’s well-practiced at facing up to trouble. He puts this down to two things:
“First, I’m used to going against the wind. The harder the conditions, the more I know what to do. If you focus, you’ll think up a hundred ways to solve a problem. Downwind, and you’re lost,” he told us, confessing that he works best when things are tough, and when he’s being attacked he feels most compelled to improve. He’s also convinced that Sogou’s work has value, and this keeps him going. “If you’re making something that no one wants, no matter how hard to try, it won’t work. But if you’re doing something that makes sense, you’ll be rewarded every time.”
Of course, as CEO, Wang Xiaochuan cannot just play “against the wind.” He also has to forge a path of development that suits Sogou. And that means AI.
A clear AI strategy
Despite being far smaller than rival Baidu, Sogou has done some impressive work in artificial intelligence. The company has gained kudos for its dictation and simultaneous interpretation tools, Sogou travel translation app, and Sogou translation pen, but most of all for its clear and focused AI strategy. Wang Xiaochuan says they aren’t just going where it’s hot, but actually stepping back to see where their strengths lie and deploying AI technology in those areas.
To understand Sogou’s strategy here, we need to first understand Sogou’s “dual turbo” model of pinyin input method and search. During Sogou’s “triple turbo” period, the search engine was Sogou’s main route to commercial success, but in the “dual turbo” period, Wang Xiaochuan believes input can be independently commercialized. This has given rise to Sogou’s AI strategy. Sogou first plans to make AI useful in search and second make the input method more natural.
Sogou has long developed image-based search based on image recognition. On the input side, Sogou derives AI functions such as speech recognition, shorthand, translation, and lip recognition based on Sogou input.
Previously, Sogou introduced a search product that could identify dog breeds, a sort of canine facial recognition software. Since it could recognize dogs, surely it could identify people too? Sogou had the chance to get into facial recognition technology, security appliances, or even autopilot functions. But Wang Xiaochuan made it clear Sogou would not step into those areas right now because, in his words, “Sogou does not currently have the security clearance or the environment to test automatic driving properly.”
The advantages of Sogou AI
Speech recognition, semantic understanding, and related technologies are becoming the crown jewels of Sogou’s AI strategy. But no matter which field, Sogou is by no means the only player in an industry that also includes iFlytek and Baidu.
“It’s really not about data, our voice technology works in the context of pinyin input, and this is an advantage.” In the face of competition, Wang Xiaochuan believes that relying solely on technology is not enough to beat the competition. The real key is having somewhere to use it: market share. Although companies such as Baidu and iFlytek are also developing input method products, Wang Xiaochuan believes their market shares are too small for them to catch up. Sogou’s dominance in pinyin input gives Wang great confidence. “We have a peak daily speech recognition call volume of 360 million, more than all the others put together,” he said.
People say Wang Xiaochuan has such belief in where he’s headed because of his experience in Sogou search. “When we started in search, we thought that if the technology was better than Baidu’s, we would be able to grab market share. What we discovered was that it wasn’t actually a matter of technology. When a business forms a closed-loop, it is not something that can be disrupted by technology alone, unless your technology is truly disruptive.”
The arrival of AI
Search and input was where Sogou’s AI capabilities were first put to the test. In seeking more fields, Sogou turned to smart hardware. Sogou has launched the Teemo smartwatch for kids, Teemo household robots, and Sogou translation pen among other items of smart hardware. Regarding the distinction between the two brands, Wang Xiaochuan said Teemo will focus on kids, while Sogou will focus on smart technology.
It is widely known that Baidu is also stepping up its efforts in AI. However, unlike Baidu’s investment strategy, and its “buy buy buy” approach to gaining hardware R&D expertise, Sogou’s smart products are developed by their in-house team of engineers.
“I recognize that acquisition may be the best route. It can cost less than internal development, and definitely saves time.” Wang Xiaochuan says he is not against gaining smart hardware through acquisitions, but in terms of the volume Sogou is after, it’s hard to find good targets. “Of course, we also have the advantage of doing our own thing. We can certainly integrate things better.”
While Baidu is working on platform products such as DuerOS and Apollo, Wang Xiaochuan tells TechNode that Sogou is not interested in developing platforms at the moment.
“Opening up solves diversification and cost issues, but only a closed system can help with innovation.” Wang Xiaochuan believes that in the start-up or early phase of a business, a closed loop makes the best products. “It’s best to open up slowly until you make it.” To illustrate this point, Wang Xiaochuan gave the example of IBM: “When IBM made computers, everything including the screens was made by them. Amazon did the same thing in sound. Only after they had finished making the product did Amazon open up.”
Wang Xiaochuan did tell TechNode, however, that by the end of the year, Sogou would be bringing out a new hardware product. He didn’t say what it would be, but he did say it wasn’t a smart speaker. “Smart speakers are quite a good direction to take, but too many people are doing it now, and the price is being shredded everywhere you look. And it takes us too far away from our purpose. The core logic of speakers is to broadcast music, which is not our forté.”
Wang Xiaochuan did note another major feature of the Sogou brand: “We like to stay close to our users and improve their convenience in communicating. Voice interaction is the very heart of what we do.”
If companies don’t move forward, they risk losing everything.
We’ve mentioned Sogou’s advantages. So to be prepared for the future, where might Sogou’s crises come from?
Wang Xiaochuan thinks that everything is changing way too fast. To keep pace with change, a company has to upgrade constantly, otherwise, it can only retreat. “You recognize that the world has changed, but you know that to adapt and change your company’s culture you need to defeat a layer of inertia. It’s like setting broken bones, painful and hard.”
In order to confront such crisis points in the future, Wang Xiaochuan says it is essential for the company to keep dissembling parts of itself, adjusting organizational structures and keeping up constant reform.
—Translated by Heather Mowbray