Since 2016, artificial intelligence has undeniably become the hottest field, with Google’s AlphaGo before, Baidu’s Dumi and Xiaodu robots after, and Xiaoice launched by Microsoft’s Bing search in 2015 constantly making news. A dizzying array of ai products has created a sense of artificial intelligence everywhere. But we find that the tech companies that made the bold bet on AI in the first wave are almost invariably search engine companies (Google, Baidu) or companies with strong search engine technology (Microsoft).
It’s interesting. Why are these search engine companies the pioneers in the “ai” battlefield? Is it just because these search engine companies are technologically superior? Or are these the only companies that see AI as “hot air”?
Giiso Information, founded in 2013, is a leading technology provider in the field of “artificial intelligence + information” in China, with top technologies in big data mining, intelligent semantics, knowledge mapping and other fields. At the same time, its research and development products include information robot, editing robot, writing robot and other artificial intelligence products! With its strong technical strength, the company has received angel round investment at the beginning of its establishment, and received pre-A round investment of $5 million from GSR Venture Capital in August 2015.
Today, we’re going to talk a little bit about that.
As a product manager, I will try to make a comprehensive analysis from the aspects of business model, technical advantages, development path and so on, hoping to judge the secret behind it by logical reasoning.
One, search engine company is advertising company in commercial essence
What a company is, not just by what products it makes, but by its business model, which track it’s on and how it makes money.
An excellent search engine, it connects ordinary users and web pages, users can easily and free to search to see the site and content, and the content of these sites is searched by the search engine, through the search engine technology presented to the user. This kind of product is a typical tool-based product, which connects the supply and demand of information and forms a large amount of search user traffic and user behavior data through operation.
Search engines in the accumulation of these traffic in the process, gradually began to consider these traffic cash. At this point, the search engine company will try to pull in a third side advertiser. Advertising is a standard business model that generates communication value through traffic and asks advertisers to pay for that value. For example, Baidu has built its own internal flow data platform and external flow data management platform, and built a set of advertising system platform, which has created huge commercial value for Baidu under the advertising model.
At this point, we look at search engines, which are actually commercial platforms that connect ordinary users, website pages, and advertisers. They increase user traffic by subsidizing ordinary users (free use of products), and then attract advertisers to pay the bill. Therefore, search engine companies must constantly through a variety of products, operations, channels to broaden their own traffic, and accumulate rich user behavior data, the value of these data reflected in the flow, so as to form high quality advertising value, to ensure their own commercial interests. Moreover, because of the accumulation of search technology and data technology, these search engine companies will get involved in the collation of external product traffic and build richer cross-domain data for users. This business model, directly help search engine companies in the past ten years to make money, daily income bucket gold but so. Microsoft’s Bing search has achieved positive cash flow earnings when its market share was only 20%, which shows the superior profitability of search advertising.
This is the business nature of search engine companies as we know them.
Why do search engine companies seek new growth points?
If search engine companies are already making so much money from search advertising, why seek new growth?
It is a complicated question. We often say that any company that wants to be sustainable must constantly innovate, or it will be abandoned by the market. Such abandonment actually reflects that some key items in the original business model have changed, which may be due to the change in the way of acquiring customers, resulting in an increase in cost, or the decrease in technical cost and shortened transaction path, resulting in lower market barriers and enhanced competition. Either way, the market is changing rapidly, and we want to understand what problems search engine companies encounter to seek new growth, to see how their business models change.
Search engine companies meet the challenge: online traffic dividends disappear, business model shortcomings exposed.
With the development of the Internet today, everyone knows that the online traffic dividend has almost disappeared. The growth of many traffic platforms is negative year-on-year, and even some traffic platforms have shown negative growth. Today alibaba and Tencent are fighting for offline retail transactions, which account for only about 15% of China’s retail market and are growing sluggishly, while 85% of all transactions take place offline.
The disappearance of online traffic dividend, the challenge to search engine companies is the future revenue growth space change school
For example, the two internal traffic products that bring the most advertising value to Baidu are Baidu Search and Mobile Baidu Assistant, which make a lot of money for Baidu through news stream advertising and App distribution, respectively. However, the online search traffic has begun to shrink, the way users get information has been infiltrated from simple search to various channels, users can get information from social platforms, we media channels and other product channels (Baidu’s self-made Feeds stream is also to counter this change), the advantages of search engines are shrinking.
In addition. As can be seen from the financial results of NetEase Q4, good mobile game products no longer rely on distribution channels, and the value of App distribution channels has been shrinking, which reflects that the App market is becoming saturated and the growth model is changing.
Baidu’s current price-to-earnings ratio and previous quarterly results suggest that revenue growth is already weakening.
At this point, you might ask, can’t Baidu monetize traffic from other products?
At present, baidu’s biggest traffic in addition to Baidu search, there are similar to know, encyclopedia, library, experience and other knowledge vertical search, as well as like tieba, map, takeout and other large traffic products, originally these products should bring huge traffic dividend for Baidu, but in fact, it is not.
First look at baidu these knowledge vertical search products, its own flow from Baidu big search points out, and they themselves are part of the search, did not become an independent entrance to contribute to baidu big search traffic, also can not be separated from Baidu big search independent cash.
And such as tieba, map, takeout and other products far away from the search model, its own traffic realization is different from the current business model of Baidu, the realization of these products themselves is quite difficult, and because far from the search model, the ability to guide the flow of Baidu is also limited.
Therefore, baidu encountered the problem is almost all search engine companies will encounter the problem, search engine business model for traffic is highly dependent, sooner or later there will be traffic bottleneck.
Three, what is the accumulation of search engine companies?
After looking at the growth bottleneck of search engine companies today, we turn back to see what the accumulation of search engine companies, whether these accumulation can help search engine companies to find new growth points.
Some people say that search engine companies accumulate technology, because judging from the name, “search engine” sounds very geeky, the accumulation of technology must be the most abundant. Well, that’s not the whole story.
1, the accumulation of search engines
Search engine companies accumulate data, technology and channels on top of their business models.
The first is rich user data
It is no exaggeration to say that the user data in the hands of advertising platforms is very rich. In the commercial product system of search engine, the massive user behavior data in its data management platform includes the user behavior data of all its products as well as the user data of external cross-domain product platform. This wealth of data ensures that search ads have enough say in pricing, and the richer the data, the more expensive they can be sold for.
The second is the accumulation of technology
Be sure is, technical accumulation is search engine company the most real accumulation. These technologies include big data analysis technology, advertising technology, deep learning technology, neural network technology and other technologies that are highly correlated with data. As can be seen from the products and revenues of previous search engine companies, such huge profitability must be built on the accumulation of extremely strong data technology. For example, because Baidu’s revenue is mainly from advertising, it is said that one of the most complex engineering projects inside Baidu was “Phoenix Nest”, which covers all aspects of big data technology. This level of data technology accumulation is not overnight.
Finally, the accumulation of channels
In the advertising sales system, it includes the supply side (advertising space), the demand side (advertisers), all levels of channel agents, and the advertising platform system under various systems. The organic combination of these resources is the premise to ensure the benefit sharing and good operation of the whole advertising sales system. The good operation of search advertising requires the accumulation of rich channel resources and interest distribution strategies in the supply side, the demand side, the intermediary agent and other links. This accumulation forms a huge interest supply chain network, and these search engine companies firmly control this supply chain network. These channel resources are able to help search engine companies in the new growth point to continue to achieve business model extension.
Business model is iterative innovation, not a new path
Some people say: innovative business model is equivalent to suicide.
When a company has grown up with an established business model, it is impossible to reverse its advantages. If you look at Microsoft, for 40 years, its business model has been ToB to sell software or services. It’s been iterating on ways to make money, but it’s still the same model.
Search engine companies are no exception, what they need is the iterative innovation of business model, need to use data for advertising profit on the road, continue to deepen.
If you want to make advertising profit model in-depth, there may be at least two ways, one is to expand the scope of traffic, previously only online these flows, now can expand the flow to the offline; The other is to increase the value of data. For example, in the old data system, a user’s single click is worth 10 yuan. Is it possible to increase this value to 20 yuan in the new data system?
No matter which way you choose, you can continue to expand on the original business model. As a search engine company, is the choice to expand the scope of traffic, or enhance the value of data?
Before we go any further, let’s talk about what ai products are all about.
What are the so-called artificial intelligence products?
The issue of artificial intelligence is too big for my knowledge system to cover complicated technical issues, so we only talk about artificial intelligence products here.
In the engineering world, think most (or almost all) technology is established based on the search technology, and search the accumulation of huge amounts of data, and to build a set of data based on huge amounts of data statistics and analysis, which can bring some application scenarios key decision guidance and support, the product model is a generic term, It’s called big data computing. Almost all scenarioalized products based on big data computing can be called artificial intelligence products.
For example, when a company has accumulated more than 10 years of industry data on transactions, records, finance, warehousing, logistics and so on, ordinary algorithms can no longer handle this complex data system. At this point, big data can effectively find some special laws through statistical analysis of data. For example, when the delivery time of goods changes, the number of transactions may double. Artificial intelligence is the product model formed by finding rules through big data and then assisting decision-making. Because it’s impossible to find this pattern, it’s impossible to make this kind of decision.
The scenarios described above are only very shallow artificial intelligence products. In the past few years, this kind of Intelligence system applied at the enterprise level is often referred to as BI (Business Intelligence), and has actually been used in some industries. In the past, BI did not have strong big data computing capabilities, because ten years ago cloud computing was not as widespread as it is today, and large scale data computing was dependent on minicomputers or distributed computing of that era.
The development of cloud computing makes it easier to collect, process and analyze data, and big data can be found in all kinds of data systems in all walks of life. Therefore, artificial intelligence has become a hot field.
So, take a second look at what artificial intelligence is.
We found that, in fact, it is based on cloud computing of big data to model and analyze the data, so as to obtain decisions and results in specific scenarios. Such products can be called artificial intelligence products.
No matter from the interaction mode of ARTIFICIAL intelligence, or from the perspective of the data accumulation that it depends on behind it, it is natural that artificial intelligence grows through search engine technology. Cortana of Microsoft, Xiaoice and Dumi of Baidu all use the technology behind artificial intelligence to reach the output of their products.
Betting on ARTIFICIAL intelligence is not a choice, it’s a must
Back to the previous question, in advertising revenue model, search engine companies will choose to expand the scope of traffic, or increase the value of data?
Let’s take Baidu as an example.
If it is the former, Baidu needs to carry out extensive layout in the offline flow, but whether it is takeout or glutinous rice, in the offline flow layout are not successful, the former dry but ele. me, Meituan, the latter dry but the new United States. Moreover, the simple increase of traffic, whether from the current accumulation of Baidu or sustainable development in the future, is not baidu’s advantage. And to realize the offline traffic, Baidu must have enough access to payment, which also needs baidu finance to continue to work hard.
So, Baidu’s choice, only to enhance the value of data. Ai has the opportunity to help Baidu improve the quality of its data accumulation, increase the value of AD revenue, and even penetrate into post-ad transactions, greatly increasing its profitability.
Looking at other search engine companies from Baidu, we find that everyone is in a similar situation.
Artificial intelligence is based on big data, so it means that no matter the input or output, artificial intelligence must be based on big data. As many search engine companies adjust, we can see that they are already looking for more external scene access, or scene extension, to deepen the depth of data.
As an example, Baidu recently launched the DuerOS operating system (you can search for it). You can see that the operating system can be accessed in various scenarios to complete some functions. For example, it can play music for you, order food for you, and control your smart home. This scene more access is actually a kind of deep precipitation data level, a user search in baidu, and baidu knows, stick these products used, baidu can analyze out roughly how old he is what, where, what kind of work, education, and so on, these data were to advertisers who used to do online advertising. After DuerOS access, baidu can get every day in the home is what appliances, this time at home, more about what the quality of life, consumption of which is more impulsive, and so on, the baidu can not only sell the user more expensive, even after can be directly involved in advertising spending recommendation (recommended itself is search technology).
When we understand it this way, we can open up our imagination and imagine.
For the past decade or so, user traffic has come through web browsing, and advertising has come through the web. But in the future, the embodiment of user traffic will change, that is, through intelligent hardware + artificial intelligence. Any form of product that can reach c-end users can be called flow, a helper robot, a smart refrigerator at home, a smart wristband, a smart home robot, and so on. Because this product becomes ubiquitous, all aspects of a user’s information is digitized, then the form of advertising at that time may change drastically, and any form of human-computer interaction will become the carrier of Baidu’s advertising. What used to be a web page for advertising may become a series of human-computer interaction intelligent hardware and ai products to carry advertising, and the business model of intelligent hardware will change.
At this time, let’s make a comparison. Under the current system, advertising is sold through traffic screening, and then launched in the form of display, click and download. For example, if 10,000 displays are sold for 100,000 yuan, it is expected to convert 1000-10,000 real users. However, this fluctuation range is too large, and it becomes very difficult for businesses to evaluate the cost and re-launch. If new data in the artificial intelligence system, because the user data is more accurate and thorough, all users of mining has become an unprecedented precision, now want to find the amount of data that can be 10000 users need accurate assessment, such as touch up to 50000 times can come the 10000 real users, advertising can even make a promise for results at this time, Then an advertisement may be sold at a higher price. For example, it costs 200,000 yuan to acquire 10,000 users. Compared with the previous strategy of 100,000 yuan to acquire 1,000-10,000 users, this kind of advertising strategy is more accurate, and it is easier for businesses to evaluate the input-output ratio of continuous advertising.
This time, we come back, we find that the search engine companies with scenes of artificial intelligence, is to exploit its depth and the value of advertising data, behind this new advertising business model may bring the earth-shaking changes in revenue, this kind of imaginary space to increase, direct promotion is the confidence of the investment market, growth would be its market value.
If search engine companies want to achieve this growth, they need to have very calm product research and development, technical system upgrade, operation management and other strategies, which are not overnight, but need a long process, huge difficulty.
Giiso information, founded in 2013, is the first domestic high-tech enterprise focusing on the research and development of intelligent information processing technology and the development and operation of core software for writing robots. At the beginning of its establishment, the company received angel round investment, and in August 2015, GSR Venture Capital received $5 million pre-A round of investment.
summary
At this point, we have completed a simple discussion on the topic of “why is the forerunner of artificial intelligence battlefield a search engine company”. From the point of view of business model analysis, artificial intelligence is the inevitable way of search engine companies, rather than bet so simple. The way we analyze it is not through superficial product relationships, or some trivial gossip, but through the perspective of business logic.
The product manager, without understanding the business model, is a beginner product feature designer.