One day, in order to make a bottle of soy sauce, you walk into the supermarket. Then ...

When checking out, there may be soy sauce, diapers or other commodities in the cart.

please don't confuse this scene with the most commonly used case of "beer+diapers" in big data, because this scene has happened to all of us.

Wen/Zhang Shule

TMT industry observer, game industry commentator, People's Daily and People's Post and Telecommunications columnist

put the supermarket as a content platform. When we skip all kinds of information and go to the required content (soy sauce), we may pass through the content area of diapers, and then suddenly think that children's diapers are almost out of stock ...

There is no necessary connection between them.

For example, put the explosions in the back, so that guests can pass through more shopping areas; For example, the downstairs elevator is on the other side; For example, buying gifts or buying activities to promote mashups ...

This is a consideration under the traditional business model. On the internet content distribution, we can regard all the content as "advertisements", and you must never let users come across things that they might buy.

Moreover, in the content distribution state, users constantly choose the content requirements, and the background algorithm will be more accurate, and it is even more unrealistic to detour. The purpose of everyone to look at the content itself is to get what they need as soon as possible.

At this time, we will find an interesting situation, that is, on the information explosion of the Internet, the user's information acquisition mode has changed into two States:

One is the most traditional search mode, and users go to Baidu with a purpose to find some kind of information, that is, soy sauce. But there is a problem, many people don't always have a clear idea of soy sauce.

Second, I blindly meet the information flow, that is, supermarket shelves, and as a result, my vague demand becomes a purchase. This has become a mainstream at present, that is, the appeal point of Baidu information flow advertising.

Xiang Hailong, the president of Baidu search company, has made a brilliant exposition on this, that is, "users not only need to find information more conveniently and intelligently, but also need personalized and targeted information to actively recommend".

from "people looking for information" to "information looking for people", it's as simple as that. Doing it is not easy at all.

Advertisements can also read your mind

With the rise of Baidu Baijiahao, Today Headlines and other content distribution platforms, information flow advertisements also have more bearing platforms, which are no longer just the product advertisements related to your search items or previous browsing items that appeared in the sidebar in the past similar to Baidu search or Taobao shopping.

Baidu, Weibo, WeChat and Today Headline have all launched information flow advertisements, but how strong is this type of advertisement?

data speaks for itself. In 214, information flow advertising in the United States reached $5.4 billion, and more than 5% and 7% of the advertising revenue of facebook and Twitter respectively came from this. In China, it is only 6 billion yuan, and in 217, it is expected to increase by 48%, reaching 34 billion yuan.

However, the expected growth does not mean that the target can be achieved simply by implanting relevant advertisements in content distribution.

Information flow advertising should not simply provide related goods according to your content preferences, such as watching cars and delivering car products.

actually, it's not bad. More often, after buying a car, we often look at the introduction of the car, and the subsequent information flow advertisement, although accurate, may often be a car.

What's the feeling? It's similar to Baidu search promotion, and Baidu search promotion has gone further now.

For example:

Tubatu, a home improvement O2O company, is a frequent visitor in baidu promotion. In the eyes of many people, the good performance is partly due to the good search engine optimization and advertising marketing. Tuba Rabbit made the most advertising profit by advertising in Baidu with the least advertising fee.

and promotion payment is an introduction. In the process of Baidu's search, a large number of content presentations in Baidu's ecosystem, such as Q&A, news and user experience, and even word-of-mouth in forums, will form a strong superposition and eventually become huge traffic, which will help advertisers go further in marketing promotion.

This case is not about how to do search, but to grasp the user's psychology:

Using Baidu to find things has a clear purpose, and searching for watermelons for watermelon shops is so straightforward, which is quite good.

while getting content in the information flow may be just a hobby in many cases, imagine a scene:

You are reading the information about home decoration, and you come across an information flow advertisement packaged as "This kind of home decoration will make your girlfriend 1% excited", which is actually a decoration company. You may be clicked, and then you will turn it off mercilessly when you find that it is an advertisement.

Because you're not looking for decoration, you're just looking for beautiful pictures. At most, you plan to remove furniture and ornaments when you get home.

The real information flow advertisement is to read the user's intention on the basis of the accuracy of search and information flow recommendation. The scenario may be like this:

When you see the cooking methods of soy sauce and other condiments, insert an information flow advertisement for diapers. Despite this feeling, it may be very sour.

how? In fact, the big move is to find more needs that users may not be very clear about through their choices. Perhaps we can call it "to-do reminder".

2 million tags =1 person? Just-needed portrait

to solve potential pain points must have a "tool" that can understand it better than the user himself, which is usually called user portrait.

"By searching data, geographical location, knowledge map, transaction data and other behaviors in hundreds of dimensions, today we can label a single user up to 2 million. Through accurate user portrait modeling, we can accurately identify everyone and understand everyone among 6 million users. "At the 216 Baidu Moments Business Summit, we announced to Hailong the amazing user tag data owned by Baidu.

2 million tags =1 person. What is the status?

ten years ago, we tagged our friends on QQ, and there were often a dozen reliable ones.

five years ago, in order to better absorb powder, we labeled ourselves on Weibo, and many people filled in three or five.

This is roughly what we know about ourselves and what our friends know about us, and a label of 2 million is equivalent to drawing ourselves from all sides that a magnifying glass may not see clearly.

how to draw it? Maybe I just searched for "the risk of caesarean section" many years ago, learned about "how to eliminate jaundice in newborns" in the Q&A, and had a heated debate on the topic of early education in the mother-infant post bar ... < P > In the PC era, such portraits are often one-off, and can only be turned into cookie. Every time targeted marketing is carried out, it can only be an advertisement to see cars and send cars.

On the mobile side, each App will require login account, and as a result, the scattered tags will fall on one account. Theoretically, the larger the product matrix and the more forms, the fuller the side of the label. Then through the information flow advertisement, put it on you who are lazy to choose information from your favorite content direction.

because it is useful and practical, in fact, it is no longer an advertisement, but a solution to the pain point.

In other words, if there is no powerful product matrix, and all kinds of labels left inadvertently are obtained from all directions and different angles, and managed effectively through artificial intelligence, then the portrait of this user is either long legs or full of labels. As a result of portrait distortion, information flow advertisements are inaccurate. Why is Baidu information flow advertising accurate? It is precisely because of the multi-product matrix, big data accumulated over the years and strong technical strength.

Rejecting false advertisements and achieving a breakthrough in the conversion rate of invalid reading

Baking your heart in Henan is a representative of entrepreneurs who have benefited from information flow advertising.

When this company started its business in 214, it was just a custom-made private cake in a 5-square-meter store. However, through effective Baidu information flow advertising distribution, it quickly exposed itself to the eyes of baking enthusiasts all over the country, and quickly developed into a professional service company with a team of thousands of square meters and 2 people.

The value of this case is as follows:

For business owners, information flow advertisements (diapers) that cut into users' inner needs can better tap users if they are used in conjunction with search promotion (soy sauce) that directly reflects their needs.

the significance of Baidu search+recommendation is actually reflected in the improvement of conversion rate besides accuracy. Compared with the platform with only content distribution, Baidu search, Baidu map, Baidu glutinous rice and other platforms can achieve multi-faceted user tag retention, and ultimately create more accurate delivery possibilities for Baidu information flow advertisements.

At the same time, the power of search is still strong, especially to meet the urgent needs. The advantage of combining search and recommendation is an important reason why Baidu advertising is more accurate and effective than other platforms.

at the same time, with the help of artificial intelligence technology, information flow advertising is also iterating.

Take Nuoxin Cake as an example. In the early days, the form of information flow advertisement was "Nuoxin Cake-Valentine's Day speaks for love, and customized exclusive love words", which was quite common, with a click conversion rate of only 1.86%. When the form becomes "Valentine's Day confession artifact, giving a girlfriend 1, romantic crits", in the fusion scene, according to common scenes such as crowd orientation and content orientation, as well as Baidu's unique intention positioning method, it naturally blends into the crowd with potential needs, and finally promotes the conversion rate to 3.24%, nearly doubling.

the scene of "I went to the supermarket to make oil sauce, but I came out with a pile of diapers on my back" may not be completely realized in information flow advertising yet, but at least a wave of opportunities has emerged, and Baidu is leading this wave.