There is no doubt that Douyin and Kuaishou Technology are duopolies in the short video industry-the former claims that its DAU has exceeded 300m, while the latter plans to reach 300m DAU by the Spring Festival in 2020. They each have unshakable basic cards in the field of short videos, successfully entered live streaming in 2018-19, and are trying to bring goods to e-commerce and even their own e-commerce platforms.

The two platforms are highly similar in strategy, but there are still big differences between user profiles and tonality-from Slogan, Douyin is "discovering a better life" and Kuaishou Technology is "seeing every life".

Kuaishou Technology, VS of Douyin, can be called "the war between two worlds". The question is: why are their user portraits and tonality so different? The answer is: because they deal with information flow differently, because their algorithm drivers are different, and because they have different attitudes towards creators.

The operation ideas of Douyin and Kuaishou Technology are not absolutely "which is better or worse"; we should be clearly aware of their differences in order to better understand the short video track, and even the entire content track, and even the entire Internet.

Frankly, we are not interested in the question of "who will win the short video war". We are only interested in the facts, and when we realize the facts, we will naturally be able to work through the whole logic and judge the direction of the situation. Moreover, the "war between two worlds" is likely to be unwinnable in the end-not all wars can be divided.

The double talents of Information flow: social attention and algorithm recommendation

Information flow can be divided into "social attention flow" and "algorithm recommendation flow", which can be called "rivers and lakes", one of which can be unique in the world. These two kinds of information flows complement each other and permeate each other, but they will not replace each other.

Social attention flow is mainly a FEED flow, which is sorted by time dimension and centered on mutual or unilateral attention of social relations. The content is mainly user-generated pictures, videos and articles, and the information is relatively divergent, which restricts the quality of information through social relationships. Wechat moments is a typical social flow, and Weibo Corp was also a social stream in his early days.

Algorithm recommendation flow, mainly based on the background algorithm as the core, comprehensive information and human relevance, importance, timeliness through the machine to give a comprehensive weight, centralized distribution. Jinri Toutiao and Douyin are both typical algorithm recommendation streams, and Weibo Corp has gradually turned to algorithm recommendation streams this year.

Recommendation flow is different from social communication. The platform has the absolute right to distribute information flow and 100% of the rights and interests in the distribution of information flow advertisements. At the same time, users have less tolerance for social exchange advertisements. This is one of the reasons why the advertising cash ability of Toutiao / Douyin is better than Wechat moments / Kuaishou Technology.

Today, pure recommendation flow and social communication are rare-for example, Bilibili Inc. 's home page has both the first screen recommended by algorithm and the third screen based on attention. Zhihu Inc. 's home page is divided into three columns: follow, recommendation, and hot list. the first column is social communication, and the last two columns are recommendation stream. However, any content platform will have a choice between "social attention" and "algorithm recommendation" and favor one side.

The three columns on Kuaishou Technology's first page have hardly changed since the transformation of the community. The first column is attention; the second column is discovery (formerly called popular); and the third column is the same city. This seems to be nothing unusual, Douyin also has attention, recently also added the "same city" column.

However, the distribution proportion of "follow", "Discovery" and "in the same City" in Kuaishou Technology is about 2:2:1, which combines "social and algorithm". Many people do not understand, as a technology, what are the differences and barriers between algorithms? The algorithm also has values, and Kuaishou Technology's values are "fair and inclusive", that is, to ensure that every user's content is displayed and every ordinary person's life is respected.

In contrast, Douyin gives creators a chance to become a "hit", but most of the opportunities are eventually taken away by a small number of people.

For the average creator, the "expected return" of becoming famous on Douyin and Kuaishou Technology may be the same-the former is a small probability of success multiplied by a large gain of success, while the latter is an appropriate probability of success multiplied by an appropriate return of success. Which one do you choose? That's the creator's own judgment.

Social distribution: "Waterfall flow + Fairness algorithm" helps to socialize, and the video circle of friends with semi-acquaintances is emerging.

Douyin makes the recommendation first and then pays attention, but so far still takes the recommendation as the mainstream, the attention chain is not very important. Kuaishou Technology directly regards recommendation as the initiator of attention, seeking synergy between "recommendation" and "concern".

From the perspective of traffic mechanism, Douyin pays more attention to "recommendation", while Kuaishou Technology tends to "pay more attention". The variance of the number of likes / plays of the same user on Kuaishou Technology is significantly smaller than that of Douyin.

Take Wang Zulan as an example. Most of the likes of Wang Zulan's short videos on Kuaishou Technology range from 200000 to 500000, while those on Douyin range from tens of thousands to tens of millions. Kuaishou Technology's "follow" relationship ensures stable traffic (mainly for medium-and long-tail creators, while head creators' short videos are restricted, mainly to cash private domain traffic through live streaming), while Douyin's "recommendation" is "popular style is king". Therefore, super belt anchors such as Li Jiaqi and Weiya will choose Douyin as the main battlefield besides Taobao, while there are tens of thousands of middle belt anchors on Kuaishou Technology.

From the perspective of user experience, Douyin is undoubtedly more "pleasing to the eye". Open Douyin to enter the playback mode directly, and rely on swiping up and down to change the video. This lazy person interaction enhances the user's stickiness and weakens the user's willingness to change. The probability of using "nearby plate" and "following plate" is greatly reduced. Users' attention is glued to the high-quality content of the head user (Douyin algorithm determines that recommended content is better, and no more user actions are needed), and centralization is intensified.

Conversely, Kuaishou Technology's three interaction modes are all "waterfall flow" (recommended can be changed to large screen mode, which needs to be set manually), and due to the algorithm mechanism, Kuaishou Technology's performance of high-quality content is not as dense as Douyin, which requires users to choose. This is why "follow" and "in the same city" are used so frequently. In this way, "scrolling" is more suitable for Douyin, "Waterfall flow" is more suitable for Kuaishou Technology, and "Waterfall flow" + "fair algorithm recommendation" work together to shape its "social attribute" space.

Su Hua, founder of Kuaishou Technology, concluded that since 2018, Kuaishou Technology has transformed into a community where semi-acquaintances and semi-strangers coexist, and a large number of social relationships have been precipitated in the platform. Alma mater, hometown, peers should be regarded as China's most extensive circle of semi-acquaintances, alumni, fellow-townsmen, peers with a natural sense of intimacy. What is the existence of Kuaishou Technology's semi-acquaintance social relationship? Based on region / occupation /. Video moments.

We choose a career-truck driver (on the road all the year round, unite and help each other), an area-Baoji (northwest region, Kuaishou Technology has a high penetration rate), and find two ordinary users with about 2000 followers. A simple analysis of the user attributes in their comments.

There will be many groups based on occupation and region on Kuaishou Technology. For ordinary users, most of the interactions come from peers or fellow villagers, and social relationships between semi-acquaintances and semi-strangers are emerging.

Truck driver @ third Brother: truck drivers communicate with the main group and have been in Kuaishou Technology for nearly a year. He has released a total of 81 short videos and nearly 2500 fans. The number of views of recent videos is about 1000, likes and comments are all around 20%, and most of the users are truck / truck drivers.

Dameiting Sir: Baoji communication group leader, has been in Kuaishou Technology for nearly a year, and has released 13 short videos with nearly 1500 fans. The number of views of recent videos is about 1000, with an average of 50 likes and 20 comments, with all comment users from Shaanxi and more than 50 per cent from Baoji.

Algorithm: Douyin popular style logic, user experience first; Kuaishou Technology's fair principle, based on creators

Recommendation algorithm, as the core competitiveness of Kuaishou Technology and Douyin, has been sealed in the black box, and there are few people who really understand the rules of its algorithm.

Outside arguments about the Douyin and Kuaishou Technology algorithms are mostly speculations, except for sporadic introductions by the management of the company, but these speculations may be very close to the truth. (the detailed rules of the two algorithms are adjusted all the time.) We will find some interesting phenomena when using Douyin and Kuaishou Technology at ordinary times, and then we will explain them from the perspective of the algorithm.

Phenomenon 1: when you open Douyin, the content recommended on the system is mostly liked at millions of levels, while the videos "discovered" on hand are basically maintained at tens of thousands to hundreds of thousands of levels.

Phenomenon 2: on Douyin, you will find that many videos were posted a few months ago, while most of Kuaishou Technology's videos were recently posted. (the measured difference has become smaller recently, and even Kuaishou Technology will appear in the video from a few months ago.)

[Douyin algorithm mechanism]

Multiple auditing: multiple auditing is the first threshold for Douyin algorithm to filter videos. The intelligent machine audit will identify the uploaded video images and keywords according to the artificial intelligence model set in advance. One of them: check whether the work and copy violate the rules. If suspected, it will be intercepted by the machine and prompted for manual attention. Second: by extracting images and key frames from the video, matching with a large number of works that already exist in the Douyin database to reduce weight and duplicate content for low-flow recommendation, or reduce the right to recommend (only fans, visible to themselves). This is the fundamental reason to guide users to "seek new, different and first".

Next, we will enter the manual review process, which mainly focuses on the video title, cover screenshots and video keyframes.

Overlay recommendation: works that have passed multiple reviews will be assigned to an initial traffic pool (about 200,300 online users), and the new Douyin account and Big V will be treated equally. Novice Douyin users may wonder: why does the system recommend more fans to others with the same amount of fans? Why do accounts with a large number of fans get more recommendations?

This is mainly due to "account initial weight", "initial flow pool user feedback" (like rate, comment rate, forwarding rate, completion rate and follow ratio, etc.), "account already has fan feedback", "external real account activation".

1) weight: the account system of continuous creation gives higher weight.

2) accuracy: continuously update vertical works, and when the system recommends pool matching, it will give more accurate users with the same attribute tags.

3) the number of fans: the stock of active users can pull the referral likes of published works to a relatively high level.

On the whole, the content creators on Douyin are the strong ones, but the role of popular styles can not be ignored and has strong media attributes.

Late detonation: there will be a very interesting phenomenon on Douyin, some of which are mediocre on the same day or even the whole month, and suddenly become popular one day. There are two main reasons: first, Douyin will re-mine the "high-quality old content" in the database to give more exposure, and the vertical content will be matched to more accurate users through the system. Second, if a creator's work gets a lot of exposure, a large number of users go to their personal home page to look at the previous work, so that some works get enough attention (referral likes), the system will re-put it into the recommendation pool, resulting in multi-point blossom.

Traffic peaking: generally speaking, the duration of high recommendation will not exceed a week. On the one hand, after the popular style content is pushed to the people with all the relevant tags, the feedback from other inaccurate tags is poor, so stop the recommendation; on the other hand, Douyin will test the sustainable output ability of content creators and do not want it to become popular quickly.

[Kuaishou Technology algorithm mechanism]

Heat weight: Kuaishou Technology's unique algorithm mechanism, as the heat increases at the beginning of the video, the exposure opportunity will also increase. The "heat weight" plays the role of "choosing the best and eliminating the bad", but when the heat of the video reaches a certain threshold, the chance of exposure will continue to decrease. "Heat weight" will play the role of "choosing the new to remove the old". In fact, it is to give users an equal opportunity to show.

Creator-oriented: short video users are mainly divided into two categories: one is video users, the other is content creators. Douyin is "user-oriented", while Kuaishou Technology is "creator-oriented". In Douyin, "high fever continues to increase exposure opportunities", head creators focus a lot of user attention, which deters ordinary creators and grassroots creators (similar to Weibo Corp).

Kuaishou Technology set an exposure limit (Kuaishou will limit the traffic of high-quality content to about 30%, and 70% of the traffic is allocated to medium-and long-tail content), which also reflects Kuaishou Technology's original intention that "every user has room to show himself." Of course, this also causes Kuaishou Technology to increase his fans very slowly, and it is difficult to drive a large number of fans by a single popular style content. It takes Kuaishou Technology an average of 854 days to reach one million fans.

Gini coefficient: Kuaishou Technology proposed the experiment of distribution mode of "GDP+ Gini coefficient". Gini coefficient (Gini coefficient) is a common index used to measure the income gap of residents in a country or region.

The Gini index (Gini Index) is between 0 and 1. The higher the Gini coefficient, the higher the degree of inequality. In the actual operation of Kuaishou Technology, they introduced the "Gini coefficient" in economics into the regulation and control of community ecology as a strict quantitative index to ensure that while the community continues to develop, it distributes resources through fair regulation of attention, avoids the content head effect, and allocates more attention resources to ordinary people.

The algorithm core of each content product is to calculate the matching degree according to the user data and the content tag, and the theory is roughly the same. However, the design concepts and values of different products are different, and there are further slight differences in the setting of quantitative indicators and algorithm rules.

From a product point of view, Douyin puts "user experience" first, and pushes the most suitable and high-quality content to users through "matching weight elimination", "initial traffic pool operation", "re-mining content library" and so on. the end result is "fashion" and centralization.

Kuaishou Technology, on the other hand, follows the principle of "fairness and inclusive benefits", focuses mainly on "creators", and ensures that the content of every ordinary user is displayed through "head traffic control" and other means. the end result is a super community ecology (users who have posted videos are significantly higher than ordinary users in stickiness, duration and other dimensions). "user experience-oriented or content-oriented" is itself a paradox, perhaps we should not expect Douyin and Kuaishou Technology "life-and-death" struggle, difficult for a bosom friend than an enemy.

User profile: Douyin and Kuaishou Technology are becoming more and more similar, but they still have their own basic disks.

The user portraits of Kuaishou Technology and Douyin are becoming more and more similar, which is normal for the APP of two TOP 10. Kuaishou Technology has accelerated its penetration among the middle-aged and elderly people, and its intercity distribution is similar to that of Douyin, but it is extremely popular in the northeast and northwest. The creators are still dominated by young users, which meets the needs of the new era of "young users expressing and being recognized".

Douyin and Kuaishou Technology each maintain a considerable proportion of exclusive users, but the trend of their mutual penetration intensifies. As of June 2019, the number of monthly active users of Douyin, Kuaishou Technology and B stations reached 486 million, 341 million and 98 million respectively, of which Douyin and Kuaishou Technology lost nearly 600 million, basically covering all short video industries.

Taking into account the exclusive ratio (using only target APP users / target APP monthly active users), the exclusive rate of Douyin users was about 37.6% in June 2019 (nearly 50% in the same period of 18 years), and the exclusive rate of Kuaishou Technology users was about 32.7% (slightly over 40% in the same period of 18 years). The contiguous area of the two users is getting larger and larger.

In addition, another Questmobile report mentioned that more than 50% of Bilibili Inc. users were also browsing Douyin in August 2018, and it is estimated that the coincidence rate between the two continues to rise. Bilibili Inc. quietly online light video, Kuaishou Technology, Douyin have opened 10 minutes of video rights, stock user competition tends to be fierce.

Douyin has transitioned from being mainly female users to gradually balanced, with Kuaishou Technology having a relatively moderate male-to-female ratio; Kuaishou Technology has further deviated from the "Generation Z" tag, and the penetration rate of middle-aged and elderly groups has accelerated. As of June 2019, the proportion of male and female users of Douyin, Kuaishou Technology and Bilibili Inc. was basically the same in the distribution of mobile Internet, accounting for about 53% and 47% of the total, of which Douyin has changed more obviously in the past year. In October 2018, female users still stood at 55%, significantly more than male users.

In terms of age distribution, Douyin Kuaishou Technology is also similar, with Bilibili Inc. accounting for 38.5% of users under 25, Kuaishou Technology 29.1% and Douyin 28.4%. In the first half of 1919, Kuaishou Technology's penetration rate among middle-aged and elderly groups accelerated significantly, of which the proportion of Generation Z decreased from 41.2% to 36.5% (Kuaishou Technology marketing platform data are different from QM data in the industry, only for time span comparison), and the proportion of users aged 30-44 increased significantly, a total increase of 15.7%, from "young" to "universal".

Kuaishou Technology and Douyin also tend to be distributed in different cities, while Kuaishou Technology still has a deep foundation in the northeast and northwest regions. In first-and second-tier cities, Douyin only has a higher 3.1pct than Kuaishou Technology, with a very small gap; Kuaishou Technology accounts for 55.9% of the sinking market (third-tier and below), especially in rural areas. Kuaishou Technology has more obvious advantages and sinks more than Douyin.

By contrast, Bilibili Inc. 's users are still mainly in first-tier cities (56 per cent in 2016), especially on the east coast, but there has been a marked increase in users in third-and fourth-tier cities in recent years, with 54 per cent of new users in 19Q1 coming from sinking markets.

In the northeast and northwest regions, Kuaishou Technology's popularity is extremely obvious, of which Qinghai, Ningxia and northwest Gansu account for only 8% of users, but their TGI index (the proportion of Kuaishou Technology users / mobile Internet users in this region * 100) are all more than 400. The proportion of users in Liaoning, Heilongjiang and northeast Jilin is only 11%, but their TGI index is all close to 200. it can be seen that Kuaishou Technology has basically become the "standard" in the northern region.

Female users prefer to create, young users have a stronger desire to shoot, and Kuaishou Technology is more obvious than Douyin in the trend. Short videos meet the needs of "users to express themselves and be recognized by others equally", which is more common among young users who are "carefree and unrealistic", which is the fundamental reason for the rise and prosperity of UGC around the world.

From the deviation of content creators' portraits of Kuaishou Technology and Douyin (creator characteristics-user characteristics), when the proportion of users under the age of 30 is about 60%, the proportion of young creators is about 8-10% higher; in addition, the proportion of female users is also higher than that of men. Kuaishou Technology content creators account for 53% of female content creators.

The person who knows you best is often your enemy. The rapid rise of Douyin has not led to the decline of Kuaishou Technology, for both of them are changing (content, traffic, algorithm) and even "convergence", and showing the momentum of "more and more brave".

Douyin has a lot of MCN, and Kuaishou Technology is gradually promoting it; Douyin has a lot of brand advertising, but Kuaishou Technology has much less; Douyin has deep roots in first-and second-tier cities, while Kuaishou Technology flows deeply in low-tier cities. All these differences can be attributed to the slogan of the two families, "recording the good life" is Douyin, and "recording the world record you" is Kuaishou Technology.

The two choose different paths: "user-centered" or "creator-oriented", some like "watching", some like "shooting", some like "tide" and some like "truth". Maybe this is not a "life-and-death struggle".

After several years of development, both Kuaishou Technology and Douyin have occupied the "basic plate", rising from vertical application to national application, infiltrating and integrating with each other. Stereotyped impressions such as "Kuaishou Technology in the North of South Douyin" and "Kuaishou Technology in the female Douyin" are all quickly wrong, and Kuaishou Technology is also catching up in PGC and OGC.

Among the Internet unicorns, Kuaishou Technology has the highest boldness and reaction speed. Although Douyin has a slight advantage in product operation and commercialization, it does not necessarily have an obvious advantage over Kuaishou Technology in overall strategy and execution.

It is the same old saying: "it is difficult for a bosom friend but difficult for an enemy."