2018 Tik Tok complete analysis report!

I haven't written anything seriously for a long time. Seeing the increasing number of local Douyin users, I remembered that I wrote a Douyin product analysis report a year ago. At that time, the number of Douyin independent devices was only 30.8 million units/month. Now the number of Douyin independent devices has reached 263.27 million units/month. After a year, Douyin short video has emerged and developed rapidly. I believe many readers have personally experienced Douyin's strong expansion momentum.

Since 2018, TikTok has almost always occupied the top spot in the App Store. In mid-July, TikTok officially announced that its global monthly active users exceeded 500 million. I plan to pick up my old writing brush and re-analyze and interpret TikTok, so that I can have a new understanding of TikTok short videos now one year later.

1. Product Introduction 1. Product Background

A year ago, as of May 1, 2017, the industry's stock exceeded 250 million, and the industry's overall daily active users exceeded 50 million. Now, according to the iResearch Index's monthly number of independent devices of 263.27 million, Douyin short video alone accounts for 17%.

According to iResearch data, the year-on-year growth rate of the number of independent devices in the short video industry in June 2018 was as high as 173.1%, far ahead of all video service sub-categories, while the year-on-year growth rate of the number of independent devices in the overall video service in the same period was only 17.8%.

From the perspective of catalyst habits, the time users spend on short videos has continued to grow. The effective daily usage time of a single machine has increased from 21.1 minutes in the first quarter of 2017 to 33.1 minutes in the second quarter of 2018.

( Data source : iResearch mUserTracker)

(Data source: iResearch mUserTracker)

After a year of rapid growth, the short video industry has turned from a blue ocean into a red ocean. According to iResearch mUserTracker data, the month-on-month growth rate of the number of independent devices in the short video industry has gradually slowed down since 2018, and the user dividend period of the short video industry has begun to fade. In June 2018, the month-on-month growth rate of the number of independent devices in the short video industry was only 3.7%.

(Data source: iResearch mUserTracker)

Do you remember Kuaishou a year ago? At that time, Kuaishou’s data was nearly 30 times that of Douyin.

According to a report released by Aurora Big Data in July 2017:

As of July 23, 2017, Douyin had 20.86 million installations, while Kuaishou’s user base had already exceeded 500 million.

In terms of daily active users, Douyin's daily active users have been rising rapidly since April, increasing from 290,000 to 1.73 million in just four months, while Kuaishou's daily active users have remained stable at more than 50 million since 2017.

However, one year later, Kuaishou's advantage has been overtaken by Douyin.

1. Product Introduction

Everyone is already familiar with the short video app Douyin, so I won’t explain it in detail. Douyin was launched on September 26, 2016. The app description a year ago was "Tik Tok is a music creative short video App focusing on the new generation", and the app description a year later was "Tik Tok is a popular original short video sharing platform in China". From the app descriptions we can see that the product descriptions were different in different periods.

( Screenshot of the Apple Store app one year ago)

(Screenshot of the Apple Store app one year later)

A year ago, whether in the app screenshots or app descriptions, Douyin focused too much on telling users what kind of product it was and what it could be used for, and more on guiding users; a year later, Douyin instead highlighted the value of the product and provided users with a better experience.

1. Product Positioning

From the slogan of Douyin a year ago, “Focus on the new generation of music short video community”, we can know that the product is positioned as a music short video community suitable for young Chinese people; a year later, Douyin is no longer so vertically focused on music. Judging from the new slogan “Record a beautiful life”, Douyin has undergone a revolution and become a diversified short video UCG platform.

1. Current status of the product

(Data source: Qimai Data)

Through the above App store ranking data, we can clearly see the product's growth to dominance. On February 14, 2017, Douyin ranked 90th on the list. A month later, Douyin's ranking quickly rose from nearly 100th to 7th, breaking into the top 10 of the list; and until today, Douyin has long dominated the first place.

In April 2018, the State Administration of Radio, Film and Television announced that it had ordered Toutiao to permanently shut down its Neihan Duanzi app and public account (poor us, our friends at Neihan Duanzi, luckily we now have Pipixia). Of course, this also affected Douyin - it underwent a major internal overhaul, which resulted in a drop in its ranking.

(Data source: Qimai Data)

• The estimated total number of downloads on the iOS platform from September 18, 2016 to September 17, 2017 is: 19.44 million;
• Estimated total downloads from September 18, 2017 to September 4, 2018: 125.87 million

From the data of the two time points, we can see that there has been no growth trend in downloads since the product was launched. It started to grow in March 2017 and showed an explosive growth trend again in December 2017. In the analysis report a year ago, the downloads on the IOS platform only accounted for 15% of that a year later, which shows the fierceness of the explosive growth trend in downloads.

Having written this far, I would like to say one thing: So far, there has been no app whose growth rate can exceed that of TikTok. I think you have no objection to this?

II. Product Analysis 1. Product Life Cycle

A typical product life cycle can generally be divided into four stages, namely the introduction stage, growth stage, maturity stage and decline stage.

A year ago, TikTok was still a newborn, in a period of rapid product growth; after a year of rapid growth, let’s take a look at which stage of the product life cycle TikTok is in now.

(Product life cycle characteristics)

From the above characteristics of different stages of the product life cycle, we can know:

• Introduction stage: During this period, the market growth rate is high, demand grows rapidly, and technological changes are large. The main focus is on developing new users and occupying the market. Not much information is available on user characteristics, etc.
• Growth stage: During this period, the market growth rate is very high, demand grows rapidly, technology gradually becomes finalized, and industry characteristics, industry competition situation and user characteristics have become clearer.
• Mature stage: During this period, the market growth rate is not high, the demand growth rate is not high, the technology is mature, and the industry characteristics, industry competition situation and user characteristics are very clear and stable.
• Decline: During this period, the market growth rate declines, demand decreases, and the number of product varieties and competitors decreases.

(Data source: Qimai Data)

Tik Tok was launched on the iOS platform on September 26, 2016. From the above data, we can see that there was no significant change in the number of downloads from the launch of the product to March 2017.

The estimated total number of downloads from September 26, 2016 to March 1, 2017 is: 171,000 - This shows that a new product has been launched on the market. At this time, users are still unfamiliar with the product. Except for a few users who are looking for novelty, almost no one actually uses the product, which meets the characteristics of the introduction period.

Since March 2017, the number of downloads has skyrocketed:

The estimated total number of downloads from March 1, 2017 to September 17, 2017 is: 19.283 million - this shows that users are gradually accepting the product, the product has gained a foothold in the market, and the number of downloads is increasing rapidly, which is consistent with the characteristics of the early stage of growth.

The estimated total number of downloads from September 18, 2017 to January 31, 2018 is: 34.585 million. We can see that the explosive growth in downloads and the upward trend of the curve are consistent with the characteristics of a high-speed growth period.

Downloads have continued to grow since February 2018:

The estimated total download volume from February 1, 2018 to September 5, 2018 is: 91.536 million. Although the download volume is twice as much as that in the introduction period and the growth period, from the upward trend of the curve, we can see that compared with the previous stability, the curve has been at a peak stage and has not shown a clear upward trend. This indicates that the demand growth rate is not high, the market is mature, industry competition is fierce, and similar products are dividing the user base, which is consistent with the characteristics of the early stages of maturity.

(Data source: iResearch)

We can see from the above data that the growth trend of the number of independent devices of Douyin short video is that the growth fluctuated greatly before March 2018, and the growth gradually declined after March. The basic month-on-month growth rate remained at around 10%, which also well confirms the main characteristics of the early stage of maturity.

(Data source: Tik Tok short video)

According to the focus of different life cycle operations :

• Introduction stage: word of mouth (exceeding user demand expectations);
• Growth stage: growth pattern and speed;
• Mature stage: user activity and commercial realization;
• Decline stage: prevent and maintain user retention.

We can identify the life cycle stage of a product.

From the above screenshots of Douyin short video products, we can see that one year later, Douyin already has advertising, live streaming , corporate certification, external links, and Taobao shopping jump functions; according to the focus of operations in different life cycles, the focus of the mature stage includes commercial realization. It can be seen that today's Douyin has embarked on the road to commercialization.

Through the analysis of the above data, the growth trend is stable, it continues to grow, and the product is commercially monetized. We can conclude that the life cycle of Douyin short video has entered the early stage of maturity.

1. Problems facing the product at this stage

A year ago, the number of comments on all versions of TikTok on the iOS platform was 2,676. A year later, the number of comments on TikTok reached 11.43 million. Let’s first look at the problems the product is currently facing through user reviews.

(Data source: Qimai Data, one year later)

(Data source: Qimai Data, one year ago)

Since its launch in September 2016, Douyin has been updated about twice a month. As of September 8, 2018, Douyin has been updated 59 times, the minor version has been updated 42 times, and the medium version has been updated 15 times, from version 1.0 to version 2.0.

According to the data from Qimai, although there were different opinions from users in the early stage of launch, overall there were still mostly positive reviews. One year later, the number of comments on Douyin had exceeded 10 million, but the overall score rose to 4.9 points.

(Data source: Qimai Data)

Judging from the content of user reviews, Douyin has indeed gained a large number of fans through continuous efforts. The key points that attract users are summarized as follows:

① Allow users to shoot unexpectedly cool videos.

② Compared with before, Tik Tok focuses more on improving user experience and adding new features to meet user needs.

③ The richer image quality, filters and other effects in video shooting have increased user participation.

(Data source: Qimai Data)

Judging from the content of user reviews, non-5-star reviews also reveal some of the current problems of Douyin, which can be summarized as follows:

① The network is sometimes not very strong, and loading fails occasionally.

② The Anti-Spam mechanism is sometimes inaccurate and may misjudge normal videos or block normal users.

③ After Douyin was commercialized, users have been complaining about its advertisements, both in terms of quality and quantity.

④ As the content becomes richer, the pushed videos can easily make users forget themselves and waste time. This is not a bad thing, but some users complain that their time is wasted.

Social shortcomings

In addition to the above problems, Douyin’s social shortcomings are one of the issues worthy of attention. The “infinite scrolling” homepage mechanism is not conducive to building social relationships. Tik Tok’s distribution model is extremely dependent on algorithms, and the company is very confident in the algorithms, which is why it can achieve “hit the mark with one shot” and allow users to watch 300 posts a day.

However, in its algorithm logic, popular content will receive more recommendations, but long-tail content is difficult to be effectively distributed.

So we will find that the content on Douyin is becoming more and more stereotyped. Some new media operators have even summarized 15 patterns of popular Douyin content: such as throwing bowls, flirting with girls and boys, being shameless, and inspirational crying poor, etc. I occasionally see content with similar patterns when I scroll through Douyin every day.

Content convergence, no risk of continuous user retention due to operations

This distribution logic will also trigger a top effect and limit the distribution of long-tail content : some content can indeed become popular easily, with hundreds of thousands of likes, but only a few of them can actually become popular. This is still a long way from Douyin's goal of becoming a UGC platform where everyone can "record a beautiful life", because there will be a large amount of long-tail and unpopular content in UGC that needs to be distributed.

UGC platforms such as Kuaishou and Meipai focus more on the distribution model that combines "algorithm + attention". The way their users obtain content does not rely on popular recommendations, or even just on recommendations, but rather on a strong desire to actively search and discover. Its content acquisition behavior is point-to-point, the supply and demand relationship is more targeted, it is easier for users to establish connections and form interactions, and a real community atmosphere is created - realizing connections between people based on short video content is one of the important values ​​of short videos.

After being banned by WeChat , Douyin underwent a revision, changing the follow page to a design similar to Moments . However, I personally feel that the design is a bit useless. On the one hand, it lacks the immersive feeling of watching Douyin videos on a large screen. On the other hand, Douyin does not have common friends like WeChat, and Moments cannot form interactions. Many times communication is one-sided, rarely results in interaction, and fails to bring the joy of socializing in the circle of friends.

In daily use, there is a chance that users you have followed will appear. At the same time, in the version updated on August 22, 2018: 2.5.0, a new video list of Douyin hot searches was added, and private messages support the function of sending voice messages, but it is still necessary to strengthen the establishment of social relationships between users.

A highly centralized model needs to transition to verticalization

The highly centralized model has led to serious homogeneity of short videos and high creative costs. Today's Douyin is a bit like Weibo in 2013. Weibo was mainly supported by big Vs , and celebrities were the focus of Weibo operations.

In 2013, Weibo's operational focus shifted from top influencers to vertical influencers. According to Cao Zenghui, Vice President of Weibo, this step of sinking operations "saved" Weibo because:

Before that, the only areas with high Weibo readership were celebrities, media, and social politics; the number of other areas could not be said to be zero, but was close. Small and medium-sized Vs contribute a lot of content to Weibo, but have not formed a brand .

Today, Weibo has 55 key vertical fields, each of which has its own big V (vertical V), MCN (agency) and business model . The entire platform is thriving and has the second rise we are seeing now.

Today, Tik Tok can only cooperate with MCN agencies at high costs to develop high-quality content, and spend money to collaborate with celebrities to shoot short videos. But these are not factors in a virtuous cycle.

The truly virtuous cycle is: users themselves generate high-quality content.

But many users only watch but don’t shoot - maybe Douyin is just a place for them to kill time when they are bored, and they are not very motivated to participate in shooting.

Even if many grassroots users shoot videos, their enthusiasm for creation will be reduced because the video content lacks creativity and does not receive effective support.

In fact, the videos we actually watch are only those high-quality videos with hundreds of thousands of likes on the recommendations. Many long-tail videos are drowned in the vast number of videos and fail to get attention.

In the field of short videos, Meipai was the first to complete vertical construction. It has established 16 vertical content channels, such as beauty and baby. According to the Meipai team's public sharing, there are more than 300 subdivided interest areas under these channels.

What can be observed is that Meipai is conducting content operations , function customization and monetization attempts in different fields.

For example, in the "dance" field that Meipai is betting on this year, Meipai has launched the "dance follower" function for this purpose. This is a method of using functions to guide ordinary users to migrate to vertical users. On the other hand, we have recently noticed in the news that Meipai has cooperated with world dance competitions such as WOD and Arena to attract authoritative and influential people in this field to Meipai. With the influence of vertical big Vs, people in this vertical field can quickly gather together when they have a leader.

The Meipai team once shared a simple logic: the number of users in a single vertical interest area may be in the tens of millions, but when many vertical interest categories gather on Meipai, the user capacity is unlimited.

Therefore, after short videos focusing on pan-entertainment such as Douyin have gone through extensive Great Leap Forward growth, intensive vertical operations are a direction worth considering.

How can internet celebrities who become famous overnight achieve sustainable commercialization?

Today, Tik Tok can only cooperate with MCN agencies at high costs to develop high-quality content, and spend money to collaborate with celebrities to shoot short videos. This problem has been explained above. I am an old user of Tik Tok and have seen many cases of people becoming famous overnight, but then they frequently disappeared from the public eye.

Weibo and Meipai have gone further than TikTok in supporting content creators. It is true that some internet celebrities become popular overnight on Tik Tok, but how can we help them remain popular? How to convert popularity into commercial value? It’s Douyin—and it’s also a lesson that Toutiao needs to learn.

Aha moment is wonderful, but if you can't control it, you will become addicted

Many people are "addicted" to Douyin. They watch Douyin before going to bed at night, and keep watching until the early morning, which leads to lack of sleep and affects their work the next day. Some people stop to watch Douyin while doing something, and several hours go by while they are watching, and the original things are delayed.

When TikTok continues to occupy users' minds and time, it is actually time to be vigilant. Because it is closely related to "human weakness", it can grasp your excitement points and know what you are addicted to, just like drugs, making you immersed in a state of complete selflessness.

However, time is wasted on other people's beautiful things. Apart from satisfying one's own "sense of voyeurism", all that is left for oneself is inferiority and emptiness.

Tik Tok has successfully satisfied the low-level consumer entertainment needs as well as the need for self-display and narcissistic self-expression.

Douyin is facing a moral test. It chooses to continue to increase the length of user stay or to prevent users from becoming addicted. However, great products should be built on morality. It is said that there have been cases of people being fined for using Douyin at work. The control and supervision of minors are issues that deserve special attention. It may go the way of Honor of Kings . This may be a bit exaggerated, but don't forget that the closure of Neihan Duanzi in April this year pushed Douyin to the forefront, which is a warning. Not only in terms of addiction, but also government control and content supervision, Douyin still has a long way to go.

1. Improvement suggestions and breakthroughs

Network optimization: According to the App Store's reviews, some users may encounter poor network performance. With the rapid increase in the number of TikTok users, further optimization is needed in this area. Video loading speed and network quality are important aspects that affect user experience .

Improve the Anti-Spam mechanism: Through the comments, we can know that many users have reported that their accounts have been blocked. On the one hand, this is to control the content, and on the other hand, it is affected by government control. As a result, when the number of users is growing rapidly, a large number of users have been blocked and their videos have been deleted.

(TikTok short video APP interface)

During the Qingming Festival holiday this year, Kuaishou launched the recruitment of 3,000 content reviewers. After clarifying that algorithms should have values, Kuaishou took action to give algorithms more values. Before Kuaishou, Toutiao has already recruited a large number of people for similar positions, and Meipai has also launched a human-machine combined content review mechanism.

Tik Tok has also added a user content reporting mechanism, but I personally feel that this feature is rather useless.

In order to test this function, I kept watching TikTok and reporting videos; I was able to join according to the application requirements, but I could never pass the last criterion. I don’t know what the real review criteria for this function are, and there is no clear prompt to users; if the number of people is full, users should be prompted, otherwise the enthusiasm of users to report will decrease, but users can never apply, and the result is the same.

An imperfect Anti-Spam mechanism will bring a bad experience to users. The relevant mechanism should be improved and optimized, and users whose accounts are mistakenly blocked should be provided with appeal channels , and manual investigation should be carried out to provide timely processing, and users should be prompted with the processing results.

Strengthen product social interaction and improve algorithm recommendation mechanisms: Whether it is Facebook or Twitter, its timeline was originally based on the Follow relationship. The logic is "the interests of the people I follow are my interests." The entire content dissemination path is based on fans and has nothing to do with the algorithm. To this day, WeChat Moments still uses this distribution logic. The logic of algorithm distribution is that "the algorithm understands your interests" and provides you with personalized content for thousands of people. How far the content can go is mainly determined by the algorithm.

However, the defects of relying solely on algorithms are very obvious. There will be the so-called "information cocoon" effect, that is, the algorithm will make more and more accurate recommendations based on your previous behavior, but it will also make the content you see narrower and narrower. For example, if you always watch technology videos, the algorithm will keep pushing technology videos to you, but you may not only be interested in technology videos.

A year ago, due to the limited user base, there would be a lot of repetitive video recommendations on UGC content, such as amazing makeup techniques and "Never Look Back". Sometimes a similar video would appear for every 5 videos viewed. The frequency was too high, which easily made users lose their sense of freshness.

Today, TikTok relies on its own powerful algorithm recommendations, but the content that users see is becoming narrower and narrower. I remember when I first used TikTok, it was full of technical content and I was deeply attracted to it. But now I see very few technical content videos. Not paying attention does not mean I don’t like watching them.

(TikTok short video app interface, one year ago)

(TikTok short video app interface, one year later)

We mentioned adding a private messaging function a year ago. The private messaging function will help active users establish connections, promote interaction between users, and thus increase user stickiness. Now you can send private messages after following Douyin. In version: 2.5.0, private messages support sending voice messages.

In today's version, we can see: the previous story shooting function is cancelled, the swipe left is replaced by the previous discovery page, the home page focus is separated and replaces the original discovery sub-page, there is one more comment than before, which increases the interaction between users and establishes connections between them.

If you want to make a breakthrough in algorithms and social networking, I personally think there are three ways to do it: recommend content that everyone should pay attention to, such as hot news. Now Douyin has added a hot search ranking function on the discovery page; the algorithm guesses your interests and recommends some content that you "may be interested in"; introduce follow-up relationships, so that the people you follow recommend more content to you, you will have the opportunity to be exposed to a wider range of interests, which can also help distribute long-tail content.

Weibo introduced discovery flow in addition to attention; Toutiao introduced social networking in addition to algorithms and launched the "Thousands of People and Millions of Fans" plan, at the cost of breaking the traffic balance and forcefully importing hundreds of thousands of fans to some gold V accounts; Meipai adopted a double feed column design on the homepage to integrate recommendations and attention, all of which are seeking a balance between algorithms and attention. However, the key to combining algorithms and attention is to take both into account at the distribution level. Douyin has been constantly improving, and it also needs to find a balance between the two.

Profit model of internet celebrities: A year ago, Douyin was in the growth stage of its product and had no clear profit model. It focused more on user growth . Today, Douyin’s profit model is basically mature. It uses a mechanism for paid gift giving like live streaming, and advertising mechanisms such as splash screen ads and information flow ads . Now that Douyin has commercial operations, internet celebrities can have channels for monetization.

The problem that Douyin is facing now is how to commercialize internet celebrities in the long term. Toutiao and Douyin are more like free-range for big Vs - I will give you a grassland with abundant grass and water, and the rest depends on you.

As long as you have a certain level of content creation ability, you can get distribution and fans, but it is difficult to get more support. This is similar to the way WeChat public accounts work, that is, apart from the infrastructure, there is no more operation or commercial support.

In contrast, players such as Weibo and Meipai pay great attention to professional support for creators. In terms of commercialization, Weibo provides a micro-task advertising platform and cooperates with many MCNs. Every year, the V Influence Summit will announce how much money creators have transferred through Weibo. Meipai has reached cooperation with more than 50 domestic MCN agencies, providing traffic support and even delivering high-quality talents to MCN agencies.

We can see that Miao Daxian, Liu Yang and other talents who grew up from Meipai have established their own companies or MCN agencies and become more important participants in the short video industry. In order to help influencers make money, Meipai launched the short video marketing platform "Meipai M Plan", which opened up influencers' e-commerce sales channels to enable them to "watch and buy".

Douyin is currently following up on commercialization and building e-commerce channels. With Douyin's traffic advantage, it is not difficult to develop these businesses. However, although we often hear media analysis on how many people have become famous overnight on Douyin and how much advertising money they have made, there are no cases like Miao Daxian and Liu Yang that can be operated as a company. Douyin still needs to provide more tools and operational support for the commercialization of influencers.

Anti-addiction mechanism: On March 19, 2018, Douyin announced that it would add two new features, “time limit” and “risk warning”, to the next version. On April 10, the anti-addiction function was officially launched, becoming the first non- gaming application to join the anti-addiction system. When the user has used it continuously for 90 minutes, a line of text will appear on the short video being played to remind the user; the 2-hour time lock is activated by a password set by the user. Once the cumulative usage time in a single day reaches 2 hours, the system will automatically lock and the user will need to re-enter the password to continue using it.

Personally, I think anti-addiction is a kind of hypocrisy and a false proposition. Before you scroll down on Douyin, you can’t guess what the next video will be. It gives you a random reward just like a slot machine. Automatic looping allows your brain to receive stimulation at the first opportunity and turns the already limited user experience into a bottomless pit.

There are many such "addictive" functions and designs, all of which are carefully designed by technology companies based on the physiological characteristics of the brain - humans often cannot resist their own instincts and desires.

The market competition is so fierce, and the "anti-addiction" function is contrary to the goals of Internet companies. Therefore, it is hard to imagine that an Internet company would sincerely promote "anti-addiction" and give away precious user time. Introducing an anti-addiction function is the same as printing a message on cigarette boxes saying that quitting smoking is good for your health.

Let’s not talk about the anti-addiction function being a question of hypocrisy on the part of technology companies. It is not uncommon for minors to lack self-control and get injured or even die from imitating online videos.

Although Douyin has launched an anti-addiction mechanism for minors, with reminders to avoid imitating dangerous actions when playing videos, it is hard to say whether it really plays a practical role; the user's usage scenarios are unpredictable and restricted, and it not only depends on the product itself, but more on parental supervision, which is not something that one product can fully control.

Recently, Didi has been on the hot search list because of the murder of a passenger, and has once again been pushed to the forefront of public opinion. For a time, many people have criticized Didi, and many netizens have expressed their hope to completely ban the Didi platform, and the voices are getting stronger and stronger.

Think about it from another angle: Is taking a taxi necessarily safe? Why not choose a private car but choose a ride-sharing car? Where has your personal self-protection awareness gone? There are too many confusing issues here. Of course, Didi also has its own responsibilities, but it will not be completely banned. At least it has improved people’s travel convenience and has become an indispensable product in daily life.

This is a bit off topic. Why do we mention Didi? If one day minors get into trouble for imitating TikTok short videos, or if minors give tens of thousands of yuan in rewards during live broadcasts, or if others are injured because of minors imitating others, will they also be overwhelmed by public opinion?

Although Douyin has established corresponding mechanisms, people are always irrational. When something angry happens, they always want to vent and find an outlet. Coupled with the "hindsight effect", when nothing happens to you, everyone is peaceful and doesn't pay attention to you. But when something happens to you, they will hold on to it and start a thorough investigation. By that time, will public opinion say that TikTok is an internet drug and that it should be shut down?

Minors are not mature and rational people, so it is necessary for parents to intervene and guide them appropriately. On this issue, Douyin’s anti-addiction mechanism is simply to put on a layer of protection for itself so that people will not accuse you of not taking anti-addiction measures.

In order to prevent addiction, Douyin needs to combine big data and usage scenarios that are beyond the control of minors, improve functions such as content filtering, anti-addiction, and risk control models for minors, as well as the balance between "anti-addiction" functions and "addictiveness". This road is indeed a long way to go.

1. The logic of content selection

After Douyin released the video, it benefited from Toutiao's recommendation algorithm. One of its biggest technical advantages is the algorithm that Toutiao is most proud of. Douyin's founder Liang Rubo is also Toutiao's technical director. Douyin, which uses the algorithm, can be said to be even more powerful. Tik Tok has added an algorithm recommendation model at the product level to ensure the efficiency and decentralization of video distribution.

Everyone should be familiar with the distribution mechanism of Toutiao. I believe that Douyin also adopts an extremely similar distribution mechanism. Generally speaking, the order of priority includes: deduplication, review, feature recognition, recommendation and manual intervention.

Deduplication mechanism

Whether there is identical or highly similar video content in the system; if so, whether the source of the video content is authoritative and valuable.

The number of identical videos in the system is not strictly controlled in Douyin. After all, the content of the videos shot by each person is different. If it is required according to the standards of Toutiao, then imitation videos will be buried in the sea.

It is mainly aimed at uploading other people's original videos, reposting or copying of similar videos to de-duplicate them, so that the simultaneous or repeated video content will not appear in the user's information flow .

Audit Mechanism

Douyin also includes machine review and manual review, with machine review as the main method and manual review as the auxiliary method.

The machine review algorithm includes a library of content that can be blocked. Once content that matches the library appears, including titles and video content, it will not pass the review.

(Criteria for failure to pass the audit)

In the spring of 2018, a series of combined measures from national regulatory authorities put unprecedented pressure on the short video industry. First, Kuaishou and Huoshan Video, a subsidiary of Toutiao, were ordered to carry out comprehensive rectification due to problematic videos.

A few days later, the rectification of Huoshan Video had not yet been completed, and another product of Toutiao, "Neihan Duanzi", received a notice of permanent closure. Douyin has been continuously improving its video content review mechanism and strengthening its review efforts. This has also led to widespread complaints from Douyin users about their posted videos being deleted and failing the review.

Feature Recognition

After users shoot/upload video content and submit it for review, Tik Tok will identify the video content and title content to determine the interest groups to which the video should be recommended. After determining these interest groups, the system will classify and compare the video with the platform videos; then, the published video content and title will contain a large proportion of interest feature vocabulary keywords , and will be labeled with the interest feature label.

Recommendation Mechanism

In order to ensure that popular content is seen by more people and unpopular content does not take up too many recommendation resources, it is often recommended in batches. It will first be recommended to a group of users who are most interested in it, and the next recommendation will be decided based on the data generated by this group of users.

Similarly, the number of new recommendations for a video will be based on the number of views, completion rate, likes, comments, and shares of the previous recommendation.

Human Intervention

As mentioned above, the short video industry is recruiting a large number of content reviewers. Video content review is many times more difficult than text review. Manual review is performed on video content that cannot be recognized by machines. A user reporting mechanism is also set up in the product to filter out various UGC and PGC content and maintain the reputation of Douyin Short Video as a "high-quality content community", thereby giving users a better experience.

(13-second video attributes of Tik Tok)

Let me analyze it from a layman’s perspective: Why is there human intervention? Why is it impossible to eliminate videos that violate platform rules and still appear?

We can see that:

A Tik Tok video is 13 seconds long and has a frame rate of 30 frames per second. A video is made up of images frame by frame, so a short video consists of 390 pictures. The number of videos released on Tik Tok every day is at least one million, so if we calculate it at 1 million, 390 million pictures will be generated every day.

Currently, the accuracy and precision of image recognition are both over 99.5%, and the response time for a single image is less than 0.2s, which means it takes 78,000,000 seconds, or 361 days. This requires an artificial intelligence supercomputing platform built with a large-scale GPU cluster to support image processing at the billion level. This is a huge financial and human investment for any Internet company that needs to conduct content review.

Artificial intelligence review uses learning algorithms to update the feature library to judge videos, but it is still a machine and there will be omissions, and users will also change the way they post, so human intervention is still needed to review and report video content.

1. The embodiment of content selection logic

(TikTok short video app interface - one year ago)

From the product screenshots taken a year ago, we can see that these four short videos all participated in the "My Body Learns English" challenge. Compared with the first selected short video, the number of comments, reposts, and likes on the following short videos are higher than the first one. After several repeated playbacks, it was found that the rhythm of the following short videos was not in sync with the BGM, and the movements were not well kept to the rhythm. It shows that whether the actions in Douyin can be consistent with the rhythm of BGM is a very important criterion.

(TikTok short video app interface - one year ago)

(TikTok short video app interface - one year later)

From the product screenshots from a year ago, we can see that the first two are selected and the last two are ordinary. If we follow the general judging criteria of the number of comments, reposts, and likes, we will find that the comments, reposts, and likes of the last two short videos are much higher than those of the selected short videos.

From here we can see: in terms of selection logic, Tik Tok is a vertical content product, mainly based on music. The latter two are inconsistent with the product positioning. Therefore, vertical music short videos have a much greater chance of being selected than other videos.

From the product screenshots one year later, we can see that the comments, reposts, and likes of the short videos are much higher than those of the short videos selected a year ago. Of course, the user base of Douyin cannot be compared with that time, but from the selection logic, these videos should all be selected.

Everyone is familiar with black face, right? It is said that every video is selected, but from the screenshots we can see that suddenly one day, the videos are no longer selected, although the data of each video performs very well. As mentioned above, Douyin has carried out a major internal rectification and adjusted the logic of short video selection, and has not yet opened the mechanism of selected videos.

(TikTok short video app interface - one year ago)

(TikTok short video app interface - one year later)

From the product screenshots taken a year ago, we can see that these four short videos are selected, but none of the four short videos are of musical style and do not conform to the tonality of the Douyin product.

At present, Douyin short video is mainly focused on acquiring new users. The number of independent Douyin short video devices has exceeded that of Meipai. The number of users who create vertical content cannot compare with the number of comprehensive users of Meipai.

From this we can see that Douyin actually changed its selection logic in order to acquire more users, and is constantly adjusting its selection logic according to the types of users it needs to acquire, in order to acquire more users.

Today, one year later, our previous judgment on Douyin's selection logic has been well verified. From the product screenshots one year later, we can see these four short videos: Typhoon Mangkhut, Steel Teeth is So Handsome, Are You Really Healthy?, Eyes Can Talk. The data of each short video can meet the selection standards, but the video content has little to do with music.

It can be seen that in order to acquire more users, Douyin changed its selection logic, from a vertical music short video platform to a diversified short video content platform, in order to meet the needs of various users and acquire more users.

1. Content weight analysis

When a video is initially uploaded, the platform will give you an initial amount of traffic. After the initial traffic, the platform will judge whether the video is popular or unpopular based on the like rate, comment rate, and forwarding rate. If the video is judged to be popular in the first round, it will be disseminated a second time. This is the recommendation mechanism mentioned above.

When you get the best feedback the second time, you will be recommended more traffic. On the contrary, if the response is not good in the first wave or the Nth wave, it will no longer be recommended. Without the recommendation of the platform, the probability of your video becoming popular is very small because there will be no more traffic to see you.

The first step to making your video popular is to be seen by others. If you block the first step, you will have to rely on your friends’ likes later on. It is not difficult to see the thinking logic behind this algorithm: intelligent distribution, superimposed recommendations, and popularity weighting.

Detailed algorithm introduction

Intelligent recommendation traffic pool. When a new video is transmitted to Douyin, Douyin will know that it is a new video through comparison, and then give you the first recommended traffic. The new video traffic distribution is mainly based on nearby and followed, and then intelligently distributed with user tags and content tags. For example, if the new video has a high playback volume, low bounce rate, high completion rate, high likes, high forwarding rate, high comments, and many interactions, this video will have the opportunity to continue to increase traffic.

Superimposed recommendation means that if new videos are intelligently distributed with a playback volume of about 100vv and the forwarding volume reaches 10, the algorithm will judge it as popular content and automatically weight the content, and give you superimposed recommendations of 1000vv. When the forwarding volume reaches 100, the algorithm will continue to superimpose recommendations to 10,000vv. When the forwarding volume reaches 1000, the algorithm will continue to superimpose recommendations to 10wvv, and so on.

So those Douyin hosts who get millions of views overnight are also confused and have no idea what is going on. In fact, it is the weighting of the big data algorithm.

Of course, the evaluation criteria for superimposed recommendations are based on the comprehensive weight of the content. The key indicators of the comprehensive weight include: playback volume, bounce rate, completion rate, likes, reposts, comments, and interactions. The weight of each level is different. When it reaches a certain level, a mechanism combining big data algorithms and manual operations is used.

After weighting the popularity, we actually checked nearly a hundred popular Douyin videos and found that all the videos that became popular overnight and the videos in the Douyin recommendation section had millions of views. The comprehensive data: view volume, bounce rate, completion rate, likes, reposts, comments, and interactions were all very good without exception.

It can be seen that after being tested by a large number of users and layer-by-layer weighted popularity, it will enter the recommended content pool of Douyin and receive the baptism of tens to millions of large traffic. The popularity weight will also be based on time to select the new and remove the old. The popularity of a popular video will last for a maximum of 1 week, unless a large number of users imitate and follow it. Therefore, a stable content update mechanism and the ability to continuously output popular products are also needed. In the popular strategy to pass the level, the algorithm only points out the path, while content is the golden key to activate human nature. The video's playback volume, bounce rate, completion rate, likes, reposts, comments, and interactions are all votes from the crowd's hearts, and the secret to passing the level is only content.

Weighted bonus items

Account basic weight, improve your own information, the more complete the better. Including avatar, nickname, mobile phone number, Weibo, WeChat, headlines, etc. The more detailed the better.

Because it is a dual review by machine and manual, once the machine conducts the review, a large number of low-quality accounts will be eliminated. When an account frequently posts advertisements or other illegal content, the account will also be put in a small black room and will not be recommended by the system.

In addition, do not upload videos repeatedly, as this will also affect the basic weight of the account. If a video is identified as having a watermark, removing the watermark and uploading it again will not result in much playback.

The video needs to be creative and have highlights. The video is only 15 seconds long. In these short 15 seconds, if there are no highlights and no turning points, people will not interact with you. There is also a blocking function. Once a user blocks you, it is a very serious matter because your short video will no longer be recommended to the user in the future.

Appearance, follow the 3B principle: Beautiful—beautiful scenery, beautiful women, handsome boys; Beast—animals (wild beasts); Baby—infant; There is a method to achieve any success, you must choose the right method.

They have their own traffic. Celebrities, internet celebrities, and companies all have their own traffic. They don’t need system recommendations and will naturally have a large number of fans to support them. It is easier for such users to generate popular content.

Video quality. A year ago, the product analysis showed that Tik Tok is a short video vertically classified based on music. In this regard, it also has its own tone and selection logic, which is mainly reflected in the background music, picture quality, and rhythm of the video content.

Platform requirements: the operational focus of different product life cycles is different. Douyin was still in a rapid growth stage a year ago, and what it needed was attention to attract new users . But today, Douyin is in the early stages of maturity, and it is paying more attention to active users. The video content has expanded from vertical span to comprehensive, and the various types of videos in each category are beginning to become saturated. When choosing video content, it is necessary to pay attention to the needs of the platform. The traffic support distributed by the platform is also different in different periods.

Interest characteristics, video distribution is intelligently distributed in conjunction with user tags and content tags. If the size of this type of user is small, the indicators of your video will not be much better. However, exceptions are not ruled out. There are too many videos that become popular overnight without any basis to follow.

Emotional resonance. Now many people have summarized many templates for popular content. This involves human nature, which is difficult to explain clearly. As the saying goes, “scarcity makes things valuable.” When the streets are filled with videos that resonate with people’s emotions, do you still think you can become an overnight sensation? It all depends on the content of the video and a bit of luck.

Hot topics, this needs no explanation. Friends who have done operations know that hot topics are valuable for operations, but please don’t chase hot topics just for the sake of chasing them. If it cannot bring value to users (emotional value/practical value, etc.), then don't pursue it. Secondly, for hot topics that suit you, if it’s not new, then it has to be fast; if it’s not fast, then it has to be new.

Update frequency. So far, I have not seen an account that can capture a large number of fans with just a few videos. The product's operational focus in different periods and when the product needs to improve data will have different weights. For example, user A posts one video per week and user B posts one video per day. Which user do you think will get more opportunities? But remember not to lower the quality of the video for the sake of comments and quantity, as this will also affect the weight.

(Tik Tok weight description)

Although these are just personal opinions, judging from the current data, Douyin is very cautious in selecting and recommending videos.

If a video can achieve all the above weights, I think it will be hard for it not to be popular. If it doesn’t become popular, then there is something wrong with its character - what it lacks is luck, so hurry up and go to the temple to burn some incense.

Therefore, for students in operations, creating a hit product is a long and arduous task.

1. Centralization and Decentralization

This topic also gives me a headache: I just finished writing one question, and the next question requires me to think for a long time; I jumped from one pit into another big pit. Perhaps it is this process that allows me to keep improving.

I have still read a lot of information and articles, and would like to express my own views here. Douyin has benefited from the recommendation algorithm of Toutiao. An algorithm recommendation model has been added at the product level to ensure the efficiency and decentralization of video distribution, but the product operation is not completely decentralized.

(TikTok short video APP interface)

Do you still remember these great gods? These are all the veterans of Tik Tok. When I was watching videos in the early days, their appearance rate was 100%. I don’t believe it if you tell me there is no centralization. At that time, the number of videos posted by UGC users could not support the number of users browsing videos, and it was very difficult to shoot videos on Tik Tok. At that time, it was all about technical skills, and it was not easy to be recommended. It was impossible to recommend some ordinary videos to users at that time. Doing so in the early stage of the product is a dead end. A new user who comes in will see only ordinary videos. How can he stay ?

Do you still remember the offline parties held by Douyin? Most of the offline advertising content used these veteran figures for promotion. From the beginning of Douyin, it has been doing centralized things, but it was not obvious in the distribution mechanism.

Of course, we can also understand that Douyin at that time did not have the current user base. If it wanted to promote itself, it had to have representative figures.

As the number of TikTok users increases, we can see that these veteran users do not have many fans, and are not given any privileges. This is in line with the argument of decentralization, but the main reason is that now there are enough videos posted and the platform’s demands are different from what they were then.

(TikTok short video APP interface)

Judging from the latest updated version, on August 22, 2018, Douyin was updated to version: 2.5.0, and a new Douyin hot search video list was added. As a representative of "decentralization", Douyin is gradually moving towards "centralization".

There are currently three categories in the hot search list: challenge themes, videos, and music. We will mainly talk about the hot searches for videos. The other two categories have little impact on traffic distribution and are updated less frequently. You can see from above that the hot searches for videos are updated every 10 minutes, and you will find that the same video will appear in the rankings.

There is another interesting point. I don’t know if you have noticed that the popularity calculation method is actually lower after ten minutes than before. At 22:50, the video playback volume was about 1.7 million, and at 20:00, the video playback volume was about 1.85 million, but the popularity showed an increase of 400,000 - this is another black box algorithm.

When the observation began, the video playback volume was 1.5 million. In just 20 minutes, it increased by 350,000. Isn’t this a manifestation of centralization?

We can also see that this function is at level three - users need to swipe left and click on the hot search list to see it. I believe that many users are still unaware of this function.

Therefore, it can be seen that Douyin is very cautious in its centralized operations. It hides the hierarchy very deeply and updates every 10 minutes to ensure that more videos have the opportunity to be shown, all in order to find a balance between decentralization and centralization.

The recommendation distribution mechanism mentioned above, although it appears to be decentralized, aims to recommend content that different users are interested in, meet personalized needs, and provide personalized services to different users. However, the recommended content is relatively concentrated. After the video is released, it will be superimposed and recommended, so as to obtain continuous traffic. The more high-quality content is exposed, the less exposure the general content will get. In other words, there is a very obvious "bias".

Moreover, Douyin’s content distribution is a complete black box for content producers, and you don’t know its internal rules. The distribution method is completely determined by the platform. From this point of view, it is already a centralized distribution. No matter how thoroughly we analyze it, it is impossible to 100% understand the algorithms and principles involved.

In content production, decentralization allows everyone to create content, but not all content created is high-quality. Centralized approaches make it easier to provide traffic support for high-quality content. Decentralized thinking allows users to be more autonomous, but centralized measures are also needed to ensure that products are in line with the company's values. Operational focus is dynamically changing, and products are constantly developing and improving. All of these reflect that centralization and decentralization complement each other and are indispensable.

III. Product Data Analysis 1. Purpose of Data Analysis

Tik Tok is currently in the early mature stage of its product life cycle, which is a stark difference from its rapid growth stage a year ago. From the vertical music technology flow to the current flourishing of various industries, let's compare the data from a year ago to see what changes have taken place in the current Tik Tok user group.

1. Changes in user data

Let’s first look at the overall user data of Douyin through Aurora Data . Douyin has achieved breakthrough growth this year, with a growth rate far exceeding that of industry competitors, and the number of active users is also higher than that of Kuaishou.

(Data source: Aurora Data)

(Data source: Aurora Data)

According to the above data, Kuaishou's penetration rate reached 25.6% in March. In Toutiao's short video matrix, Douyin's short video penetration rate was 16.5%, a month-on-month increase of 121.3%. In June, Douyin surpassed Kuaishou to rank first with a penetration rate of 29.8%, an increase of 80.2% month-on-month, and the average DAU exceeded 100 million; Kuaishou, which ranked second, had a penetration rate of 24.8% in June, a decrease of 3.2% month-on-month. We can see that Douyin is growing very fast. As of today, it has surpassed Kuaishou.

(Data source: 2018 TikTok Research Report)

Through the data from the above 2018 Douyin research report, we can see that compared with the same period last year, the proportion of male users has increased and has shifted to people over 26 years old.

(Data sources: iResearch, Baidu, 360 Index)

By comparing the above user data before and after, we can see that there have indeed been some changes in Douyin users. For example, the data in the research report jointly released by Douyin and Haima Data are consistent, the age group of users is decreasing, and the proportion of male users is increasing.

The iResearch Index is user usage data, while the Baidu and 360 Indexes only focus on product data. It is not difficult to see that the data gap between the iResearch Index and the research report is the smallest.

(Data source: Aurora Data)

If we look at the Aurora data from a year ago, the majority of users are female, accounting for as high as 78.8%. In terms of age distribution, we can see that users aged 20-24 and 25-29 account for the highest proportion, accounting for 37.3% and 29.4% respectively.

In the early days of Douyin, it attracted a lot of young ladies. Without so many young ladies, how could so many users follow it?

According to the statistics of Jiguang Data in February 2018, female users are still the majority, accounting for 66.4%; in terms of age distribution, users aged 20-24 and 25-29 still account for the highest proportion, but it can be seen that the age distribution shows signs of sinking.

(Data source: 2018 TikTok Research Report)

Through the data from the above 2018 Douyin research report, we can see that the user city distribution has sunk, and people in offline cities have become the main growth force. Seeing the current Douyin user growth chart reminds me of Kuaishou's user group, which is basically overlapping.

(Data source: Aurora data from one year ago)

This data report is the earliest user data. Since its inception, Douyin has established its position as a "music short video community for young people". First- and second-tier cities account for 50% of the total, while a year later, third- and fourth-tier cities account for 50% of the total.

Judging from the age distribution and geographical distribution of users, it is not difficult to see that Douyin’s current strategy is to become “Kuaishou-like”. Its strategy of starting from the high-end population and then moving down to the low-end users is colliding head-on with Kuaishou, which puts Kuaishou under great pressure. Who will be the winner in the future? I believe we will see the result in another year.

(Data source: Qimai Data)

From the above data, we can see:

• One year ago, the proportion of VIVO model users: 33.3%, the proportion of OPPO model users: 18.6%
• One year later, the proportion of OPPO users: 39.39%, the proportion of Huawei users: 28.93%

VIVO is highly similar to OPPO in terms of products and marketing . One year later, OPPO became the number one model for Android TikTok users. Huawei's growth rate also significantly surpassed VIVO in previous years, becoming the mainstream model for users.

(Data source: Analysys one year ago)

(Data source: Analysys one year ago)

(Data source: Analysys one year ago)

According to the usage data of Douyin short videos from Analysys Consulting Group one year ago, the average number of times Douyin was used per person was 5.21 times per day, and the average daily usage time was 31.23 minutes. The peak usage hours for users are around 12 noon and 6 pm.

(Data source: Douyin Blue V White Paper)

According to the "TikTok Enterprise Blue V White Paper" report, the average daily launch times of the short video industry in China is as high as 8 times, which is much higher than other video formats, and the user behavior level shows a higher user stickiness than a year ago.

(Data source: Douyin Blue V White Paper)

From the "Douyin Enterprise Blue V White Paper" report, we can see the data on Blue V release time periods, Douyin playback time periods, and Blue V video playback volume. It is not difficult to see that Douyin's peak hours have not changed significantly. The user usage peaks are concentrated at 12 o'clock, 18 o'clock, and 22 o'clock, which correspond to the usage scenarios of after-get off work lunch break and after-meal entertainment, on the way home from get off work and after-meal entertainment, and before going to bed at night.

This is the end of the user data. Although some of the data comes from product monitoring data, most of the data comes from third-party data reports. I will directly recommend them to everyone for detailed review. I will not select and explain them one by one. I will recommend a few analysis reports and recommend that you take a look at them to have a more comprehensive understanding of Douyin’s current user data.

• Douyin Enterprise Blue V White Paper
• White Paper on Short Video and City Image Research
• "2018 · Douyin Research Report" - jointly produced by Miaozhen System and Haima Cloud
• Kuaishou & Douyin User Research Report

These are jointly released and do not represent real data, but they can still serve as some guidance, mainly focusing on Blue V operations, short video research, Douyin product analysis, and comparative analysis of competitor Kuaishou.

1. The battle between TikTok and Kuaishou

A year later, Douyin has moved from vertical to diversified in terms of content. The above data comparison records the changes in Douyin users. Douyin is frantically sinking to third- and fourth-tier cities to acquire users, following the strategy of entering from high-end groups and then sinking to low-end users.

Although Kuaishou and Douyin have different product positioning and face different user groups, if we compare Kuaishou and Douyin, Kuaishou is a square and Douyin is a theater. The square is just a simple platform. If you want to dance, you can come and dance. If you want to chat, you can also move a stool and chat.

The theater has admission requirements and even more importantly, performances. Whether it is the soundtrack of "Seaweed Dance" or other editing techniques, Tik Tok is trying its best to help users make the video "performances" more entertaining.

If the two products are facing different user groups, when the user volume reaches a bottleneck, should they snatch users from competing products? Today's Douyin has moved from vertical to diversified in order to expand its user base. When Douyin and Kuaishou encounter bottlenecks, there will inevitably be a battle to grab users. Of course, we cannot ignore the existence of Tencent Weishi here. Maybe it will be a more interesting battle. As for the opponent Meipai, let’s give him a song called "Liangliang".

4. Bad ideas about advertising

I have previously written an article about improvement suggestions for Douyin advertising optimization . At that time, the ads had just started to run and user feedback was not very good. So I had a sudden idea to adopt the user review mechanism of Qiushi Encyclopedia , so that users could also participate in the ad review and add user weight to decide whether the ad should appear in the information flow.

(Data source: Qiushi Encyclopedia review page)

Today, a year later, advertisements on TikTok have become the norm. An advertisement will appear after watching a few videos. Not only has the number of advertisements increased, but the quality of advertisements has also deteriorated. In the past, advertisements were at least somewhat interesting, but now advertisements are simply impossible to evaluate. So they came up with another crooked idea, which is to improve the quality of advertising content and reduce the frequency of shoddy advertisements.

(Data source: Ad screenshot from one year ago)

(Data source: Ad screenshots one year later)

Today, TikTok has begun to move towards commercialization, and a large number of advertisements have appeared in the product. Just when I was writing this article, the advertisements were upgraded again, and information can be filled in directly on the current playback page. A lot of thought has also been put into the optimization and modification of the advertisements.

From the above advertising screenshots, we can see a very obvious gap: during the advertising test period, the main merchants were large companies, and the quality of the advertising content was also very high, which was in line with the tone of the vertical music short video platform at the time; after the diversification and the opening of advertising , the quality of these advertising videos fell to the bottom - there is really no harm without comparison.

The poor quality of advertising content is also related to the size of the company. There is not enough money to invest in a high-quality advertisement, but the impact of low-level advertisements is also great. It is completely inconsistent with the tone of the product. I believe that such advertisements will not have good data performance after being released, and it also has a certain impact on users.

People in modern society have no way to escape from advertising, which has invaded every corner of society.

There are ads on TV, online, in newspapers and magazines, on car bodies, on buses, and outdoors. Even on TikTok, there are ads now. As audiences, we find good ads pleasing to the eye, but bad ads make us feel like our eyes are being raped.

The current advertising methods for Douyin are: splash screen, CPM, CPC, and CPA. I won’t talk about splash screen here. Generally, companies that can place splash screen ads will not place a spam ad and be criticized - of course, they can do so, which will cause negative discussion and spread, and the cost is also extraordinary. It will be bad if they fail.

• CPM: Cost per thousand impressions, which is the cost required for an ad to be displayed 1,000 times.
• CPA: telephone consultation, that is, the cost of a customer consultation call; form submission, that is, the cost required for a user to submit personal information once; single download cost, that is, the cost required for an APP to be downloaded once, limited to Android APP.

Generally, the most commonly used advertising methods are CPM and CPC, which adopt a bidding model: it can be simply understood that the higher the bid, the higher the ad position , and the more high-quality advertising resources you get.

(Optimization ideas for Tik Tok advertising)

Through the above bad ideas, I believe many people have understood what I mean - the main purpose is to allow high-quality ads to get more exposure. For low-quality ads, we must consume the limit as soon as possible and reduce the frequency of appearing in front of users.

Of course, advertisers will be informed when adjustments are made, and those who operate advertising will keep an eye on the background data.

Let's explain this through two user scenarios:

Operations student A: Use CPM to place ads, with a price of 15.8 yuan per thousand impressions and a bidding price of 22 yuan per thousand impressions, ranking in the 4th advertising position. During the advertising delivery, the system automatically distributed the advertisement to 10,000 precise users for display. The video content was well-liked by users, and the advertising video playback volume, bounce rate, completion rate, likes, reposts, comments, and interactive data all performed well. After data monitoring, the system determined that the advertising content was of high quality and reduced the thousand exposures to 20 yuan. Other weights remained unchanged and it was still ranked in the 4th advertising position. The system automatically distributes it to 20,000 precise users for display, and the data performance is still good, reducing the price per thousand impressions to 18 yuan and rising to the third advertising position. And so on. The more popular the advertising content is with users, the higher the chance of being displayed.

Operations student B: Use CPC to place ads, with a single click price of 1.5 yuan and a bidding single click price of 5.6 yuan, ranking in the first advertising position. During the advertising delivery, the system automatically distributed the advertisement to 10,000 precise users for display. The video content was not liked by users. The playback volume, bounce rate, completion rate, likes, reposts, comments, and interaction data of the advertising video were all poor. After data monitoring, the system determined that the advertising content was of poor quality and increased the single-click bid to 6.8 yuan. Other weights remained unchanged and it was still ranked as the first advertising position. The system automatically distributes the ads to 20,000 precise users for display. The data performance is still good. The single-click bid is increased to 7.8 yuan and dropped to the second ad position. And so on. The less the ad content is liked by users, the fewer opportunities it has to be displayed, and the ad quota is quickly consumed.

Through the above user scenarios, everyone should know what the process is. There are some algorithms and weight calculations that I will not delve into. In fact, the principle is very simple. Since Tik Tok uses Toutiao’s algorithm to recommend videos, it can also be used in advertising, and the algorithm can be flexibly combined to optimize the advertising effect.

As for whether the effect is good or bad, let me give you my superficial opinion. We cannot make rigid requirements on the quality of advertising content. If we adopt rigid requirements and restrict video images and content, the amount of advertising will decrease over time and advertisers will not place ads.

If this method is used to improve the quality of advertising content in disguise, I believe advertisers will be very happy, because these ads are displayed to users and are judged entirely based on user feedback. Not only can they get more display opportunities, but the cost of advertising investment will also be reduced.

On the contrary, ads that users don’t like will also have very poor effects. Reducing the appearance of such ads is also a good thing for the product itself, thereby indirectly encouraging advertisers to improve the quality of advertising content. It doesn’t necessarily have to be high-end, as long as users like it, it is good, so it will not increase the cost of advertising production too much. Compared with the previous user review, this bad idea is more practical. Everyone thinks about how to improve the quality of advertising content and maintain user experience.

Summarize

At present, Douyin still has an overwhelming advantage over other competitors.

Douyin has also launched its overseas strategy, moving from vertical to diversified, with its user population distributed downwards. With people in offline cities becoming the main growth force, Douyin can maintain a good growth trend in the short term.

With the addition of Tencent Weishi, the short video field is bound to usher in another round of chaos, with Kuaishou, Douyin and Tencent Weishi competing for supremacy.

The future of Douyin is not easy: although Kuaishou is not a direct descendant of Tencent, it has a reliable godfather behind it; while Weishi is a direct descendant of Tencent, and the rich second generation is not short of money. However, after using Douyin, I can often see familiar video content on Weishi and Kuaishou, and the frequency of repetition is very high. I wonder if the battle between Weishi and Sina Weibo this time will be like the battle between Tencent Weibo and Sina Weibo back then.

As for what is the final one? It may take time to witness.

I am not an analyst and I cannot predict any results, but as of now, TikTok is still in the leading position.

The only concern is that after diversification, the video content will become worse than before and lose the previous fun. In addition, how to maintain the output of content is also a very critical issue: if the content seen on each platform is the same, it doesn’t matter which one is used. In addition, Weishi has added shooting templates, which has reduced the difficulty of shooting a lot.

I feel that TikTok will also add such a function. I'd like to bet that within half a year, TikTok will definitely make a major overhaul of the filming process to reduce the difficulty of filming. Ten packs of Weilong spicy strips .

Let’s write it here this time. One year later, we will continue. This bet is valid!

Source: Occupation, Xiaobai