Traffic Analysis of the BIG 4 live broadcast platforms in China: Douyin, Kuaishou, Taobao and Tencent
Live broadcast e-commerce is getting hotter and hotter. Even after reading the live broadcast e-commerce report of China Merchants Securities, Luo Yonghao announced a high-profile entry into live broadcast e-commerce, and soon signed a contract with Douyin.
Although Luo Yonghao hasn’t made a mobile phone in recent years, he has made himself a “super influencer” who does not have a topic. Here I will not judge whether Luo Yonghao will be successful in live broadcast e-commerce, and whether Li Jiaqi, who surpasses Taobao live broadcast, and Simba, a fast-hand live broadcast e-commerce, will become the “first brother” of Douyin live e-commerce.
In this article, the flow logic of the four live broadcast e-commerce platforms, Douyin, Kuaishou, Taobao Live and Tencent Live, is thoroughly researched and sorted out for merchants to use as a reference when choosing a live broadcast platform and actual selling.
Douyin’s heavy algorithm light fan’s flow logic comes from the success of today’s Toutiao. As an information recommendation model that is different from search and social, matching content with users, accurate recommendation through the system is the core of this algorithm. So some people call this logic: content-oriented planned economy.
There is a simple culvert formula behind the vibrato and headline recommendation algorithm: y = F (Xi, Xu, Xc)
This function includes three dimensions of variables, namely user, environment, and content.
The first dimension: content. Each kind of content has many tags, what category, what field, broadcast volume, number of comments, number of reposts, etc., need to consider how to extract the content characteristics to recommend.
The second dimension: user characteristics. Including interest, occupation, age, gender, etc.
The third dimension: environmental characteristics. Where is the user, on what occasion, work or travel, or in the subway.
Simply put: who am I, where am I, and what do I want to see.
To match these three is a very complicated mathematical problem, there are several commonly used models. Such as Douyin, which has a large amount of data and strong real-time performance, is generally mixed with multiple models. In the end, the system will calculate the index of a video based on multiple factors, and then recommend it step by step according to the index.
The first step is a cold start.
After the video passes the review, the system will allocate an initial flow pool, which consists of two parts.
1. Fans of the account, but not all fans can push, and must obey the principle of algorithm priority.
2. Users who may like the video.
It is recommended to have about 300 playbacks for cold start.
The system will weight the video based on the data. There are 4 core data: play rate, comment rate, like rate, and finish rate.
Then do weighted calculation:
The order of weights is probably the end rate> like rate> comment rate> forward rate.
The reason is very simple. Your video may attract users at the beginning, the title may attract users, or the cover may attract users, but these cannot prove that the quality of your entire video is high, only that a certain part is attractive.
If the user can watch your video, it means that your video is really high-quality, so it is not surprising that the right to broadcast rate is replayed in the first place.
In addition to these four data, the weight of the account is also a consideration.
According to the algorithm experience of Toutiao today, if two accounts send the same message (the text can be crawled for analysis), the algorithm will give priority to the account with the highest weight. But the video should be more difficult to encounter this situation.
The second step, after weighted calculation, meets the requirements of the second recommendation, the video will be recommended to the second flow pool, about 3000 .
Then repeat the second step. Statistics, and then recommend, each recommendation will get more traffic. If the data does not meet the standard, it will be temporarily recommended. The video traffic will stop.
Finally, an inverted triangle recommendation mechanism was formed.
The above is the flow logic of Douyin’s short video, so most of the live e-commerce will continue this traffic recommendation algorithm, but the live e-commerce will also involve conversion rate, repurchase rate and other e-commerce parameters, which will make Douyin face New traffic distribution challenges.
Quick hand flow logic: social + interest
Kuaishou recommends content based on social + interest, using a decentralized “market economy.” The platform focuses on waterfall-style dual-column display, and the fan arrival rate of published content is about 30% -40%.
Kuaishou prioritizes the regulation of traffic distribution based on the user’s social attention and interests, focusing on the recommended content of the “Focus Page”
The quick operation of weak operation control directly “links” content creators and fans, deepens the stickiness of both sides, precipitates private domain traffic, and creates a “trust relationship” with high trust.
Let’s take a look at the rules and paths recommended by Kuaishou for “mutual fans”. The platform limits the daily attention limit to 20 people, and it can no longer be added when the number of concerns reaches the upper limit of 1500. Of course, the purpose of this platform design is not to make people love each other.
There are several types of recommendation mechanisms:
- Recommended based on the people you follow
- N friends follow together
- someone you may know
- He is following you
Fans who get through mutual fans generally pay more attention to “mutual fans”, he may do fan management: often check whether the people they follow are also paying attention to themselves, if the other party no longer pays attention to themselves, then get rid of.
The following model diagram drawn by the product manager can probably demonstrate the fast hand flow logic of changing from “stranger social” to “fan old iron social”, from “public domain traffic” to “private domain traffic”, and the poster’s “ The more obvious and powerful the “style” / “personal setting”, the closer the private domain traffic will be.
According to QuestMobile statistics, the retention rate of fast-acting active users on the 7th reached 84.4%, ranking first in short video apps, and the retention rate was second only to WeChat.
The flow logic of Taobao live broadcast: the “experience + professional” graded operation of the anchor
Taobao live broadcast has gradually moved from content transition to the “experience + professional” grading operation of the anchor. The dimensions involved in the experience include: live broadcast sessions + duration, platform activity completion rate, and fan retention rate.
The dimensions involved in the profession include: single live broadcast column settings, effective baby delivery, monthly live broadcast orders, in-store conversion rate, order returns and poor after-sales service capabilities.
The anchor is divided into three major levels:
- TOP anchors: MCN institutions, artists, big coffee, KOL
- Waist anchor: high conversion, high ability, high face value
- New anchors: fans, online duration, live broadcast cover
In March 2019, Taobao Live launched a system called “Anchor Growth”, through which the anchor can understand where his rank is.
Taobao ’s anchor level reflects the influence of the anchor. The anchor needs to accumulate experience and professional points if he wants to upgrade his level.
1. How to get experience value and professional points:
Basic experience value task: You can get 1 experience point for every 1 minute of broadcasting, up to 200 points per day, and the excess will no longer be accumulated.
Additional experience tasks: After the viewers generate interactive activities such as likes, comments, concerns, and sharing, the platform grants additional experience value rewards, which are settled on a daily basis. The maximum reward is 100 points per day, and the excess will not be accumulated.
Basic professional tasks: Every time you add 1 product to the live room, you can get 2 professional points, which are settled on a daily basis. You can get up to 200 points per day. The excess will not be accumulated. Repeated addition of the same product will not result in additional points.
Additional professional tasks: After the audience in the live broadcast room enters the store through the product list or produces purchases, the platform gives additional professional points rewards, which are settled on a daily basis. The maximum reward is 100 points per day, and the excess will not be accumulated.
2. Matters needing attention:
The experience value and professional score value will be brought to the next level. The accumulated experience value of the anchors of Taobao live broadcast is only valid for the anchor, and the professional score is only valid for the professional category of the anchor. The experience value and professional score data of level 4 and above anchors will be stored in the bottom table, but the front desk only displays the current month value for the monthly top anchor sorting.
In addition to the hierarchical operation of the anchor, the Taobao platform also has a set of rules for the distribution of traffic, which mainly has the following three evaluation principles:
1. Label Competition
The tagging of live broadcasts is actually to accurately locate your live broadcast attributes for officials and fans, and match the corresponding traffic according to your attributes. But if there are more people using tags, there are more options.
Under the label, compete with competitors for traffic.
2. Level climbing
This is beyond doubt. The higher the climb, the more the live broadcast rights, the greater the chance of being seen by officials and fans, and the natural traffic will also tilt towards the high-level anchor or store.
3. Activity ranking
The large and small events organized by the Amoy Department, live broadcasts of various themes and end-of-month qualifying are all a shuffling process.
The better the official activities and official tasks are completed, the higher the ranking, which proves that you have the strength, and will not waste the traffic that the official hard to “buy”. “preference”.
In the process of traffic competition, the rational use of live broadcast tags, climbing live broadcast ratings, and seizing activity opportunities, ranking on the list has become a few core actions.
The “distribution according to work” in the live broadcast industry is always reserved for a small number of “risk” people.
Of course, in the Taobao department, the judgment point of the inclination of traffic will also focus on content construction. Therefore, doing a good job in content construction is the core point of increasing traffic.
How to make the content series of the live broadcast system can be judged from these five parts:
1. Content visibility
That is, the breadth of content that can cover consumers is mainly through the people who appear in the live broadcast room with weights and Weitao. The wider the audience covered, the greater the chance that the content can be seen.
Mainly investigate the operational capabilities of live broadcasting.
2. Content Attraction
Taking unit time into consideration, whether fans can stay, buy, and interact with each other (comments, likes, sharing, etc.) as a consideration depends on the live broadcast atmosphere, product selection, and anchor guidance.
The main investigation is the product composition and anchor attractiveness.
3. Content guidance
It is closely related to the attractiveness of the content, from keeping the fans to guiding them into the store and actively understanding the ability of the product. This part can be improved by the anchor’s speech construction.
It mainly investigates the construction of the speech system, the control field of the anchor and the attraction.
4. Content gains customer power
Represents the ability of content and consumer purchase behavior to lead conversion, that is, the purchase behavior after understanding the product, from the previous grass planting to the success of pulling the grass, through the content to obtain the precise consumer group of the purchased goods.
5. Content power
That is, through continuous content output, visitors who only stay briefly will become purposeful, long-term “hardcore” fans.
Taobao is currently the most mature platform for live broadcast e-commerce, which is mainly divided into popular people bringing goods + merchant self-broadcasting. 90% of live broadcasts and 70% of the transaction volume come from merchant self-broadcasting.
The conversion rate of Taobao live streaming into the store exceeds 60%, but the return rate is higher. The monthly activity of Taobao APP is 650 million, and the monthly activity of Taobao live app is 75 million. The user base is huge, but the social attributes of the application are low.
In 2020, Taobao will focus on live broadcast stores, supplemented by private operation of traffic operations, refined incubator incubation, and hierarchical organization operation, and will continue to work hard to bring goods live.
The flow logic of Tencent live broadcast: instrumentalized social fission
In December 2019, Tencent Live is fully open for public beta, which is divided into watch live + applet live.
Tencent mainly takes the form of a “watching live broadcast” tool, through the existing personal WeChat, circle of friends, public account, WeChat group, corporate WeChat and placing Tencent advertisements (Guangdiantong), in a “decentralized” way. Anchor gets platform traffic by itself.
In 2020, the layout of WeChat mini-programs focuses on building business scenarios, and the latest official mini-program live broadcast component “Watch Live” is launched to help merchants create their own business closed loop.
WeChat adopts the S2B2C model. The high stickiness of platform users and the high trust of private domain traffic can bring high conversion and high repurchase of e-commerce. The future performance in the e-commerce live broadcast market is worth looking forward to.
🇨🇳This original English article was published by ChinaPotion. Please contact us for any deeper insight & data; or to report unique tech stories in China.
ChinaPotion.com | Breaking Tech Stories from China
