5,000 words to reveal the dilemma of live streaming traffic!

I have done a point-to-point breakdown of the operating cycles of a large number of accounts. I believe that after reading this issue, friends will have a deeper understanding of traffic.

01. Why is there so little traffic when a new account is launched?

It is not terrible that a new account has very little traffic when it starts broadcasting. In my live broadcast room, there is also a phenomenon that there is no traffic at all when I broadcast the first show.

As a new live broadcast room, without any labels and weights, the algorithm cannot identify what kind of users you need.

In order not to waste traffic, the most common traffic allocation number ranges from 30 to 500. As mentioned above, it is normal to have no traffic at all.

In addition to the live broadcast room, the same is true for account short video traffic. We will find a large number of new accounts that release works in the early stage. Even if they are of very high quality, the number of views is still pitifully small.

What you need to do is not to panic, but to correctly understand the rules of account traffic distribution and keep broadcasting every day.

At the same time, you should know that if you want to further obtain more traffic recommendations, it is useless to just start broadcasting. You need to create the data needed by the algorithm in the live broadcast room to feed the algorithm.

Generally speaking, 3-5 games are enough to get relatively stable traffic recommendations.

If you don’t know how to get traffic through the feeding algorithm, please refer to the previous article "300,000-word article on Douyin live streaming and selling goods, series 2": Analysis of the latest algorithm of Douyin live streaming" for details.

Of course, there are extreme cases where no matter how the show is played, there will be no natural recommendations.

In this case, the first thing to do is to check the problems with people, goods and places, whether the host's rhetoric is rubbish, and the cost-effectiveness of the traffic-generating products. If no one comes in even after they are delivered,

Then you can throw away this account after a week. This is not metaphysics. I can prove it with my experience of having so many live broadcast accounts.

02. Traffic suddenly dropped, and I don’t know why?

Don’t ask why the traffic has decreased first, but first find out what traffic has decreased.

Any live broadcast room has multiple traffic entrances, and the method of analyzing the problematic traffic is very simple.

By comparing the traffic share of the past two weeks, we can analyze which traffic entrance has problems.

If the decline is due to natural recommendations, first distinguish the account cycle. If it is the account cold start phase, it is mostly due to a decline in interactive data.

If the account itself has reached a very high level of market visibility and then drops, it is most likely that the conversion rate, transaction density and market visibility level do not match.

Of course, the factors leading to the decline are far more than just that. The above explanation is only to make a point clear: the decline in traffic from different channels must be analyzed on a case-by-case basis.

If the proportion of videos decreases, then it is directly a problem of video traffic, and it is necessary to check whether the video format has been changed recently.

Whether the product displayed in the video has selling advantages is also affected by factors such as release time and frequency.

If the proportion of Qianchuan decreases, first determine whether the Qianchuan traffic has really decreased. Sometimes it may appear that the Qianchuan traffic has not decreased, but the proportion of other traffic has increased.

If there is indeed a decline, you can directly check the data of Qianchuan's backend and analyze its production, click rate, and thousand times.

Of course, in addition to the above factors, we also need to judge the recent live broadcast room product grouping and product arrangement.

Check whether it is caused by changes in products and the health of the account. For example, if an account violation leads to credit deductions, it will also affect the push of traffic.

03. Will interrupting broadcasting or changing time periods have any impact on traffic?

First, let’s talk about interruptions. Different live broadcast modes have different degrees of impact on account interruptions, and accounts that rely on natural recommendations are most affected.

Among the accounts that have been operated in the past, there has been a phenomenon where the natural recommendations dropped by 50% after one day of suspension of broadcasting, and this happened more than once.

The possible reason for the impact of the interruption is that live broadcast accounts of the same category and at the same stage are always competing according to the algorithm's horse racing mechanism.

When an account stops broadcasting, its peers will share the day's traffic. When it resumes broadcasting, the account's traffic level will change. This is not absolute, but it is definitely one of the factors.

The second is the accounts that rely on Qianchuan traffic. The biggest impact of the suspension is that I think the impact of the plan is far greater than the account itself.

Every time the plan is launched, the traffic pool is re-drawn. For a plan that increases volume, the longer the live broadcast time, the better while the volume continues to increase.

If the broadcast is suddenly interrupted, there is a high possibility that the volume will drop completely.

On the contrary, the impact of flat broadcasting is the smallest. In the past, several flat broadcasting rooms had to stop broadcasting for one day due to problems with the allocation of anchors. There was not much impact if they continued the live broadcast on the third day.

Changing the time period will have an impact on any live broadcast mode.

On the one hand, changing the time period is out of the viewing time of old fans. Especially for live broadcast rooms that rely on fan transactions, the time period should be preheated in advance and should be avoided if possible.

On the other hand, it disrupts the algorithm's allocation of traffic. Of course, this does not mean that the algorithm has made regulations on traffic allocation for each account, but that changing the time period will easily cause the algorithm to re-sort the traffic, thereby causing traffic fluctuations.

04. The proportion of old fans is too high. Are they suppressing new fans?

Too high a proportion of old fans will not suppress new fans, but the conversion rate of new fans is too poor, causing the algorithm to determine that the live broadcast room cannot take on new traffic, thereby reducing the recommendation of natural traffic.

So from the results, the number of new fans has decreased, and the number of old fans has naturally increased.

If this problem arises, there are several proven suggestions that can be adopted.

First, promote more past hot-selling products in the live broadcast room. Since old fans have already purchased them, the order rate will naturally decrease. Since they are proven hot-selling products, the transaction rate of new fans will naturally increase.

Second, the live broadcast room is set up exclusively for new fans, with clear provisions that only new fans can participate, and it is coordinated with them in product organization and event planning.

Third, guiding users to stay, like, follow and other interactive indicators can also achieve the effect of stimulating new fans.

05. If the proportion of paid traffic is too high, will it suppress natural traffic?

The same logic applies to the above question. Payment will not suppress natural traffic. The different conversion results of paid traffic have both advantages and disadvantages for natural traffic.

If the interaction data and conversion transactions are good after the paid traffic comes in, it will not only not suppress the natural traffic, but will stimulate the natural traffic.

This is also the reason why a new account will use the "sui xin tui" method to bring in initial traffic when it starts broadcasting, and then obtain recommended traffic through the initial traffic.

Another phenomenon is that if the campaign attracts paid traffic, but the live broadcast room does not do enough to carry out conversions, the recommendation ratio of natural traffic will be affected.

This is also why I have always recommended that if you are not sure you can afford the payment, invest as little as possible in payment.

06. For Pingbo accounts, why is the proportion of natural recommendations low?

If you analyze a large number of live broadcast rooms, you will definitely find a pattern in the proportion of traffic. Basically, video traffic and fan traffic dominate the market. This is a very normal phenomenon, especially when short videos become popular.

The different modes of account creation will largely determine the traffic share.

For example, in the live broadcast room where the account is created through activities, natural recommendations are the mainstream. In the live broadcast room where the account is created through Qianchuan, Qianchuan is the mainstream. The same is true for video accounts.

The logic behind this is that due to the excessive live broadcast format, the flat broadcast room lacks behaviors like event registration to arouse interactive indicators, so it is unable to obtain a large amount of natural referral traffic.

Of course, there is one exception, that is, when a large number of hot products are sold in the live broadcast room, it will quickly stimulate large-scale traffic to flow into the live broadcast room.

From a safety perspective, I suggest that when you create a video account, you should just focus on generating video traffic.

If the operation is not appropriate, the live broadcast room that was originally able to sell goods every day relying on accurate video traffic can accidentally become a gathering place for freeloaders.

If you really want to solve the problem of recommended traffic, you must follow the logic of event creation.

In the original live broadcast format, appropriate adjustments should be made to the product and live broadcast language, and attention should be paid to being appropriate.

For specific operation details, please refer to the previous tweets "Creating monthly GMV exceeding 10 million, the first comprehensive dismantling of progressive gameplay"

07. The recommended traffic suddenly becomes 0. What is the reason?

This is a problem that occurs very rarely, but it does exist and a team has experienced this phenomenon before.

I have three accounts, and their daily viewership is over 50,000. Suddenly, on the same day, the number of live broadcast recommendations for the three accounts dropped to 0.

To address this phenomenon, first check recent violation records to see if the account has been restricted due to multiple violations.

Otherwise, continue broadcasting and check for a week to see if the recommended traffic is restored. Do not make any changes during this period.

Secondly, conduct self-examination and self-correction of the account to see if the same anchor is broadcasting multiple accounts at the same time, or if there are a large number of false orders on the network used.

Of course, this does not mean that sharing a network will affect account streaming, but that if no conclusion can be drawn, as much investigation as possible should be done.

If no changes have been made after completing the above steps, it is recommended that you directly use low-price traffic diversion to stimulate new users to stay and interact by placing a large number of orders. If there is no effect over a long period of time, it is recommended that you shelve the account and start over.

What I want to say here is that not all phenomena can be concluded. The algorithm of TikTok is manually set and automatically learned, and many factors do not completely follow the established rules.

08. Why does Qianchuan Promotion have no traffic?

If there is no consumption or low consumption in paid delivery, there will naturally be no traffic. Some people even raise prices and increase volume without any consumption, so they suspect that there is something wrong with the account. This shows that we do not understand Qianchuan's material review mechanism enough.

As a content advertising platform, Qianchuan needs to objectively review the quality of the materials delivered. For example, if the materials are not competitive, it will be decided from the very beginning that no more traffic will be given.

The idea that you can gain traffic by raising prices and increasing sales is even more wrong.

If everyone can get traffic by paying more money, then the result will be the low quality of paid advertising, which is absolutely not allowed to happen in Qianchuan Ecology.

When Qianchuan has no consumption volume, the first thing to check is the price increase. If the price increase cannot be consumed, the targeting is checked. If there is no problem with the targeting, it is basically a problem with the materials.

In daily promotion, at least 50% of the content is not consumed due to material reasons.

So when there is no quantity and you have tried all kinds of measures, there is no doubt that it is a problem with the materials.

Please note that there is also an extreme situation, that is, there is indeed a phenomenon that the volume increases quickly by changing the account number.

This is a very rare situation. If you want to avoid this kind of phenomenon from the beginning, you can use the matrix mode to solve it.

09. What’s the reason why Qianchuan suddenly has no volume?

Plans with different cycles have different reasons when faced with a sudden drop in volume.

If it is a cultivation plan, it means that the plan itself is unstable and sudden drops in volume are common. The solution is not to change the plan and observe for two days to see whether the plan will recover.

If there is no volume, then raise the price and observe. If there is no volume, then suggest relaxing the targeting. If it doesn’t work, then close the plan.

If it is a mature plan, it is not recommended to modify the plan too much, the most you can do is increase the price.

If the price increase cannot solve the problem, simply put the plan aside for one to three days to see if the plan can be restored. If not, make adjustments according to the above-mentioned incubation period plan.

Of course, mature plans are very valuable. In addition to making modifications, it is recommended to copy more mature plans with different orientations, creative classifications, and bids, and throw the plans into different traffic pools for further testing.

10. What should I do if the traffic of my account decreases after my account has a hit product?

What this means is that a hot product appears in the live broadcast room, and after the hot product period is over, the traffic continues to decline, and is even lower than the traffic before the hot product appeared.

These are two aspects of the problem. First, according to the normal weight mechanism, if an account has experienced a hot product, the account labeling will be more accurate.

Of course, the premise is that the styles you sell are consistent and the prices are within the regular price range.

However, if the product with explosive sales is a low-priced model, then after a large number of transaction tags are added, the subsequent user model push flow will change, resulting in a decrease in UV value.

Therefore, this also reminds operators that when planning hot products, they must take into account the overall product arrangement, price, and user audience of the live broadcast room.

Another aspect is that in the live broadcast rooms where hot products appear, the algorithm will be more strict in evaluating them. That is, future data conversion and transactions will be based on the data during the period when the products became hot.

Then in the future, if the quality of product placement cannot keep up, it will inevitably affect the conversion rate and the recommended traffic will decrease.

So I said something before, that is, when a hot product appears in your live broadcast room, the most important thing you should do is to quickly cultivate the next hot product.

Similarly, we can also see many mature live broadcast rooms, often after a hot product appears.

When the sales of hot products enter the middle and late stages, we will take advantage of the popularity of the scene to quickly test new products.

The purpose of doing this is to ensure that new hot products can catch up after the old hot products fall in price.

11. There is a lot of traffic at the beginning, how to best handle it?

It is impossible to undertake extremely fast traffic, we can only undertake it to the maximum extent possible, and these are the nodes where extremely fast traffic generally appears.

The first is the fifteen minutes before the opening, and the second is the online pull-up during the opening process.

As for how to handle the extremely fast traffic at the beginning, the first thing to understand is that in a live broadcast, the live broadcast data is calculated in real time, and the opening data in the first half hour is extremely important, which determines the flow rate of the traffic in the middle and late half.

We can often capture a phenomenon in daily life, that is, when the overall data in the first half hour is good, the subsequent live broadcast process will be very smooth. On the contrary, the flow rate in the middle and second half is slow, and there is a lack of online pull-up.

The second is the extremely fast traffic at the opening, most of which comes from the recommended traffic from the live broadcast square. Users have not been educated by front-end videos and previous live broadcasts, and it is easy for them to churn away. This is the characteristic of extremely fast traffic.

This determines that when taking over high-speed traffic, the focus will be on two data.

The first is the interactive data mainly based on the user's stay, which means that you can effectively take on the user's stay. The better the interactive data of the live broadcast room, the higher the online push will be.

Because the live broadcast room itself is a leaky barrel, the better the retention is, the smaller the leak will be, and the online level will continue to rise like water.

The second is the user conversion rate, that is, whether we can complete the conversion of corresponding indicators when high-speed traffic comes in.

After understanding the above logic, you can then work out a method for handling extreme fast traffic, that is, when extreme fast traffic comes in at the beginning, don’t just use the full-price model.

Instead, they use welfare funds to attract users to stay, and use the host’s marketing language to create user interaction, continuously raising the online level until they finally use welfare funds to obtain a high transaction conversion rate.

Most mature live broadcast rooms will choose to use benefits to pamper fans at the beginning, which also indirectly reflects the method and importance of opening traffic.

12. The convergence effect of traffic and its significance to the growth of live broadcast rooms

This is the last question, and also the initiative first proposed in June. For a high-quality live broadcast room, I suggest adopting the three-horse model.

Looking at all the ways of playing in the live broadcast room, we can roughly distinguish several major mainstream modes, including event start-up mode, short video mode, Qianchuan delivery mode, etc.

We can clearly see that from the perspective of traffic entrances, activity creation mainly relies on live broadcast recommendations, short video testing mainly relies on video traffic, and Qianchuan mainly relies on paid traffic.

Most live broadcast rooms can be established by relying on a single traffic source, but this also has hidden drawbacks. The spiral decline of data indicators and account violations can easily affect the traffic scale of the account.

Once short video models face the homogeneity of formats, the reduction of video push will lead to a decrease in video recommendations. Qianchuan itself is a buying volume model, and the corresponding market traffic is becoming more and more expensive.

Since June, I have been advocating the "three-horse carriage" model, namely natural recommendation + short video + Qianchuan traffic. This is because after working on the account for a long time, I can clearly feel the disadvantages brought by single traffic.

The three-horse model avoids the above problems better. I will review and analyze the specific gameplay with my own cases later.

The above is a recent summary of traffic-related issues, but don’t think this is a standard answer.

Since its development, the Douyin platform has not given any official standards for algorithms, traffic, and weights. Most of the information we can obtain comes either from bystanders or from practical summaries.

Just like the traffic analysis I am writing now, it is a practical summary for me, but it is just hearsay for the viewers. Can you ensure that it is correct?

But the output of any point of view is not to prove the correctness of the argument, but the collision of multiple thoughts.

At the intersection of different viewpoints, readers can think deeply and draw their own conclusions. This is the meaning of the discussion and the purpose of the content of this article.

Author: Yin Chen's live streaming

Source: Yin Chen's live streaming