The latest algorithms and gameplay for live streaming sales in 2022

No matter what the problem is, it can be solved using algorithms.

A comprehensive analysis of the live broadcast algorithm

In daily live broadcasts, we often encounter many live broadcast problems, such as live broadcast room streaming, peak, flow rate, live broadcast room popularity and transfer issues.

No matter what the problem is, it can be solved using algorithms.

Algorithm is the underlying architecture of all Internet products. Whether it is Douyin or Kuaishou, the underlying operation is called algorithm.

No one can understand an algorithm 100%.

The algorithm, as the operating architecture of the platform, comes from human settings, but after human settings, the operation of all algorithms will go through a machine learning process.

Second, the platform rarely talks about algorithms. Why not?

Algorithm is the core of a platform. As a platform, telling you the most basic logic of the algorithm is tantamount to ignoring the interests and risks of the platform and telling you nakedly how to make money on the platform, which is impossible.

Algorithms can be guessed but will never be made public.

Most of the algorithms we come into contact with in our daily lives are derived from a posteriori.

What is a posteriori?

The understanding of algorithms that we can gain comes more from certain practical operations, including the accumulation of data in the live broadcast room and the discovery of existing patterns.

Algorithm architectures do not change easily, but there is evolution.

The source of any algorithm evolution is not that the platform can change it as it pleases, nor is it that the merchants have any needs and tell the platform whether they can operate in a certain way.

The underlying basis of algorithm evolution is the development of the platform's commercialization path.

In the early days of live streaming, you’ll find that as soon as you start broadcasting, there will be traffic.

At that time, the algorithm assessment was that I would give traffic as long as the broadcast started, but later I found that after the broadcast started, you had to stay, and after staying, you had to convert, and you had to have UV and GPM.

As a profitable product, the platform needs to use algorithms to drive users from an ordinary content user to a transaction user.

The live broadcast room is the operating form presented by the algorithm.

A good live broadcast room must have a relatively large audience, and the users who come in to watch the show must also have high-quality user tags.

When your live broadcast room can ensure the scale of the audience and the accuracy of the users, then the live broadcast room will definitely make money. If you don’t make money, there will be only three problems: people, your goods, or your venue is not good.

With the help of the evolution of algorithms, how do we understand the evolution of a live broadcast room?

The evolution of a live broadcast room from 0 to 1 is actually the evolution of two elements.

The first is the evolution of the field perspective, whether your field perspective will become larger and larger, and the second is the evolution of labels, whether your user quality will become more and more accurate. You put these two things together, which is called weight.

Weight is for us live broadcast merchants. Weight can obtain more scene views and user accuracy. For the platform, I can make the live broadcast room bring me more value.

This is an evolution of weight for us and an evolution of monetization value for the platform.

The evolution of the live broadcast room: looking at the gameplay from a dialectical perspective

There is no one way of playing that is always effective, because the gameplay is also evolving, but the long-term effectiveness of different gameplays is different.

Based on long-term classification, you will find that all gameplay can be divided into two types. The first type will maintain continuous optimization to allow the live broadcast room to survive as much as possible.

The second possibility is that he only appeared for a very short time, for example, when everyone saw some violent gameplay in a live broadcast room.

To summarize a common feature, as long as it is a short, flat and fast way of playing, what it pursues must be quantitative evolution.

Regardless of the 13-line gameplay we saw some time ago, or the other gameplays that will be promoted in the future, it is a quantitative change. It is difficult to complete and satisfy both user tags and the live broadcast room audience.

The second point is to only pursue a single indicator.

What is a single indicator? We often talk about retention, conversion rate, GPM, UV and so on.

In an ideal live broadcast room, all indicators develop in a balanced manner according to the rules of the platform. But 99% of live broadcast rooms cannot achieve the ideal model, including myself. We can only get close to this level.

If your live broadcast room only pursues a single factor to become popular, the audience of your live broadcast room will also become popular.

However, traffic is limited. Although short-term gameplay can make the scene very large, the traffic of the entire platform is very limited from the perspective of the platform.

Since you are given so much traffic, you must pursue the corresponding commercial value.

This is called maximizing traffic value.

The platform gives you 100,000 views and also gives the same to another live broadcast room. You earn 50,000 yuan, while others earn 500,000 yuan.

Your data is very bad. The platform thinks you may have a good start in the next show, but if you have poor performances for several shows in a row, do you think the platform will still give you traffic? It won't be given to you.

This is called the horse racing mechanism.

For live broadcast rooms of the same category and level, I gave you the same amount of traffic, but you had no way to achieve the same unit of monetization granularity. If you were out of balance, you would be eliminated from the horse race.

This is how we look at all the short, flat and fast gameplay we see now from a strategic and high level perspective.

Regarding traffic, you have to enjoy the delayed gratification of traffic.

What is the delayed gratification of traffic?

When the platform gives you a very large flow, it will not only assess your basic indicators, but also your deep indicators.

The crowd, goods and venue in your live broadcast room, including your conversion capabilities, are not up to standard. If you cannot meet the platform's deep-level indicators, the traffic in the live broadcast room will decrease.

This is why most excellent live broadcast rooms choose to use initial weights to gain volume at the beginning, then test the overall conversion, and then expand the traffic.

We have summarized the rules as: obtain traffic, take over traffic, and continue to obtain traffic. We also conclude that we should not lose sight of quality but pursue quantitative development.

If you look at all the ways we play, this is basically the way we do it.

Fairness is a product of algorithms.

The competition in the entire TikTok industry is very fierce, but from the perspective of ordinary people, TikTok's algorithm is very fair. Without this algorithm, it would be impossible for ordinary people to make a comeback through live streaming.

So the essence of the algorithm is fairness.

The algorithm sounds complicated, but it is not certain whether all live broadcasters have the ability to analyze the algorithm. It depends on your ambition for live broadcasting. A couple can have successful cases even if they don’t understand any algorithms, but they lack the ability to replicate in live broadcast rooms.

It's like doing math problems. You may get one problem right by chance, but if you are not familiar with the solution, you are not sure you will get it right the second time, even if the answer is the same.

How the algorithm guides the behavior of the live broadcast room involves the breakdown of indicators in the live broadcast room algorithm system.

We often hear the following sentence: What kind of indicators should we set in the live broadcast room?

There are three kinds of illusions we commonly face.

The first one is that you don’t know what indicators exist.

The second type is that among so many indicators, we don’t know what the relationship is between them.

The third type is that you don’t know which indicators are important and which ones to focus on.

First of all, we need to understand the importance of an indicator.

Why is there no traffic in the live broadcast room? Why is the traffic very high at the beginning but then drops? Why can't you transfer the money? And so on.

Take one of the cases as an example. For example, why is there no traffic when a new account is launched?

Remember, it is best understood from the platform's perspective, not from the user's perspective.

The platform doesn't trust you because you can't prove the value of your account.

You are a new account, the algorithm has no idea about your customer situation, and cannot evaluate the conversion rate of the anchor.

Many things in Douyin can be understood with reality, just like in the workplace, when you go to work in a company, the boss will not give you a high position right from the beginning because you cannot prove your value.

If I give you 100,000 traffic but you cannot monetize your live broadcast room, it is a loss. If you want to gain traffic, you have to prove to the platform and the algorithm that you can indeed bring value and do the KPI assessment set by the platform.

When you find that you can create value for the company, he will give you a higher position and higher equity compensation, so it makes sense.

To meet the algorithm's indicators, you need to know what indicators the algorithm has.

I suggest that to understand the indicators, we should first have a framework, which is composed of multiple parts rather than being scattered.

All the indicators we have heard of can be summarized into four types: interaction indicators, funnel indicators, transaction indicators, and traffic indicators, which constitute the entire indicator system.

Interaction indicators are the most basic user behaviors in the live broadcast room. Staying, commenting, liking, following, and fan groups are all interaction indicators.

Interaction indicators are basic indicators for stimulating the "scale" of traffic.

The key word here is scale. Traffic can be divided into two types: traffic scale and traffic quality.

In the algorithm, as long as you keep people and comment and like, you can complete the live broadcast recommendation, but it only stimulates the scale of traffic for you. If a live broadcast room wants more accurate traffic, there is only one way -

Only with a large number of transactions will the live broadcast room have a real transaction label.

Users who do not engage in transactions are just viewers.

The broadcast viewing user is a very magical existence. It is more accurate than the miscellaneous entertainment live broadcast rooms and PK rankings, but not as accurate as the serious trading users.

If there are no transactions in the live broadcast room, the algorithm cannot know what products are sold in the live broadcast room, what the average customer spending is, and the overall quality maintenance capabilities of the live broadcast room.

A live broadcast room that only has viewers but no transaction behavior is garbage.

This is what we call the interaction indicator, which is the most basic indicator.

Only when a live broadcast room has the initial interactive indicators, even if it is just a stay, can it arouse the second-level indicators of the live broadcast room, namely the funnel indicators.

These users will take some actions, such as clicking on your shopping cart, clicking on your products, or creating orders.

In order to cater to the funnel indicators, many people often use collection links to complete a stay in a live broadcast room. It also triggers shopping cart click-through rate and product click-through rate.

The closer the user is to purchasing behavior, the more weight advantage he or she has. The behavior of clicking on products and shopping carts will be more accurate than simply asking users to like and comment on user labels.

Funnel indicator, which is the basic indicator for obtaining conversion efficiency.

The key word here is conversion efficiency. If each layer of the funnel from exposure click-through rate to order conversion is done well, the conversion efficiency of the live broadcast room will be very high.

The click-through conversion rate in the live broadcast market represents the marketing efficiency of the funnel to a certain extent. You should pay attention to it and use it to see the real-time conversion effect of the live broadcast room.

Only after completing the conversion of funnel indicators, users will generate the third layer, transaction indicators.

Analysis of transaction indicators, focusing on click-through rate, ATV, UV, and GPM.

Let's take a real example. In a 30-minute broadcast, one item is shown in the first 15 minutes and another item is shown in the last 15 minutes. The flow rate of the whole show is similar, but the click-through rate of different items is different.

To give another example, the model is the same, but the two anchors are different, but your click-through rate is different.

Why is this happening? Assuming that the quality of traffic is balanced, different products can generate different conversion rates, and the conversion capabilities of the same product but different anchors are also different.

When we look at click-through conversion rates, we must be very clear about what we are looking at.

If there is only one product now, what you are looking at is the ability of the anchor. If you are the same anchor in your live broadcast room, but you use different models, you need to distinguish them specifically.

Why is the click-through conversion rate of this product so low? This is the purpose of analyzing the click-through conversion rate.

ATV is very important in trading indicators. We call the role of ATV the conversion rate of the live broadcast room to users.

Whether a user buys something for 19.9 yuan or 199 yuan is not only determined by the product, but also by the user's purchasing power.

If you want to test ATV, it is very simple. Just raise the overall average order value of the live broadcast room to a higher level to see if your overall GPM can be raised. If you cannot, this can indirectly reflect the problem, but it is only an indirect reflection.

The purpose behind the ATV test is to see how large a price range I can transfer when transferring funds in my live broadcast room.

A good transfer must be done step by step, rather than raising the price too high all at once.

You will find out why there are problems when we do activities and start selling ATV. It’s because you suddenly raise the price from 9.9 yuan to 99 yuan. How can you possibly make the transition?

Is the slow transition a process of cultivating users? This is a wrong idea. If you think that the consumption level of a live broadcast room can be cultivated through your live broadcast room, that is wishful thinking.

You are not trying to cultivate users for your live broadcast room, you are just filtering out those who cannot be consumed, leaving only the users who can truly be converted by you.

This is called the carrying capacity of ATV, and the carrying capacity is also reflected in the anchor.

What is anchor carrying capacity?

For example, when a host has 50 people online and the traffic rate is 200 in 5 minutes, he can convert 50,000 yuan.

But if there are 200 or 300 people online now, can we achieve the same transaction amount?

This is the carrying capacity of the anchor. Some anchors have exactly 50 people online and can achieve a 17% click-through rate. But when the number of online users reaches four or five hundred, the click-through conversion rate may drop.

GPM has become a must for live broadcast room data analysis because it can prove how much transaction volume I can achieve by giving you 1,000 users in the live broadcast room.

Then why not use GMV to measure? Wouldn’t it be more convenient to count the entire event?

Behind GPM is the conversion rate of flow rate. There are bugs when using the GMV of the entire show to review afterwards. For example, if the show starts for four hours, and there is extremely fast traffic in the first 30 minutes, compared with the extremely fast traffic from 30 minutes to one hour after the show, are the quality of these two batches of traffic the same?

If you really want to analyze a live broadcast room, you must split the indicators into multiple stages and time periods in real time, rather than analyzing GMV and conversion rate throughout the entire broadcast, which is meaningless.

This is the role of GPM. It tells you that I don't need you to analyze the whole thing. What I want you to see in real time how much money you can convert from these 1,000 people coming in. This is called the conversion rate of flow rate, also called phased traffic.

Therefore, we need to know how to look at GPM in the overall market, if GPM is going down. You need to especially understand why my conversion rate dropped when the number of units reached 1,000? The overall transaction amount has gone down?

Compared with GMV, GPM can reflect the presentation results of the live broadcast room data in real time, which is the general trend of data review.

Simply pursuing the overall effect of the event is like counting money afterwards. Only by analyzing data in real time and implementing optimization can you make money in the process.

UV represents how much value a single user can create in the live broadcast room.

How much user value can be obtained from a show with 100,000 viewers? This can also be combined with GPM accordingly. Because UV is the molecular factor of GPM.

If the GPM of a live broadcast is good, the value of UV will not be bad. This is why people like to use UV to measure the traffic value of the live broadcast room.

Click-through conversion rate, ATV, GPM, and UV are the keys to obtaining traffic quality. Only by doing these four points well can the traffic quality be improved.

Finally, let’s analyze the fourth indicator, the traffic indicator.

The flow index is divided into two intervals, one for field observation and one for flow rate and peak value.

The field view is simple and clear. Why are flow velocity and peak value analyzed together?

No live broadcast room can be separated from the peak flow rate analysis. This is a common misunderstanding. You often hear how many people are online today and how many people will be online in a certain period of time.

For example, if 20 people are online during the halftime live broadcast and 20 people are online at the end of the live broadcast, are the flow rates the same?

The peak value can easily mislead people into thinking that there are still users in the live broadcast room.

For the peak analysis, first, we need to look at it in combination with the flow rate, and second, we need to look at it separately in different stages.

Looking at data based on the entire live broadcast is the root cause of ineffective review.

A qualified live broadcast analysis can be divided into the first 30 minutes before the broadcast, 30 minutes to 90 minutes, 90 minutes to 180 minutes, etc. The finer the time period can be divided into for the flow rate and peak value, the more useful it will be when you analyze it.

Many people know about the 5-minute flow rate analysis, but in fact, five minutes is just for identification, but it does not necessarily have to be five minutes, it may also be at the minute level.

Sometimes there is a misunderstanding: if my traffic rate in this live broadcast is very good, will the next one be even better?

This statement is also wrong

Flow rate and peak value are not the reasons, but external manifestations of gaining recognition from the platform.

For example, if there are suddenly so many flow rates and such high peak values, it is because the data is well prepared, so there are such high flow rates and peak values.

It is not the reason that leads to your higher market view and GMV, it is just a component, and you must be very clear about this.

Similarly, traffic levels are a result rather than a cause.

Someone posted a comment on the Planet saying that today's traffic level C has reached level B. Then can I move up to a higher level?

You can't judge like this. If you use this to judge, you will have problems.

For example, if you broadcast for 3 hours today and have 5,000 views, and you broadcast for 10 hours tomorrow and still have 5,000 views, then you will find that you are still at the same traffic level.

Traffic Square is not without its advantages. The function of the traffic square is that it allows us to constantly watch live broadcast rooms of a similar level or a higher level as our own and absorb their advantages. This can improve the details in all aspects.

Since there are so many indicators, how do we determine which indicators should be used?

This is a question that most people would be confused about. I put forward a point of view, which may not be correct, but this point of view has accompanied me in creating all the ways of playing Douyin since I started, namely -

Gameplay design affects indicator design

Through a large amount of practical operations, we have discovered a pattern, that is, the combination of playing methods is actually a combination of indicator satisfaction. We use two playing methods to explain this pattern.

First is the progressive gameplay

Progressive gameplay, as a type of activity-based gameplay, meets the algorithms of different stages through cold start, follow-up period, and hot product period.

During the cold start period, we aim to attract traffic by gaining retention, interaction, and online presence, and then open live broadcast recommendations. However, in order to avoid excessive audience attendance, the cold start only keeps the initial recommendations open.

During the acceptance period, when the recommendation is opened, in order to be able to open the recommendation continuously while stabilizing the audience and washing the label, the live broadcast room will adopt the method of intensive transaction of welfare funds to complete GPM, conversion rate and UV

During the hot-selling period, traffic tends to be stable, and you only need to gradually add full-priced models and continuously test models until you find a hot-selling product.

To sum up, the progressive gameplay is in line with the principle of uniform development of the Tik Tok algorithm, so it will meet the algorithm's indicator requirements in segments.

Next, let’s look at how to play with transaction density.

When playing with transaction density, the core indicators pursued are retention and GPM. You will find that the indicators are different, and you don’t design too many likes and comments.

Make a strategic loss at the beginning, place orders continuously within 5 minutes, and complete high-density transaction density within 3 games, then you can open the live broadcast recommendation in a very short time.

The above two methods are both played in the live broadcast room, but the focus of the indicators is different. The difference between the two is that the former method pursues the balance of indicators and is therefore more long-lasting, while the transaction density only uses a single indicator, so it is a short-term and fast method.

But no matter what kind of gameplay you design, you must meet the most core indicators in the end: click-through rate, GPM and UV. These three are indispensable for any gameplay you create.

Finally, let’s summarize the indicators.

First, what kind of indicator dimension is most friendly to Douyin and merchants?

All the data are not analyzed individually, but exist with cross-weights.

If you have a single-line indicator structure, your live broadcast room is doomed to be short and fast and die quickly. But if I cross-check and evenly develop the indicators of each live broadcast room, then your live broadcast room will be able to last longer.

What does it look like exactly? If you start broadcasting with a new account, we don’t require you to set trading indicators right from the start. You should first make sure the retention rate is good and at the same time make the popularity of your entire live broadcast room very good. That is, if you can keep users, then I can recommend traffic to you.

At this time, you will find that most of these low-cost live broadcast rooms can be opened in this way, and then the initial live broadcast recommendation of traffic can be completed. This is the first step.

Every push is a test, and every test has a cost.

At this time we can explain a word called spiral upward, the spiral of data.

Every time the platform pushes traffic, it is a test, and there is a cost for testing. For example, if you have 500 views today, and it gives you 2,000 views tomorrow, it has an extra 1,500 views, which is a cost.

He gave these 1,500 views to those who could be converted in the live broadcast room. Isn't he nice? Why did he give it to you? Because he hopes that you can grow, and really wants to see whether you can receive and make the conversion. Okay, then he will appropriately push another wave of 2,000 traffic to you.

If you catch it, it means that this traffic is valuable, so you will give more traffic at this time, so you will find why the data is spiraling upward.

Of course, we will also find that some accounts can soar to the top in the first game, but such accounts are extremely rare and it is just luck to come across them.

If your data is not good, the algorithm will give the traffic to your peers to try and see if they can bring higher profits. This is horse racing.

Every push of traffic is a two-way journey of value.

Why do you say that? Every time I push something out, on the one hand the platform gives you traffic, which is a benefit to you, right? It is equivalent to a reward given to you by the platform.

When you have achieved relatively good retention, conversion rate and GPM, does it also mean that you have created daily active users and conversions for the platform? Well, he has also made money.

So this is called a win-win situation.

This is the gameplay feature of the balanced indicator structure.

Next, let’s talk about a point other than the indicators in the weights, the evolution and structure of labels and labels.

The process of a live broadcast room developing from 0 to 1 is the process of continuous increase in the audience and user tags of the live broadcast room. As long as you do a good job in people, goods and places, your live broadcast room will definitely sell well, because your users are becoming more and more accurate, and your audience is getting bigger and bigger.

Let’s continue to review what I said. Did we just solve the problem of quantity, that is, the problem of market perspective, through indicators?

This is for us to think about how we can understand qualitative evolution from the perspective of labels and trading indicators.

What is the label evolution path of a live broadcast room?

First of all, you will find that in the live broadcast room, we divide all the tags into four: basic tags, preference tags, potential tags and transaction tags.

The labels of users in a live broadcast room are initially basic labels, such as gender, age, region, etc.

You will find that the first person who comes into a newly launched live broadcast room may not even have basic tags. But when you allow your users to stay in your live broadcast room, like and comment, you will find that your tags start to become more accurate.

The refinement of basic tags is preference tags. What characteristics do users of preference tags have? He may have watched a live broadcast room similar to yours, and even liked or commented on a live broadcast room similar to yours.

Going a little deeper, there are potential tags. Users with potential tags may not have bought anything in a live broadcast room similar to yours, but they have most likely clicked on the shopping cart or product.

Going deeper into the labels, we will find transaction labels. These users are likely to have similar transaction behaviors, or even transaction behaviors for the same products. Moreover, as the labels go deeper, even the consumption frequency of users is divided.

In summary, no live broadcast room will naturally have a transaction tag, and any live broadcast room will be "tuned" by users.

From shallow to deep labeling systems, there is only one way to label, and that is user behavior.

The form of a transaction label in a live broadcast room is that a large number of qualified products have been sold, which will result in you having a transaction label.

Once we have created transaction labels, will our user portrait be accurate?

A sentence that is often heard is that our live broadcast room labels are too accurate. This is wrong. No one can say that the user portrait of his or her live broadcast room is more accurate, because the accuracy of the user portrait of the live broadcast room is infinite and is in dynamic change at any time.

For example, if you have been running a live broadcast room for one or two months, the labels may be relatively accurate. However, if you make one or two mistakes in product assembly, a large number of low-price transactions occur, or if you overuse live broadcast room activities such as super lucky bags, the labels will definitely be disrupted later.

Regarding accounts, can they be labeled manually? The answer is yes.

For example, by pushing short videos and adding influencers, and taking 1 to 3 hours of transaction delivery, you can label the account with certain labels after several consecutive broadcasts. This is effective for both new and old accounts, but it is ineffective for live broadcast rooms that have already carried a large number of label-biased behaviors.

Finally, let’s summarize our thoughts on user tags:

1. Payment is the best way to directly resolve labeling.

How can I ensure that my traffic is very accurate when I start my first live broadcast? That is to use Qianchuan or Suixintui, and bring these users in to make transactions from the very beginning. It is equivalent to that when we use Qianchuan products, we never consider adding labels. We just focus on transactions and orders in the live broadcast room.

2. If it is free flow, the evolution of labels is a necessary process for free flow.

If you want to move to a long-term live broadcast room, you can first wash out these low-priced users through welfare payments, and then use these users who can be taken over by you at a high price to slowly convert them. This is the process of label evolution.

3. Only large-scale transactions are the ultimate method of label washing

If our live broadcast room really wants to have more and more good transaction labels, we must complete large-scale transactions, and the goods in this transaction cannot be too low.

Of course, if your live broadcast room personality is very strong and the anchor is awesome, you can use a high price to wash it. The welfare fund is not a must, but it is a must for most people.

In summary, we have completed the overall thinking of the live broadcast room algorithm.

Live broadcast room gameplay design model under algorithm

The following describes my derivation of gameplay since I started using Tik Tok in September 2019.

First of all, I would like to introduce a very important word to you, which is called analogy and first principle.

What are analogies and first principles?

I see that many people nowadays would want to imitate other people’s live broadcast rooms. They would see that others are doing well and then try to do the same. However, many times the imitations fail. Why did it fail?

Because many people use analogies.

Do what your peers do. If you see your peers placing orders, place orders too. If you see your peers arranging payments, arrange payments the same way. If you see your peers setting up their live broadcast rooms, imitate them. See what their anchors are like and look for the same.

Of course we say that absolute imitation is best.

But there is a problem with this analogy method: you don’t know how the other party actually designed it.

When I disassemble other people’s live broadcast rooms, I don’t use analogies. I work from first principles.

What is the first principle? I think there are three core steps.

First, if you want to create a gameplay, the first thing you need to do is to think about what the commercialization path of the entire platform is? For example, it is February 2022, platform, what does he want now?

If you look at all my gameplay, for example, our entire live broadcast gameplay from 2020 to now, we will think about what Douyin needs.

Second, when we understand what the platform needs, we should think about how the algorithm will be driven to meet the needs of the platform?

If you think about this question, you won’t have to rely on rumors to figure out your own understanding of the algorithm.

Third, when you know that you have an understanding of the algorithm, you need to think about the last question: what kind of SOP should I use to reflect and execute this gameplay?

To make it easier to understand, let’s take progressive gameplay as an example.

The background for the emergence of progressive gameplay is that the platform has increased the speed of traffic monetization and pursued the granular value of users.

The corresponding goal is to hope that users can buy more positive products, and at the same time merchants can realize the healthy value of live streaming. In order to meet this goal, in addition to evaluating daily stay and conversion, the algorithm will also tend to evaluate the granular monetization of real-time live streaming and the value of a single user. This is how the considerations of GPM and UV come about.

Goals, packaging, content, scenarios, anchors, goods, rhetoric, activities, and processes run through the entire SOP model.

1. The first step to achieve the gameplay is to set a goal.

That is, what kind of algorithm should be met at what stage, such as the cold start, the succession period, and the explosive product period of the progressive account. The cold start opens the live broadcast recommendation through interactive indicators, the transaction density is met through welfare products in the succession period, and the explosive product is tested through precise traffic in the explosive product period.

2. After the goal is determined, the process comes.

The goal planning sets out the stage to be reached, and the process addresses the entire life cycle of the account, including how many days the cold start is expected to take, what to do on the first day, what to do on the second day, how to conduct activities on the first day, and what products to match, and it is inferred from the cold start stage to the follow-up period and then to the explosive product period.

3. Below the process is the activity combination

At each stage, I need to plan what kind of activities to stimulate user behavior and complete the assessment of the live broadcast room. For example, in the cold start, I can stimulate users to like and comment by offering extremely cost-effective products. During the follow-up period, I need to have intensive transactions to complete the KPI algorithm assessment.

4. The product mix is ​​under the activity

In order to cooperate with the activities in each stage, I should have some product grouping and selection matters. In the cold start stage, you need to attract traffic, and you need to select the traffic-generating products that meet the requirements. The same applies to the welfare funds needed in the follow-up period.

5. Underneath the product is the rhetoric

The process, goods and activities are the premise of the sales talk. What should be said in the first 30 minutes before the broadcast, and what should be said in the first 30-90 minutes after the broadcast? The sales talk revolves around the payment, the activity sales talk revolves around the activities, the sales talk revolves around placing orders and forcing orders, and the Q&A sales talk revolves around comments and atmosphere. When the sales talk is formed into a whole, it becomes the script for the anchor's live broadcast.

6. The scene is underneath the rhetoric

In order to facilitate the smooth progress of the event, what kind of live broadcast scene should be designed so that users can be attracted and the event format and product selling points can be reflected?

7. The rest is content and packaging

Combined with the gameplay of the live broadcast room, how to determine the packaging of a live broadcast room's account and personality to gain user trust and increase conversion, and how to combine videos to serve the live broadcast room's traffic while improving account recognizability.

Goals, processes, activities, products, scripts, content, and packaging in the live broadcast room can form the cornerstone of SOP. At this point, I want to say that many people will have this lesson.

We often encounter a problem on Planet: my traffic is very high, but I don’t know what to do next.

Why does this problem arise?

Because many immature traders do not have the ability to coordinate in advance.

Any trader should think clearly about all the details of the account, including the purpose, process, activities, and scripting, and know what to do at each stage of the account.

The account may not be 100% according to the plan in the end, but if you have this thing, at least you will feel at ease and the anchor will not panic.

How closely you can make this process meet expectations depends on the maturity of the trader.

But if you haven’t thought through this process clearly, then when your traffic control is improved and you start thinking about which product to use, your account will definitely die.

But in reality, many novices may not have the ability to design a complete account process.

What should I do then? Your peers are your best teachers.

If you don't have enough ability, don't try to create an original way of playing. Just see how your peers do it. How do you do the first 30 minutes of the show, how do you do the period from 30 to 90 minutes, and how do you do the period from 90 to 150 minutes? Based on all his elements, do three levels of thinking and analysis using the first principles. Play the same way they do it.

Although many live broadcast rooms have different categories and products, the principles are the same.

Whether you are a novice trader learning from your peers or a professional player designing live streaming gameplay, you can design the SOP model of the gameplay in an industrialized way based on the first principles. It cannot determine that your live broadcast room will be awesome, but it can greatly reduce your trial and error costs.

I will review the entire SOP model idea based on my actual gameplay case.

Since I founded the club, I have shared more than six original or improved ways of playing, which can actually be divided into two categories, one is called short, flat and fast way of playing, and the other is called long-term way of playing.

1. The hot-selling product launched in June.

It is a relatively short and fast way of playing, and the strategy is very simple. At the beginning, by placing pods (push as you like), you can control the field very well. Adopt the AB link model, use the low-priced model of A and the profit model of B to make it. As long as the profit model is booming, the live broadcast room will explode.

2. Lucky Bag Card Plaza

This is my relatively famous way of playing. Through the rhythm of the lucky bag, I keep going to the graded Dika Square for 10 minutes. This is also recommended for everyone in June, but it is also relatively short, flat and fast.

3. Extremely vertical number starting gameplay

Through flat broadcast + welfare payment + free push, and using the collapsed triangle mode, the conversion rate of the product is first completed and accurate live broadcast recommendations are obtained with accurate transaction data.

4. Progressive gameplay

By dividing the cold start-up, acceptance period, and explosive product period, we cater to the initial indicators and trading indicators of the live broadcast room in stages, and complete the soft landing of the account from the starting traffic to the explosive product test.

5. How to play the transaction density card list

The way to play food last year was played from October to February. That is, from the beginning, it was a large and intensive order. You can use 3-5 games to increase the conversion rate and then select a high conversion time period to start the broadcast to obtain the transaction scale.

6. Fan club gameplay

This set of presentation gameplay is based on the evolution path of Douyin algorithm, and pays more attention to the growth and repurchase of old fans. In the entire live broadcast room, you don’t have more likes and comments. As long as you ensure the addition rate of your fan club in the 7 strong games at the beginning and stimulate old fans to make transactions through the opening, the live broadcast recommendation will be opened.

The remaining short video test models, Qianchuan single products, three-horse carriage and other gameplay are no longer listed, and what we follow are the first principles.

Your ability to understand algorithms determines the creative ability of gameplay.

If you are a relatively ordinary trader, I suggest that you follow my ideas to dismantle the live broadcast room of your peers and imitate it.

If you have reached a certain level, in addition to following the hot topics, you should also try to design your own gameplay.

At the end of the article, I will give a long narrative summary of today's content, which is also my personal thoughts on the algorithm and gameplay in 2022.

First, see the essence through phenomena.

Most of the gameplay can be summarized as a cater to the algorithm, but there are differences in the form of expression.

Second, the failure of most people's live broadcast gameplay comes from the mismatch between ability and gameplay.

You obviously don’t have the ability to launch Qianchuan, you want to play Qianchuan; the anchor is a novice, you insist on playing the event and start the account; the short video ability is not good, you want to play short video test products.

Third, traders should think carefully about every decision.

As a trader, all your actions will determine the direction of the operation of the live broadcast room. Don’t hold a meeting and ask for execution when it’s hot. First think clearly about why and estimate the effect, and then make decisions. This can greatly avoid the cost of making mistakes in the live broadcast room.

Fourth, change is common, and unchanged is the cannon fodder spot.

I admit that the live broadcast room is always changing. Don’t think that if I have a set of gameplay, the gameplay is constantly innovating.

Fifth, embrace rational thinking and abandon metaphysical analysis of internal causes.

Don’t be fooled by things you don’t understand. All the ways of playing must be simple inside the road. Embrace rational thinking and truly analyze what the live broadcast room should look like.

Sixth, set the ultimate standard for gameplay based on long-term fan value.

With Douyin requiring such a high GMV this year, but the user growth is slow, it will definitely require the labels in the live broadcast room to start to deepen.

Seventh, live broadcast rooms with low traffic and high conversion are the general trend.

Don’t enjoy the climax of traffic too much. It doesn’t matter if the online is low. It doesn’t matter if the anchor starts for a long time. As long as the fans are accurate and the conversion rate is high, they can defeat 99% of the live broadcast rooms.

Eighth, content traffic is the last cost-effective weight that small and medium-sized teams can outperform.

If you still want to do live broadcasts this year, you must pay attention to the content traffic. Content traffic is my last opportunity, and most of them do not have much capital, but they can stick to the highest cost-effective traffic weight in the Douyin track.

Ninth, payment is still the best way to fight traffic uncertainty.

I will still say this sentence I said last year. In three years and four years, Taobao Express will become more and more expensive. When you want to release it again, you will not have a chance.

Tenth, character design is the best way to fight against algorithmic uncertainties.

A live broadcast room with a person is fully equipped, and it has both product professionalism and trust, which can greatly reduce the impact of the algorithm on traffic. Even if the traffic algorithm changes at any time, the live broadcast room will at least not die.

Eleventh, look at the gameplay from 0 to 1, look at the long-term effect from 1 to 10, look at the company pattern from 10 to 100.

If you want to do Douyin well, don’t ignore the importance of gameplay in the stage from 0 to 1. From 1 to 10, you must be effective and make the live broadcast room refined. When you reach a very awesome level, thinking about your corporate thinking and company layout will directly determine the ceiling of the live broadcast.

Author: Yin Chen's live broadcast

Source: Yin Chen's live streaming

No matter what the problem is, it can be disassembled using algorithms.

A comprehensive disassembly of the live broadcast room algorithm

In daily live broadcasts, we often encounter many live broadcast problems, such as streaming, peak, and flow rate in the live broadcast room, popularity and transfer of money.

No matter what the problem is, it can be disassembled using algorithms.

Algorithms are the underlying architecture of all Internet products. Whether it is Douyin or Kuaishou, the underlying operation is called an algorithm.

No one can understand algorithms 100%.

As the platform's operating architecture, the algorithm comes from human settings, but after artificial settings, all algorithms will have a machine learning process.

Second, the platform rarely talks about algorithms, so why not talk about algorithms?

Algorithms are the most core thing on a platform. As a platform, telling you the lowest logic of the algorithm is tantamount to being indifferent to the interests and risks of the platform, and telling you nakedly how to make money on the platform. This is impossible.

The algorithm allows to be guessed, but it will never be disclosed.

Most of the algorithms we come into contact with in daily life come from posteriori.

What is posteriority?

The understanding of algorithms we can gain is more about discovering existing rules through certain practical practices, including data precipitation in the live broadcast room.

The algorithm architecture will not change easily, but there is evolution.

The source of any algorithm evolution does not mean that the platform can change whatever it wants, nor does it mean that the merchant has any needs, telling the platform whether you can operate in a way.

The underlying layer of algorithm evolution is the path development of platform commercialization.

When you first did live broadcasts, you found that as long as you started broadcasting, you would have traffic.

At that time, the algorithm assessment was that I would give traffic as long as I started broadcasting, but later I would find that you still have to stay after broadcasting, convert when you stay, UV and GPM.

As a profitable product, the platform needs to use algorithms to promote users from ordinary content users to trading users.

The live broadcast room is the operating form presented by the algorithm.

A good live broadcast room must have a relatively large audience, and the users who come in the audience also have high-quality user tags.

When a live broadcast room can ensure the scale of the audience and the accuracy of users, then the live broadcast room will definitely make money. If you don’t make money, there will only be three problems: people, your goods, or your market is not good.

With the help of the evolution of algorithms, how do we understand the evolution of a live broadcast room?

The evolution of a live broadcast room from 0 to 1 is actually the evolution of two elements.

First, the evolution of the field view, will your field view become bigger and bigger, and second, the evolution of the label, will your user quality become more and more accurate. If you summarize these two things together, it is called weight.

Weight is for us live broadcast merchants. Weight can gain more scene view and user accuracy. For the platform, I can make the live broadcast room bring me more value.

This is the evolution of weight for us, and the evolution of monetization value for the platform.

In terms of the evolutionary form of the live broadcast room, viewing the gameplay from a dialectical perspective

No gameplay is always effective, because gameplay is also evolving, but the long-term effectiveness of different gameplays is different.

Based on long-term classification, we will find that all gameplay can be divided into two types. The first type will maintain continuous optimization so that this live broadcast room can survive as much as possible.

The second possibility is that he only appeared very briefly, for example, everyone will see some violent live broadcast rooms.

To summarize a common feature, as long as it is a short, flat and fast gameplay, the pursuit of quantitative evolution must be quantitative evolution.

No matter what the 13-line gameplay I saw some time ago or if I go forward to push other gameplay, it is a quantitative change. It is difficult to complete and can satisfy the user tags and live broadcast room view.

The second point is to pursue only a single indicator.

What is a single indicator? What we often talk about is stay, conversion rate, GPM, UV, etc.

In an ideal live broadcast room, all indicators are developed balancedly according to the rules of the platform. However, 99% of live broadcast rooms cannot achieve an ideal model, and I can't do it either, either. We can only get close to this level.

If the live broadcast room only pursues a single factor to explode, your live broadcast room will also explode.

However, the traffic is limited. Although the short-term gameplay can make the scene very large, the traffic of the entire platform is very limited from the platform's perspective.

Since you are given so much traffic, you must pursue corresponding commercial value.

This is called pursuing the maximization of traffic value.

The platform gives you 100,000 views, and also gives you another live broadcast room. You cashed in 50,000 yuan, and others cashed in 500,000 yuan.

Your data is very poor, and the platform thinks that you may have a good start in the end, but if you have a bad start in a row, do you think the platform will give you traffic again? It won't give it to you.

This is called the horse racing mechanism.

In the live broadcast room of the same category and level, I gave you the same traffic size, but you have no way to complete the monetization granularity of the same unit. If you are unbalanced, you will be eliminated by the horse racing.

This is what we see from a strategic perspective and from a very high perspective.

Regarding traffic, you need to enjoy the latency satisfaction of traffic.

What is the delay satisfaction of traffic?

When the platform gives you a very large stream, it will not only evaluate your basic indicators, it will evaluate your deep indicators.

The people and goods yards in your live broadcast room, including your conversion ability, have not met the standards. If you cannot meet the deep-level indicators of the platform, the traffic in the live broadcast room will continue.

This is why most excellent live broadcast rooms choose to use initial weights to first use quantity, then test the overall conversion, and then expand traffic.

We summarize the rules as to obtain traffic, undertake traffic, and continue to obtain traffic. We also summarize it as not being separated from quality and pursuing the development of quantity.

You see, all of our gameplay is basically this way.

Fairness is the product of algorithms.

The competition in Douyin is very fierce, but from the perspective of civilians, Douyin's algorithm is very fair. Without this algorithm, there would be no ordinary people making a live broadcast counterattack.

So the essence of the algorithm is fair.

The algorithm sounds complicated, but whether the live broadcaster has the ability to analyze algorithms is not necessarily the case. It depends on your ambition for live broadcast. A couple can even have successful cases if they don’t understand any algorithm, but they lack the ability to copy the live broadcast room.

This is like doing math problems. You can answer one question by chance, but if you are not familiar with the solution, you must not be sure to do the second question correctly, even if the answer is the same answer.

How the algorithm guides the behavior of the live broadcast room, this involves the split of indicators of the live broadcast room algorithm system.

A sentence we often hear is what kind of indicators we should make in the live broadcast room.

We usually face three illusions.

The first one, you don't know what indicators exist.

The second type is that among so many indicators, I don’t know what the relationship between indicators and indicators is.

The third type is, among the indicators, you don’t know which indicators are important and which focus on doing it.

First of all, we need to understand the importance of an indicator.

Why is there no traffic in the live broadcast room? Why does the streaming traffic go on and go down later? Why can't you transfer money? etc.

Take one of these cases as an example, for example, why is there no traffic when a new account is broadcast?

Remember, it is most appropriate to understand it from the perspective of the platform, not from the perspective of the user.

The platform does not trust you because you cannot prove the value of your account.

You are a new account, and the algorithm has no idea about your people and cargo yard situation, nor can you evaluate the anchor's conversion power

Many things on TikTok can be understood in reality. Just like in the workplace, when you go to a company, the boss will not give you a high position at the beginning, because you cannot prove your value.

Give you 100,000 traffic, but the live broadcast room cannot be cashed in. This is a loss. If you want to obtain traffic, you must prove to the platform and the algorithm that I can indeed bring value and do the KPI assessment formulated by the platform.

When you find that you can create value for the company, it will give you higher positions and higher equity compensation, so it is also true.

To meet the algorithm's indicators, you must know what the algorithm has.

I suggest that in the understanding of indicators, there is a framework first, which is composed of multiple parts rather than fragmented.

All the indicators we hear can be summarized into which four types: interactive indicators, funnel indicators, trading indicators, and traffic indicators, which constitute all indicator systems.

Interaction indicators are the most basic live broadcast room behavior of users, such as staying, commenting, likes, following, and fan clubs. These are interactive indicators.

Interactive indicators are the basic indicators that stimulate the "scale" of traffic.

There is a key word here: scale, and traffic is divided into two types, one is called traffic scale and the other is called traffic quality.

In the algorithm, as long as you ensure that people are kept and liked by comments, you can complete live broadcast recommendations, but it only stimulates the traffic scale for you. If a live broadcast room wants more accurate traffic, there is only one way-

Only with a large number of transactions can the live broadcast room have a real trading label.

Users who have no transaction behavior are just watch broadcast users.

Watching broadcast users is a very magical existence. They are both more than the various entertainment live broadcast rooms and accurate PK rankings, but they are not as good as the serious trading users.

There is no transaction in the live broadcast room, and the algorithm cannot know what products the live broadcast room sells, what unit price the live broadcast room sells, and the overall quality and maintenance capabilities of the live broadcast room.

A live broadcast room that only has users who watch the broadcast but lacks trading behavior is garbage.

This is what we are talking about in interaction indicators, which are the most basic indicators.

When a live broadcast room has initial interactive indicators, even if it is to stay, it is possible to evoke the second level of the live broadcast room, namely the funnel indicator.

These users will do some behaviors, maybe clicking on your shopping cart, clicking on your products, or maybe creating an order.

In order to cater to the funnel indicator, many people often use collection links to complete the stay in a live broadcast room. It also drives the click-through rate of shopping carts and product click-through rate.

The closer the user is to buy, the more weight the advantage. The behavior of clicking on products and shopping carts will be more accurate than simply asking users to like and comment on users’ tags.

Funnel indicator, it is the basic indicator for obtaining conversion efficiency.

The key words here are conversion efficiency. If the funnel is made from each layer of the funnel from exposure click-through rate to order conversion, the conversion efficiency of the live broadcast room will be very high.

The click conversion rate in the live broadcast market represents the marketing efficiency of the funnel to a certain extent. You need to pay attention to it and see the real-time conversion effect of the live broadcast room through this point.

Only when the funnel indicator is converted can the user generate the third layer, trading indicator.

The analysis of trading indicators focuses on analyzing click conversion rate, ATV, UV, and GPM.

To give a realistic example, if it takes 30 minutes to start broadcasting, one model will be displayed in the first 15 minutes and another model will be released in the next 15 minutes. The flow rate of the whole game is similar, but if it is used in different models, the click conversion rate will be different.

Let me give you another example. The model is the same model, but the two anchors are not the same, but your click conversion rate is different.

Why is this happening? Assume that the conversion rates of different products can be different based on the premise of balanced traffic quality, and the conversion capabilities of different anchors are different.

When we are looking at click conversion rates, we must understand very well what we are looking at.

If there is only one product now, what you are looking at is the ability of the anchor. If you have the same anchor in your live broadcast room, but you have different models, you have to distinguish them specifically.

Why did I use this model? The click conversion rate is so poor? This is the purpose of your analysis of click conversion rate.

ATV is very important in trading indicators. Regarding the role of ATV, we call it the conversion rate of the live broadcast room to users.

A user buys 19.9 yuan and 199, which is not only determined by the product, but also by the user's purchasing power.

It is very simple to test ATV. You can increase the overall average customer price of the live broadcast room to see if your entire GPM can be raised. If you can't raise it, this can reflect the problem from the side, but from the side.

Behind the ATV test is how much price I can transfer when I transfer money in the live broadcast room.

For good transfers, you must transfer them step by step, rather than making the price particularly high.

You will find out why I had a problem with the ATV when we started the event. It is because you immediately increased from the price of 9.9 to 99. How could you make a transition?

Is the slow transition a process of cultivating users? This is a wrong idea. If you think that the consumption level of a live broadcast room can be cultivated through your live broadcast room, it is Tianfang Night Hall.

You are not trying to cultivate users in your live broadcast room, you are just filtering out those that cannot be consumed, and what you leave behind is users who can truly be converted by you.

This is called the carrying capacity of ATV, and the carrying capacity is also reflected in the anchor

What is anchor carrying capacity?

For example, when an anchor has 50 people online, the 5-minute flow rate ratio is 200, and he can convert 50,000 yuan.

But if there are 200 or 300 people online now, can we achieve the same transaction amount?

This is the carrying capacity of anchors. Some anchors have exactly 50 people online and can achieve a click conversion rate of 17%. But when it is as high as 400 or 500 online, the click conversion rate may drop.

GPM is already a must-have for data analysis in the live broadcast room, because it can prove how much transaction volume I give you 1,000 users in the live broadcast room can help me complete.

Then why not use GMV to measure it? Shouldn’t the whole game statistics be more convenient?

Behind GPM is the conversion rate of flow velocity. Afterwards, there are bugs when using the entire GMV to review the movie. For example, four hours of broadcasting, in the first 30 minutes of extreme speed traffic. Compared with the time from 30 minutes of broadcasting to one hour of broadcasting, the traffic quality of these two batches is the same?

If you really analyze a live broadcast room, you must split the indicators into multiple stages and periods in real time, rather than analyzing GMV and analyzing the conversion rate throughout the game, which is meaningless.

GPM is what it does. It tells you that I don’t need you to analyze the whole game. What I want is that you can see in real time how much you can convert these 1,000 people come in. This is called the conversion rate of flow rate, also called phased traffic.

So you should know how to look at GPM in the market, if GPM goes down. You have to understand particularly why my conversion rate has declined in 1,000 units? The overall transaction amount has gone down?

Compared with GMV, GPM can better reflect the live broadcast room data presentation results in real time, which is the general trend of data review.

Simply pursuing the entire performance is like counting money afterwards. Only by analyzing data in real time and completing optimization can you make money in the process.

UV represents how much value a single user can create in the live broadcast room.

How much user value can be obtained in a 100,000-game view can also be combined with GPM. Because UV is a molecular factor of GPM.

If a live broadcast GPM is done well, the value of UV will not be bad, which is why people like to use UV to measure the value of traffic in the live broadcast room.

Click conversion rate, ATV, GPM, and UV are the keys to obtaining traffic quality. Only by doing these four points well can traffic quality improve.

Finally, let’s analyze the fourth indicator, traffic indicator.

The flow index is divided into two intervals, one is field view, and the other is flow velocity and peak value.

The field view is simple and clear, and the flow rate and peak value are combined to analyze it?

There is no live broadcast room that can escape the peak of flow rate analysis. This is a misunderstanding that you can often see, which is that you often hear how many people are online today and how long it takes to stay.

For example, 20 people live broadcasts online in the midfield, and 20 people are online in the end. Is the flow rate the same?

The peak is likely to cause misleading, thinking that there is still user flow in the live broadcast room

For peak analysis, first, look at it in combination with flow velocity, second, look at it in separate stages.

Watching data based on the entire live broadcast is the root cause of invalid review.

A qualified live broadcast analysis can be classified as the first 30 minutes before the broadcast, 30 minutes to 90 minutes, 90 minutes to 180 minutes, etc. If the period when the flow rate and peak can be divided, the more useful it will be when you analyze it.

Many people know about the 5-minute flow rate analysis, but in fact, five minutes is just for discernment, but it may not be five minutes, but it may also be at the minute level.

Sometimes there are also misunderstandings. My flow rate in this live broadcast is very good, so is the live broadcast even more powerful?

This statement is also wrong

Flow rate and peak value are not the reasons, but external manifestations recognized by the platform.

For example, there are suddenly so many flow rates and such high peaks. That is because the data is done well, so there is such high flow rates and peaks.

It is not the reason why you have a higher field view and GMV, it is just a constituent element, and this must be very clear.

Similarly, the traffic level is also a result rather than a cause.

Someone posted a comment on Planet saying that today's traffic level C has reached B. Then can I go up to a higher level next?

You can't make a judgment like this, if you use this to make judgments, you will have problems.

For example, if you start broadcasting for 3 hours today, 5,000 views, and 10 hours tomorrow, you will still have 5,000 views, then you will find that you are still in the same traffic level.

The traffic square is not useless either. The role of the traffic square allows us to keep watching live broadcast rooms at similar levels or higher-level live broadcast rooms to absorb their advantages. This can improve the details in all aspects.

Since there are so many indicators, how do you judge what indicators should be made?

This is a question that most people will wonder. I put forward a point of view, which is not necessarily correct, but this point of view has accompanied me in creating all the ways to play since I was engaged in TikTok, that is,-

Design of gameplay influence indicators

In a lot of practical practices, we found a rule, that is, the combination of gameplay is actually a combination of indicators. We use two ways of playing to explain such rule.

First, the progressive gameplay

Progressive gameplay, as a type of activity-based gameplay, meets different stages through cold start, acceptance period, and hot product period zone.

During the cold start period, we pursue the method of attracting traffic, to get stay, interact, and online, and then open the live broadcast recommendation, but in order to avoid too high the scene, the cold start only keeps the initial recommendation open.

During the acceptance period, when the recommendation is turned on, in order to continuously open the recommendation, stabilize the view of the venue and wash the tags, the live broadcast room will adopt the integrated transaction method of welfare payments to complete GPM, conversion rate, and UV.

During the period of hot products, the traffic will stabilize. You only need to gradually add the regular price model and continue to test the model until the hot products are measured.

The summary of progressive gameplay is in line with the principle of uniform development of Douyin algorithm, so it will meet the algorithm's index needs in segments.

Secondly, let’s look at how to play transaction density

The core indicator of trading density is to pursue as a stay and GPM. You will find that the indicators are different and do not design too many likes and comments.

If you make strategic losses at the beginning, you can place orders continuously within 5 minutes and complete high-intensive transaction density within 3 games, and you can open the live broadcast recommendation in a very short time.

Both of the above methods are live broadcast room gameplay, but the focus of indicators is different. The difference between the two is that the former method pursues balance of indicators and is more effective, while the transaction density is only a single indicator, so it belongs to the short and flat fast gameplay.

But no matter what gameplay you design, you must meet the most core indicators in the end, click conversion rate, GPM and UV, these three things you cannot escape from in any gameplay you do.

Finally, let’s summarize the indicators

First, what kind of indicator dimension is the most friendly to Douyin and merchants

All data are not analyzed separately, but data exists with cross-weights.

If you have a single-line indicator structure, your live broadcast room is destined to die soon. But if I have crossed the indicators of each live broadcast room and made uniform development, then your live broadcast room can live longer.

What does it look like specifically? If you are on the airing of a new account, we do not require you to make trading indicators from the beginning. At the beginning, you first make the retention rate well and make the popularity of your entire live broadcast room very good, that is, if you can keep users, I can recommend traffic to you.

At this time, you will find that most of the low-priced live broadcast rooms can be opened in this way, and then the initial live broadcast recommendation of traffic is opened. This is the first step.

Every push is a test, and test has a cost.

At this time, we can explain a word called spiral rise, the spiral of data.

Every time the platform pushes and streams, it is a test, and it has a cost to test it. For example, if you have a field view of 500 today, it will give you 2,000 tomorrow, and it will have an additional 1,500 views, which is a cost.

He gave these 1,500 views to those who could convert the live broadcast room. Isn't he smelly? Why did he give it to you? Because he hopes that you can grow, it really depends on whether you can receive it and make the conversion. OK, he appropriately promotes you another wave of 2,000 traffic.

If you catch it, it means that the traffic is valuable, then you will give more traffic at this time, so you will find out why the data is spiraling up.

Of course, we will also find that some accounts can soar in the first game, but there are very few such accounts, and it is luck to meet them.

If your data is not done well, the algorithm will give this traffic to your peers to try these traffic to see if it can get higher profits. This is horse racing.

Every push is a two-way value rush.

Why do you say so? When I push every time I push the traffic, on the one hand, the platform gives you traffic, which is a benefit to you, right? It is equivalent to the reward given to you by the platform.

When you have also completed better stays, conversion rates and GPM, does that mean that you have also created daily active users and conversions for the platform, and well, he has made money.

So this is called two-way win-win.

This is the gameplay feature of the balanced indicator structure

Next, let’s talk about a point other than the indicators in the weight, the evolution and structure of labels and labels.

The process of a live broadcast room developing from 0 to 1 is the process of increasing the audience's view and increasing the user tags of the live broadcast room. As long as you do a good job in the cargo yard, your live broadcast room will definitely sell well, because your users are becoming more and more accurate and your view is becoming bigger and bigger.

Let’s continue to review what I said. Did we solve the problem of quantity through indicators, that is, the problem of field view?

This is how we think about how we understand qualitative evolution from the perspective of labels and trading indicators.

What is the evolution path of a live broadcast room?

First of all, you will find that in the live broadcast room, we divide all labels into four, basic labels, preference labels, and potential labels trading labels.

The user's tags in a live broadcast room are initially basic tags, such as gender, age, region, etc.

You will find that the first person who flashes in the newly launched live broadcast room may not even have the basic label. But when you ask your users to stay in the live broadcast room, like, and comment, you find that your tags are beginning to become accurate.

Basic label precision means preferring labels. What characteristics do users who prefer labels have? He may have seen a live broadcast room similar to you, and even liked and commented in a live broadcast room similar to you.

Going further down, it is the potential tag. Users of the potential tag may not have bought anything in your similar live broadcast room, but they are likely to have clicked on the shopping cart or products.

Further down the label is the transaction tag. These users are likely to have transaction behaviors for the same type of purpose, and even the same product trading behaviors. As the label deepens, even the user's consumption frequency is divided.

In summary, no live broadcast room will naturally have a transaction label, and any live broadcast room will be "trained" by users.

There is only one way to mark the labeling system from shallow to deep, that is user behavior.

The form of a live broadcast room transaction label is that a large number of qualified products are completed, which will lead to you having a transaction label.

When we make the transaction label, will our user portrait be accurate?

I often hear a saying that our live broadcast room labels are accurate. This is wrong. No one can say that the user portraits in their live broadcast room are more accurate, because the accuracy of the user portraits in the live broadcast room is infinite and is undergoing dynamic changes at any time.

For example, if you have been in a live broadcast room for a month or two, the label is relatively accurate, but as long as you make one or two product group mistakes, there are a large number of low-priced transactions, or if the live broadcast room activities are too fierce, the label will definitely be disrupted in the future.

For accounts, the answer is OK whether they can be tagged through artificial means.

For example, by pushing at will, short videos are similar to those of Jiada people. They take 1 to 3 hours of transactions and placement, and can label the account in a few consecutive games. Both new and old accounts are valid, but they are invalid for live broadcast rooms that have already carried a large number of label-oriented behaviors.

Finally, summarize the thoughts on user tags—

1. Payment is the best way to directly solve the label.

How can I ensure that my traffic is very accurate when I first broadcast, that is, Qianchuan placement or freely push placement. From the beginning, I used these users to make transactions. As for us, we never considered making labels when playing Qianchuan single products. We played with the live broadcast room and orders.

2. If it is a free stream, the evolution of tags is a necessary process for free streaming.

If you want to go to a long-term live broadcast room, then you first wash out these low-priced users through the welfare payment method, and then use these users that can be purchased by you to slowly change them. This is a process of label evolution.

3. Only large-scale transactions are the ultimate way to clean up tags

If we want to really have a good trading label in the live broadcast room, we must have completed a large-scale transaction, and the commodity in this transaction cannot be too low.

Of course, if your live broadcast room has a strong personality and the anchor is very awesome, you can wash it at a high price. It is not a necessary item for welfare payment, but for most of them, it is a must.

In summary, we have completed the overall thinking on the live broadcast room algorithm

Live broadcast room gameplay design model under algorithm

The following tells the story of my gameplay deduction since I started working on Douyin in September 2019.

First of all, let me give you a word, which is very important, called analogy and the first principle.

What is analogy and first principle?

I see that many people nowadays think about imitating other people's live broadcast rooms, watching others do well, and then doing it too, but many times imitation fails. Why does it fail?

Because many people use analogy.

You do whatever you see your peers. When you see your peers place orders, you place orders. When you see your peers place orders, you arrange them like this. When you see your peers set up the live broadcast room scene, you imitate it to build it, see what the anchor is like and what you look for.

Of course we say that absolute imitation is the best.

But this analogy method will have a problem, you don’t know how the other party designed it.

When I was dismantling other people’s live broadcast rooms, I didn’t need to compare them. I am using the "first principle".

What is the first principle? I think there are three core steps.

First, if you want to create a gameplay, the first thing you have to do is to think about what the commercialization path of the entire platform looks like? For example, it is February 2022, what does the platform want now?

Look at all my gameplay, for example, from 2020 to the current live gameplay, we will think about what Douyin needs.

Second, when we understand what the platform needs, we will think about how the algorithm will promote it in response to the needs of the platform?

If you think about this question, you don’t have to hear from it, but you can figure out your own understanding of the algorithm.

Third, when you know that you have an understanding of the algorithm, you have to think about the last question. What kind of SOP should I use to reflect and execute this set of SOPs.

For easier understanding, we use progressive gameplay as an example.

The background of the progressive gameplay is that the platform has strengthened the speed of traffic monetization and chased the granular value of users.

對應的就是希望用戶能夠購買更多正向性商品，同時商家能實現良性的直播價值，為了滿足這點，算法除了考核日常停留、轉化外，也會傾向考核實時直播的顆粒變現及單用戶的價值，GPM、UV的考量就是這么出來的。

目標、包裝、內容、場景、主播、貨品、話術、活動、流程貫穿了整個SOP模型。

一、實現玩法第一步要做的是目標。

即什么階段要滿足怎樣的算法，比如遞進式起號的冷啟動、承接期、爆品期，冷啟動通過互動指標打開直播推薦，承接期通過福利品滿足成交密度，爆品期通過精準流量測試爆品。

二、目標確定之下就是流程。

目標規劃了要達到的階段，流程解決賬號的整個生命周期，冷啟動預計幾天，第一天該做什么，第二天該做什么，第一天里面活動如何進行、搭配什么品，從冷啟動階段推斷到承接期再推斷到爆品期。

三、流程之下是活動搭配

我需要在每個階段，策劃什么樣的活動來激發用戶行為，完成直播間的考核，比如冷啟動通過極致性價比商品，激發用戶來點贊、評論，承接期需要密集成交的形式，完成KPI算法考核。

四、活動之下是貨品的搭配

為了配合每個階段的活動，我應該會有怎樣的組品跟選品事項，冷啟動階段需要引流款，你就需要去挑選符合條件的引流款，承接期需要福利款，也是同樣的道理。

五、貨品之下就是話術

流程、貨品、活動是話術的前提，開播前30分鐘話術要怎么說，開播30-90分鐘要如何說，圍繞講款的講款話術、圍繞活動的活動話術、圍繞打單、逼單的成單話術，圍繞評論、氣氛的百問百答話術，當把話術構成一個整體，就成了主播直播的腳本。

六、話術之下是場景

為了配合活動流暢進行，應該設計怎樣的直播場景，用戶能夠被吸引，活動形式、商品賣點能夠被體現

七、剩下的就是內容與包裝

結合直播間的玩法，如何去確定一個直播間的賬號與人設的包裝，讓用戶信任并提升轉化，如何結合視頻在提升賬號辨識性的基礎上，為直播間的流量服務。

目標、流程、活動、貨品、話術、內容、包裝直播間能形成SOP化的基石，這時候我想說一個很多的人會有這個教訓。

在星球我們經常會遇到一種問題，我的流量拉得很高了，但是卻不知道接下來怎么做。

Why is this problem?

因為很多不成熟的操盤手，不具備事先統籌的思維。

任何一個操盤手，都應該想清楚賬號的目的、流程、活動、話術等所有的細節，知道賬號到了每一個階段該如何去做。

賬號最終可能并不是100% 按照規劃來，但是你有了這個東西，至少心里面會很穩，主播也不會很慌。

你能夠把這個流程制造得越符合預期，取決于操盤手的成熟程度。

但是連這個流程都沒有想清楚，那么等到你的流量場控上去，你再去想說我用哪個品來接，你的賬號一定會死掉。

但現實當中，對很多的新手，可能并不具備設計一個完整的賬號流程的能力。

What should I do then?同行是你最好的老師。

如果能力不夠，就不要去原創玩法，就去看你的同行他是怎么去弄的，你把她的同行的開播的前30 分鐘怎么弄，開播到30 分鐘到90 分鐘怎么弄、90 到150 分鐘怎么弄，根據把他的所有要素，按照第一原理的方式去做三級的思考與分析，他怎么玩，你給我怎么玩。

很多直播間雖然品類不一，產品不一，但是道理相通。

不管是小白操盤手學習同行，還是專業選手設計直播玩法，都可以以第一原理的方式，工業化的去設計玩法的SOP模型，它不能決定你直播間一定會做得非常牛逼，但是可以大幅度降低你的試錯成本。

我以我實際的玩法案例，復盤下整個SOP的模型思路。

創建俱樂部以來，我分享過原創，或者改良的玩法超過六種，其實都可以分為兩類，一個叫短平快的玩法，一個叫長效型的玩法。

一、6月份推出的隨心推測爆品。

是比較短平快的玩法，玩法策略很簡單，在一開場通過投放豆莢（隨心推），把場控拉得非常高。采用AB 鏈接的模式，用A 的低價款跟B 的利潤款穿插去打，只要利潤款爆，直播間就會拉爆。

二、福袋卡廣場

這個是我相對有名的玩法，通過福袋的節奏性，10分鐘不斷地去分級地卡廣場。這也是6 月份是給大家推薦的，但它也是比較偏短平快的。

三、極度垂直起號玩法

通過平播+福利款+隨心推，用倒閉三角形模式，首先完成商品的轉化率，用精準的交易數據獲得精準的直播推薦。

四、遞進式玩法

通過劃分冷啟動、承接期、爆品期，分階段迎合直播間的初始指標跟交易指標，完成賬號從起流量到測爆品的軟著陸。

五、成交密度卡榜玩法

去年玩食品的玩法，從 10 月份開始玩，到現在2 月份一直在玩，即從一開始放單是大密集放單，3-5場即可通過轉化率遞增撕開直播推薦，再配合卡榜選擇高轉化時間段開播獲取成交規模。

六、粉絲團玩法

這套呈現玩法基于抖音算法的演變路徑，比較多關注老粉用戶的增長與復購性。在整個層直播間里面，不過多打點贊評論，只要在開場的強7 場里面，保證你的粉絲團的添加率并且通過開場激發老粉成交，直播推薦就會打開。

剩下的短視頻測爆款、千川單品、三駕馬車等玩法，不再列舉，所遵循的，均是第一原理。

你對算法的理解能力決定了玩法的創作能力。

如果說你是個比較平平淡淡的一個操盤手，我建議你按照我的那個思路去拆同行的直播間去模仿是最佳的。

如果已經到了一定級別，除了緊跟熱點，也要嘗試去自行設計玩法。

到文章最后，針對今日的內容，做一個長敘性的總結，也是我個人對2022年算法與玩法的思考。

第一，透過現象看本質。

大多數的玩法可以歸納為對算法的迎合，只不過表現形式是有區別而已。

第二，大多數人直播玩法的失敗，來源于能力跟玩法的不匹配

你明明沒有千川投放能力，你要去玩千川；明明主播是個新手，偏要去玩活動起號；明明短視頻能力不行，偏要玩短視頻測爆品。

第三，操盤手每一個決策都應當深思熟慮。

作為一名操盤手，你的所有行為將決定直播間的運轉風向，不要一時頭熱就開會要求執行，先想清楚為什么怎么做，并預估效果，再去做決策，能夠極大避免直播間的犯錯成本

第四，變是家常便放，不變是炮灰點放。

承認直播間永遠都在變化，你不要覺得好像我有套玩法就怎么樣，玩法也不斷在創新。

第五，擁抱理性思維，拋棄玄學分析內因。

不要被自己聽不懂的東西忽悠，所有的玩法都一定是大道內簡，擁抱理性思維，真正地去分析直播間應該是什么樣子。

第六，以長效粉絲價值去制定玩法的終極標準。

抖音在要求今年有這么高的GMV ，但是用戶增長又緩慢的情況下，他一定會要求直播間里面的標簽開始更加深化。

第七，低流量高轉化的直播間是大勢所趨。

不要太多去享受流量的高潮性，在線很低沒關系，主播開播時間長點沒關系，粉絲只要精準，轉化率只要高，就能打敗99%的直播間。

第八，內容流量，是中小團隊能跑贏的最后性價比砝碼。

如果你現在今年還想做直播，內容流量一定要抓好，內容流量是我最后的風口，也是大多數沒有太多資金，但是能夠堅持在抖音賽道最高性價比的流量砝碼。

第九，付費依舊是對抗流量不確定最佳手段。

我去年講的這句話，我今年還是會講，過三年，四年以后，像淘寶直通車越來越貴，等到你想要再投放，你都沒有機會。

第十，人設是對抗算法不確定因素的最佳手段。

一個有人設直播間，只要人設足夠飽滿，既有產品專業度，又有信任感，能極大降低算法對流量的影響，哪怕任何時候流量算法在改變，直播間至少不會死。

第十一，從0到1看玩法，從1到10看長效，從10到100看公司格局。

如果想把抖音做好，在從0 到1 的階段，不要忽略玩法的重要性，從1 到10，則一定要長效，把直播間精細化做好。當做到非常牛逼的程度，思考你的企業思維，公司的格局，將直接決定直播的天花板。

Author: Yin Chen's live broadcast

Source: Yin Chen's live streaming