How to analyze live streaming traffic!

There are two topics today. The first one is "Cold start traffic design for new accounts" and the second one is "Rescue methods for account level decline". These are actually two problems at different stages of an account. I will give you a comprehensive analysis model, and I suggest that those of you who are still in the cold start stage read it carefully.

The first part is the account assessment mechanism, which every trader should memorize. The account assessment mechanisms are different at different stages and at different levels of traffic recommendations.

When a new account is cold-started, the platform will assess the popularity index of the live broadcast room. As long as the data indicators such as stay, comments, and conversion to fans are increased, the recommendation will naturally be enabled.

However, once you have more and more live broadcasts and their audience grows, the platform's assessment of you will be multifaceted. It is not enough to just meet the interactive indicators. Your transaction indicators will also be required to increase. If you cannot increase them, your level will naturally drop. Why does your level drop?

Because of the horse racing mechanism

Accounts of the same category that started out like you have the same traffic as you, but the commercial value of their conversions is higher than yours. So why should the algorithm continue to give you traffic? This is a very simple truth. If you want to get better data when doing live streaming, you should race with your peers. If you beat your peers, Douyin will not treat you unfairly.

The second part naturally recommends how to open it.

I always ask everyone to pay attention to the combination of several elements, including scenes, activities, products, words of speech, and rhythm. Because I am doing a cold start, no matter which live broadcast room, I use these five elements to make a combination.

First, break it down from the scene. Pay close attention to the details such as the background of your live broadcast room, the display of the main products, and the stickers in the live broadcast room. Ask yourself a question: if you were a user and you entered your live broadcast room, could you stay for 3 seconds and still have the desire to continue watching the live broadcast?

If not, then your live broadcast scene must be unqualified. Don’t we have such experience ourselves?

Sometimes when I enter some live broadcast rooms, the host doesn’t speak, but just by looking at the scene, I can’t help but stay. Many times it’s just because I’m lazy, or I don’t care about the importance of the scene at all.

Vision is always more important than hearing. If you optimize the scene well, you have achieved the most basic stay.

Second, activities. If you want to attract traffic, activities are a must. You need to think clearly about what kind of activities you should design so that these general traffic can be attracted by you.

9.9 low price flash sale? Sure, a big lucky bag? That’s fine. Since you’re a new account and regular-priced products can’t keep people there, right? So you can only do promotions, right?

When an offline store opens, every store has to hold activities. How do they do it? This question is really very simple. Don’t fill it in your mind. You are a newbie and you are not that good. Spend some time and use tools like Chanmama and Douchacha to crawl accounts that have just started broadcasting like yours and see what activities others are doing. If others can increase their traffic, why can’t you?

Third, after you have an event, you need to consider your own products and whether you can choose to support your event.

Sometimes people will have another way of thinking, which is to check my products first, and then see what activities can be combined. I don’t think this logic is contradictory.

Based on my own supply chain capabilities, I can find the supply chain for whatever product I want, so I design the activity path first and then look for the products to be replenished.

If your supply chain is limited, then you can bring products to screen the form of activities. It is the same. The final point is still the same. You must have products that can support the execution of your activities, and you need to have a lot of these products. Not many live broadcast rooms can make the live broadcast room explode by selecting a product to attract traffic.

Many newcomers are either blindly confident or lack cognition. They have been trying to use traffic-generating products for half a month but still cannot increase traffic. They just optimize the wording and scenes but never think about whether it is a traffic-generating product.

Fourth, sales pitch. Once the activity and product are launched, it’s time to design the sales pitch.

New traffic does not have that much patience, and the holding time cannot be too bad, so you need to design a five-minute sales pitch.

Don’t know how to talk? Just copy it. All the peers you can think of are doing it. Just copy their words and adapt them according to your own product. It’s not a difficult thing to do.

Finally, it is rhythm. I think rhythm is the most difficult element to overcome among these five elements, except for quality.

The rhythm is actually linked to the ability of the anchor. Many live broadcast rooms are hosted by new anchors, but they insist on participating in activities to create an account. Isn’t this a paradox?

You can give any anchor the script and he can speak very well as long as he can open his mouth, but the rhythm requires experience. So what I have always suggested is that if conditions permit, it is better to find a more mature anchor to do a cold start, rather than a new anchor, because it is not that fast to train a new anchor.

Even if you have passed the cold start of recommended traffic, due to the lack of relative experience in product scheduling, speech response, and rhythm control skills, the traffic will be widened in the later stage. This is a problem of the host's carrying capacity.

To come back to the topic, if you are a new anchor, I really recommend that you let your anchor watch more live broadcast rooms of excellent peers to see how others play the rhythm. Then when you start broadcasting yourself, let the anchor imitate at any time. Review with the anchor more after the broadcast and compare the gap between you and others. As a result, you will see the anchor's progress in a week or two, but if you want to make great progress, you must broadcast continuously.

The third part is the problem of dropping levels, that is, after broadcasting a few data shows, the traffic began to decline.

I won’t talk about the cause of the problem first, because there is a mistake that too many people make, which is the cold start of a new account. Everyone always focuses on how to tear off the natural recommendations, but does not consider in advance how I should take over the live broadcast room after the scene is opened.

I used a word here, called "in advance". For those friends who have dropped in level now, you should ask yourself, when the account has not yet been fully launched, have you ever seriously thought about these issues? My viewership has reached 10,000, which data should I focus on? In order to prepare these data, are my products ready? Has the anchor’s script for the follow-up period been written? What should I do if my traffic drops?

Many people will not think about this question in advance. We have all heard of an anecdote called Tian Ji's Horse Racing, and there is no need to explain the content in detail. But this story tells us a truth, that is, when we run any live broadcast room, we should think clearly from the very beginning about how we should adopt strategies at each stage.

If you don't do this, when the audience increases, Douyin will not ask you to stop for a few days and think clearly before starting live broadcasting. Instead, you have to solve the problem of succession the next day after the audience increases. What will you do? This is where many people make mistakes. You can see that the monthly GMV of Chuangzao exceeded 10 million, and the progressive gameplay was fully disassembled for the first time. It explained it clearly. This is also my habit when planning a live broadcast room. I seldom focus on the details to plan a live broadcast room plan.

Under normal circumstances, once the scene viewing is opened, the data to focus on are conversion rate, transaction amount per thousand, and UV value. You can analyze most of the live broadcast rooms with problems and find that the traffic level has dropped. Is there a problem with these data?

Because these data represent how much commercial value you can create after Douyin gives you so much traffic. The conversion rate represents the advantages of your product, the transaction amount per thousand represents the efficiency of your use of traffic, and the UV value represents the overall consumption level of your live broadcast room traffic.

We can reverse this model. If you can't achieve incremental growth in these data, why would Douyin give you traffic? If you achieve these data and bring commercial value to Douyin, why wouldn't Douyin give you traffic?

If you want to do the above three data well, you have to think clearly about the role of the acceptance period

How many people can clearly explain the role of the takeover period? If you don't read my text below, can you think about it yourself? Have you figured it out?

In the progressive gameplay, I have said that as long as it is an account that uses natural recommendations, there is no such thing as user accuracy. You are not paying, and you are not getting traffic through short videos. How can natural recommendations be accurate? Natural recommendations will always be the least accurate traffic among all traffic entrances!

Especially when cold starting a new account, have you created a large number of transaction tags? If you don't have it, you dare not place large orders for the traffic-generating funds. If you place large orders, you will suffer losses exceeding your costs!

The first thing to do during the acceptance period is to wash the labels. The traffic-generating funds can only help you wash the basic labels and preference labels at most. Only the welfare funds are the best group of products for washing transaction labels, because you don’t have to lose money on the welfare funds. You make transactions through a large number of welfare funds, which in itself is a process of exchanging no money for transaction labels.

The second function is to help you filter out users who are looking for bargains. Everyone should think about this problem. Your drainage products are so low-priced, and the regular-priced products are also regular-priced. Unless your regular-priced products are really good, then what reason do you have to make a group of people who want to buy your bargains willing to buy your regular-priced products?

You don't even have to think about it. You just need to think about one thing clearly, that is, as long as it is wool, it will be wool for life. What you need to do is to get a contract and drive away the wool party that should be driven away. Don't be afraid that these people will lose money.

People who won’t buy your full-priced products will never buy your full-priced products, but people who can be sold by your products will have a much higher probability of buying your full-priced products.

The third function is to obtain conversion rate. Why can't you use full-price products to achieve conversion rate? Because full-price products are affected by prices and cannot achieve large conversion rates. Why not use drainage products? Can you afford to lose money on the goods?

Fourth, if you use traffic-generating funds to increase transaction density, it will still cause the problem of high costs. Moreover, it is difficult to attract traffic with too low prices now. As for profit funds, your conversion rate will not go up. How can you use profit funds? You can play with any transaction density you want with welfare funds, because you will not lose money.

The fifth is the UV value. The requirement for the UV value does not require you to increase the UV value indefinitely. As long as you have a welfare product with a certain price, your UV value will not be worse than that of your peers. However, with the price and transaction volume of the drainage product, you cannot achieve such a high UV value.

This is the role of the succession period. The reason why the progressive gameplay has to design this link is to allow the account to complete a soft landing during the process of natural recommendation and opening. Otherwise, if the account is opened with high traffic and high volume, it will die in a month or so.

If you really want to play it this way, play it unmanned, play super low prices for flash sales, play low-value high-density transactions to attract traffic, why go through so many live broadcast processes, are you wasting your time?

There will be another problem: what should I do if I can’t combine my products? It’s a shame. If you can’t even put together such a product structure, how can you expect to receive recommendation traffic?

I often say that you should not let TikTok adapt to you, but you should adapt to TikTok.

The same goes for the gameplay. You have to make all your elements, including the venue, anchor, product combination, etc., adapt to the gameplay, rather than letting the gameplay adapt to you.

But this doesn't mean that you can't play if you can't combine them. You can choose to create your own original combination of gameplay. Let me summarize it. If you choose to apply the gameplay on the market, then you have to make all your elements cooperate with the gameplay.

If you find that your hosting skills or goods do not match each other, then you have to think for yourself how to combine them to suit your own gameplay.

Therefore, the reasons for the drop in traffic levels are mostly due to the above problems. Many people have not visited hundreds of live broadcast rooms and do not know the complexity and closed-loop nature of process design.

But the result has already happened, what should we do next? Don't rush to determine the problem with the data. The analysis of the problem must be from the shallow to the deep. First consider whether it is caused by a shallow problem, and then judge the deep-level problem.

First, consider whether the word-of-mouth points have been lost, resulting in a drop in traffic. If so, make up for the word-of-mouth points.

Second, consider whether the broadcast time has been changed. Changing the broadcast time period will also affect the traffic.

Third, consider whether there are changes in the elements of the live broadcast room, such as product scheduling. Improper product scheduling will have a huge impact on traffic. Is the host changed or in a bad state? The more important elements include product scheduling, words of speech, host status, rhythm, scene, and delivery. Review whatever has changed.

Fourth, consider whether it is a problem with the overall market traffic. Sometimes the overall market traffic will fluctuate unstably.

Fifth, consider the violation issues in the live broadcast room and check the violation records through review.

The above five problems are easy to solve. The most difficult one is the problem of losing levels caused by data. Remember these data: conversion rate, transaction amount per thousand, and UV value.

For general adjustment strategies, we must first focus on the field view. What influences the viewing experience is the flow rate, which can be divided into the broadcast flow rate and the real-time flow rate.

The opening flow rate is affected by the overall data of the past few games, while the real-time flow rate is affected by both past data and the follow-up of the opening flow rate.

You will find that if the data of a live broadcast room spirals downward, two obvious characteristics will appear in the online graph of the live broadcast room.

First, the opening online score is no longer as high as before.

Second, the live broadcast only showed the online high and low trends, with almost no wave-like bulge afterwards.

We want to improve the audience's viewing experience again. The first thing we need to solve is the problem of extremely fast traffic at the opening. The best way to handle the extremely fast traffic at the opening is always to attract traffic + hold back the traffic + release orders.

The opening speed is highly dependent on the traffic of the square, which is accompanied by inaccurate traffic. However, the opening does not strictly assess the UV and transaction density, so what you need to design is

First, how can I adjust a traffic-generating model to be able to handle extremely fast traffic?

The second is to hold the flow. Only by holding it as long as possible can the role of taking over be brought into play.

Third, it is to place orders. Placing orders requires a rhythm, one to two, two to three, and using the drainage funds to maximize the number of users who can run into your live broadcast room 30 minutes after the opening. The purpose of doing this is to enable the algorithm to obtain the popularity of your opening live broadcast room, while retaining a large number of people.

In the following 30 minutes of streaming, the algorithm will continue to provide you with a good flow rate, and these flow rates constitute the traffic entering your live broadcast room.

So many times we find that no one enters the live broadcast room after half an hour. In most cases, the opening is not well maintained. After the opening traffic, the live broadcast room should be set to perform the live broadcast process for 30-90 minutes. The purpose of this stage is to increase the conversion rate and transaction density.

Choose good welfare products, use 5-minute sales tactics + flash sales + traffic-blocking tactics + multi-level order placement + large-scale order placement.

The 5-minute loop of sales speech can ensure that the people who keep coming in are educated by you. The flash sale is to increase the selling point of welfare products. The holding back flow routine is a technique to prolong the release of orders and continuously accumulate the number of online people. The multi-level release of orders is to avoid online gaps caused by one-time release of orders.

If you manage the traffic well for this hour, your real-time traffic rate will not be particularly bad in the next 2-3 hours.

But many times, it is difficult for us to control the audience after just one show. What the live broadcast room needs is a spiral increase.

We can assume that when the market starts to decline, there will be no traffic after half an hour of broadcasting. However, after a few increases, you will find that the market has a better audience in the first hour and a half of broadcasting, but it is not optimistic after 2 hours.

But if you broadcast a few more shows, you will find that the flow rate will not be particularly bad after 2 hours. When you can continuously extend the broadcast time, a good phenomenon will appear. In a certain period of time, your overall data will outperform your peers and you will enter the rankings. Your live broadcast room will naturally enter the live broadcast square and gain explosive live broadcast online.

You can summarize it as follows: the opening determines the flow rate 30 minutes later, and the intensive transactions and conversion rate 30 minutes later determine whether you can push the flow into a higher live broadcast square and obtain high online.

In summary, if you connect the broadcast flow rate well and make dense transactions within 30-90 minutes of the broadcast, your audience will be pulled back again. In the process of pulling back, you need to take into account the increasing trend of conversion rate and transaction amount per thousand, while ensuring that the UV value is not lower than that of your peers, and your audience will not drop too much.

Occasionally, the market data fluctuates, but it doesn’t actually have much impact.

What remains at the end is to gradually and continuously test the profit items among the welfare items. As long as the profit items explode, a large number of regular-priced goods will be sold in the live broadcast room in a short period of time, and the live broadcast room will become popular.

This is the strategy to use when there is a problem with traffic. Please note that this is not a conventional method. In a normal live broadcast path, the frequency of welfare funds will not be so long. Instead, it is necessary to continuously observe the real-time performance of the data in a gradual manner. This tests the trader's data adaptability.

It is precisely because most of the operators in the live broadcast room lack the ability to respond that the account dies midway.

Finally, let me talk about two misunderstandings that many people have when reviewing their past performance.

First, it is useless to count the data of the live broadcast room throughout the entire show. Let’s take the example of counting the audience. A live broadcast room with a higher opening flow rate may have a larger audience even if the broadcast time is shorter.

Second, when counting the retention rate, the retention rate of the opening traffic is completely different from the retention rate during the broadcast. Does it make sense to count them normally? It is meaningless, so this is why I rarely use the Traffic Square tool. It only feedbacks the normal results, but does not feedback the interim data of the live broadcast room.

However, TikTok assesses periodic data, so the best way to collect data should be the first 30 minutes, 30 minutes to 60 minutes, 60 minutes to 90 minutes, and so on.

Third, if you want to review the situation in a more detailed manner, you should review and adjust in real time. The data for the post-review has been formed and the judgment of the Douyin algorithm has been determined. What’s the point? Once the five-minute data of the live broadcast room drops, such as the click-through rate, it is right to make adjustments at any time.

I accidentally wrote more than 5,000 words. I still hope that everyone who does live broadcasts can think more deeply about things and not always stay on the issues that 80% of people can see.

Creating an account for a live broadcast room is a low-probability event in itself. If you can't see the problem in more dimensions than others, how can you do better than others?

Author: Yin Chen's live streaming

Source: Yin Chen's live streaming