Summary of issues regarding live broadcast advertising

A:

This is an excellent case study that sorts out the problem.

1. Why is the payment rate so low when the previous link data is so good?

2. Is Tongtou accurate and intelligent? What is the payment logic behind it?

3. For situations like low exhibition costs, stable and continuous high audience, and low transaction rate, what can be done in terms of advertising?

Question 1 answer:

In fact, the actual reason has been discovered: a large number of people who do not need the commodities have emerged.

The elderly do not even have the payment function, but may observe that the order rate is very good, with many orders created but few payments. The payment rate is low, and it has been observed that most of the people are elderly. It can be judged that the traffic is not accurate.

And the cost of development is low. Generally, the cost of development for this group of people is relatively low, such as the middle-aged and elderly people in remote areas. The conclusion is that the population is not accurate

Question 2 answer:

Why did I choose Tongtou? It should be smart enough. Why would there be problems with inaccurate crowds? Douyin’s paid advertising is in the form of ocpm, which optimizes thousand impressions to pay for traffic. How to understand this optimization of thousand impressions?

I went out to send out survey reports today. I sent out 100 copies and found that 80 of them were filled out by women. In order to improve the efficiency of sending out survey reports tomorrow, I need to optimize my behavior tomorrow and specifically select women to communicate with to improve the efficiency of my behavior. OCPM payment is a similar process.

In the case of full-scale investment, which groups of people will be exposed in the future will be determined by who is more willing to enter the live broadcast room, stay, click on the products and realize purchase conversion. Which group of people can complete this link better will get continuous exposure.

You may wonder, why do we continue to expose this group of people when there is obviously no conversion? Because we don’t just look at the conversions, but actually look at the group with a high click-through conversion rate.

The ocmp paid advertising bidding means that the system can guarantee that the exposure obtained within your bidding range will achieve a conversion. As long as it can be achieved, it will continue to be consumed. It happens that this part of the population has few competitors and a large base, and they can have a high exposure by spending a little money. The system defaults to being able to achieve this goal, so in the case of full bidding, it is possible to continue to expose to this part of users.

Question 3 answer:

For such a situation, you just need to calibrate the direction of the account. I think many people are wondering why others can invest in general while I have the same situation.

This is related to too many factors. It may be that the people who were displayed and clicked in the early stage happened to be this group of people. It may also be that the bid is too low. In order to complete the task, the system thinks that only this group of people can achieve the goal within the targeting you selected. Other targeting groups may not be able to compete with others based on the current bid.

There are many reasons. Generally, we don’t delve into the reasons, but only ask what to do. The answer is that the flow test needs to continuously correct the crowd. The more accurate the circle can be, the more accurate it can be. Let the crowd that fits the product portrait continue to behave in the live broadcast room. The label of the person will label the live broadcast room, and things will gradually get better later~

【Q2】

Why does interest take precedence over behavior for products with low repurchase rates? Do you understand that the repurchase rate is low, so the willingness of people who have bought before is low, so you should give priority to those who are interested

A: You can understand it that way. When the behavior is locked to e-commerce behavior and the corresponding category or word is selected, the users who have historically performed related behaviors in the Douyin system are circled. They have a low repurchase rate and already have this behavior, so the corresponding purchase probability will of course be lower.

At this time, you should find those people who continue to pay attention to the corresponding product information, are interested in the corresponding products but have not yet made a purchasing behavior, so that interest is better than behavior.

However, this cannot be evaluated based on the average order value, but rather, as the question suggests, whether there is a high repurchase rate. For example, for luxury goods, the average order value is very high, but once purchased, customers will continue to purchase again. For example, when it comes to sweeping robots, it’s rare that people say they’ve already bought one and will buy another.

【Q3】

Hand selection: Interest takes precedence over behavior, how to do it? Is it that the number of people covered by interests is greater than the number of people covered by behaviors? Or are there any other operating techniques?

A: This is actually not that mysterious. You can choose fewer keywords for behavior. Taking the sweeping machine as an example, we can directly select the household appliances category. There is no need to select the sweeping machine keyword. However, in the interest column, you can choose sweeping machine, or even directly check the expansion word.

In terms of behavioral choices, except for sweeping machines, there are no expanded words that are clearly related to sweeping machines, such as clean and cleaning products, which can still be selected. When we talk about low-repurchase products, interest takes precedence over behavior. Under a single product name, we attract interest rather than behavior.

【Q4】

So the question is, how to find the right people?

A: There are many channels. Of course, to draw conclusions, we need to conduct online tests and get actual data.

1. Third-party platform to check the audience and transaction crowd of competitors’ live broadcast rooms

2. Official platform, Qiance on the delivery side, and product data on the store side

3. E-commerce platform data of other platforms for corresponding categories. For example, the eight major consumer groups of Qian Ce actually come from Taobao, which can be used as a reference

In addition to these, there are many clues, such as user insights. I observe myself to see whether there are more boys or girls, more young people or more old people; for example, the product manager may think about what group of people this product was originally designed for; no matter what channel the information is obtained, it can be tested online and conclusions can be drawn.

First determine whether there are differences in the basic population, that is, gender and age, and then choose behavior and interests. At this stage, manual selection is recommended. For products with high repurchase rates, behavior takes precedence over interest, and for products with low repurchase rates, interest takes precedence over behavior.

【Q5】

How much material is currently required for a live broadcast or how to determine it?

A:

What is the basis for judging the amount of materials needed for a live broadcast? From several aspects

1. The GMV target of the live broadcast room (usually determined based on the number of historical sessions)

2. How many views have been attracted by historical short videos? How much should I amplify it based on my goals?

3. How many people can each short video attract on average in history, and what is the explosive rate of short videos?

4. What is the overall advertising budget?

For example, my GMV target for historical sessions is 1 million. In the past, my direct screen projection consumed 50,000, which was enough to achieve the GMV target for the session, the book consumption and ROI targets. Paid traffic accounted for 50%, and 5 short videos were released at the start of the broadcast. Short videos entering the live broadcast room accounted for 10%.

My GMV target for this event is to reach 3 million. It is known that when the amount consumed by direct screen investment is greater than 50,000, the ROI will drop. The remaining amount needs to be supplemented by the promotion of short videos. By default, three times the current number of people is required to achieve my three-fold GMV target. The rest of the traffic will come through short videos.

Then we can calculate the average number of people attracted by each of the five short videos, and finally calculate the number of short videos that need to be added to achieve this goal.

The actual effect may be better than the calculated one, or worse than the estimate. The more historical data there is and the more stabilizing factors there are in the live broadcast room, the more accurate the estimate should be.