Are there any differences in the traffic pools of information flow accounts in different industries?

During the delivery process, we often encounter this problem: will there be differences in traffic pools for accounts in different industries? I think there is. So for us optimizers, how to make good use of the media's effect of dividing the traffic pool can be considered from the following aspects:

1. Taking Tencent Guangdiantong as an example, the framework categories are divided into K1-consumer goods KA category, K2-online effect category, K3-life service KA category, and K4-regional license. In addition to the easy management of industry-based frameworks, different frameworks may correspond to different QPS traffic levels. We often hear that when opening an account, we hope to open a framework account, which makes sense. In addition to rebates, there may also be differences in traffic size. In this way, we can consider different frameworks to better balance advertisers' budgets.
2. Appropriately break the circle in terms of traffic, adopt different targeting to explore more appropriate traffic targeting, reduce the dead end of delivery due to excessive traffic overlap, and just keep competing with rivals in the same industry, making the conversion cost higher and higher. In the end it depends on who exits the market first.
3. Open multiple accounts. Each account may have a different initial traffic pool. In the initial launch of the plan, we can touch different traffic pools. This is the same as adding multiple plans. Although this method is relatively stupid, sometimes the simple method is the most efficient.

Many times when we are dealing with accounts, we always have many questions. One of the classic questions is, if I am dealing with accounts in two different industries at the same time, will there be any difference in the traffic pool when delivering the traffic? If you are a client optimizer, you may not be very sensitive to such issues. After all, due to business needs, the client's products are relatively simple, while the B-party optimizers are always in contact with different products, and often run many products from different industries at the same time. This often leads to the above-mentioned problem: Do traffic pools for different industries exist? This problem has actually existed for a long time, and I have often heard similar questions since I entered the industry. Most optimizers gave a consistent answer that traffic pools do exist, but after asking many people, there is still no satisfactory answer as to what traffic pools are and why they exist. Here I would like to give some of my own thoughts on whether traffic pools exist or not.

Why do we think that traffic pools exist?

First of all, this is inconsistent with the facts we have observed. On the media platform, products of all conversion prices and conversion demands exist. If we believe that the media's advertising bidding is ranked by ECPM, then products with high conversion prices will naturally have a greater advantage. This will cause the entire platform to tend to favor products with high conversion prices, causing the advertising ecology of the entire platform to decline. Although media revenue has increased in the short term, in the long run an oligopolistic market has been formed, and the media's bargaining power has decreased (here we first ignore other CVR and CTR factors to make a simplified assumption).

On the other hand, we need to consider the matching degree between traffic and advertisers. Suppose when a traffic request comes, there are only two advertisers A and B in the entire market bidding. When A's ecpm is higher than B's ecpm, advertiser A obtains the traffic, but finds that the matching degree of this traffic is low. In the next bidding, advertiser A will tend to lower ecpm to reduce waste. In the long run, this may cause advertiser A to withdraw from the market because of large or poor fluctuations in conversion. If a more stable system is needed, the media's advertising algorithm model will automatically look for similar traffic to convert, which creates a natural evolution of the traffic pool. However, if it can be divided from an artificial perspective, it can have many benefits.

In fact, traffic pools in different industries do exist. First of all, the media's revenue is calculated based on eCPM. Specifically, each traffic request is calculated based on eCPM for each traffic. This is a basic division. If we consider it based on the idea of ​​classification, the eCPM of different industries is different. Therefore, based on the starting point of maximizing revenue, the media will make a rule to divide the traffic pools for different industries. As can be seen from Figure 1, the industries that mainly received information flow advertising in 2020 are culture and entertainment, games, etc. For the media, the profits from these industries are obviously greater, so there is a possibility of setting up certain traffic pools for these industries to allow these industries to obtain more traffic.

Figure 1 Comparison of advertising volume in key industries in 2020

The second aspect is that under the current rules of the ocpx algorithm, it is necessary to build and predict models based on historical data. Dividing the traffic into different traffic pools in advance can reduce the number of candidate advertising creatives, which can lay the foundation for subsequent sorting of advertising creatives and reduce the computational workload of the entire system. This is considered from the perspective of rules and is also conducive to reducing the computational pressure of the advertising system.

The third aspect is that due to the needs of each advertiser's own business, advertisers in different industries actually face similar or highly overlapping groups of people. This leads to advertisers only screening the groups they need when conducting targeting. The traffic screening based on this leads to the media being motivated to divide the traffic pools into different industries.

From Figure 2, the industry scale and ecpm of advertisers do have an impact on the media's division of traffic pools. The scale of advertisers and ecpm are positively correlated with the size of the traffic pool. The larger the ecpm, the larger the traffic pool the media is willing to define; the larger the industry scale of the advertiser, the larger the traffic pool the media is willing to give. Just like the E and F industries in Figure 2, the scale and ecpm of industry E are greater than those of industry F, so its circle is larger. In addition, the audiences of advertisers A and B in industry E overlap, indicating that they are in competition for traffic, which will push the overall ecpm higher and higher.

Figure 2 The impact of advertiser industry scale and ecpm on media traffic pool

As for the concept of traffic pool, I have only given a brief explanation here. I personally believe that there are different ways to divide the traffic pool in different situations such as general and targeted, CPC and OCPX, but I will not go into details one by one.

Therefore, from the perspective of the media, the media is motivated to divide the traffic pools of different industries, but whether the media has the ability to divide the traffic pools of different industries is another question. This may involve the media’s own traffic scale, algorithm capabilities, and the degree of perfection of the advertising mechanism.

Author: Enthusiastic Aji

Source: Advertisement placement tutorial class