Advertising strategies for the Internet automobile industry (Part 3)

Whether it is a user product or a commercial product, traffic distribution is always an unavoidable topic. But the prerequisite for good traffic distribution is that we must have sufficient understanding of our own traffic.

Advertising strategies for the Internet automotive industry (I)

Advertising strategies for the Internet automotive industry (Part 2)

1. Panoramic understanding of commercial traffic

What is panoramic awareness of commercial traffic? Many junior commercial product managers may simply understand traffic based on the PV and UV of the page. Such understanding is not only superficial but also exposes the fact that they are laymen or rookies. In fact, to have a panoramic understanding of your own business traffic, you need to answer the following five questions:

1. How big is the traffic plate?

The answer to this question is that in addition to daily UV and PV, the commercial advertising field has its own statistical indicators for traffic: the average daily number of advertising requests and exposures. How big your own traffic plate is, that is, how many billions of ad requests you have every day, and how much traffic you have available for sale, that is, how many billions of impressions you have every day. Sometimes, indicators such as requested UV, exposure UV, visible exposure, and visible exposure UV are also referred to to accurately measure how large the traffic plate is.

2. The proportion of stock and increment in the traffic plate

In addition to the most basic understanding of commercial traffic size, what needs to be considered further is the ratio of existing traffic to incremental traffic.

This is an indicator that is often overlooked, but in fact, when the proportion of stock is much greater than the incremental proportion, the performance advertisements settled according to CPL and CPS will be significantly affected.

Especially in automobile industry advertisements, a user is willing to leave information for an average of 4-8 car models before buying a car. After buying a car, the user value drops significantly. If there is no sufficient and stable new traffic to supplement it, the effect promised by performance-based advertisements cannot be guaranteed.

3. What is the quality of the traffic?

Understanding the quality of your own traffic can be said to be the most important step in the process of understanding commercial traffic.

The indicators for measuring traffic quality are similar to those for C-end products, and are reflected through indicators such as the average number of page views in a session, bounce rate, and average session duration.

However, since the quality of traffic varies, it is difficult to judge the quality of traffic by simply looking at the overall average. Therefore, it is necessary to stratify the traffic. For example, first divide it into on-site and off-site, then divide the on-site into different locations or functional modules, and then divide the off-site into traffic source channels. This will provide a detailed and clear insight into the overall traffic quality.

4. What is the value of traffic?

Since the essence of commercial advertising is to monetize traffic, in addition to being aware of the quality of traffic, we also need to be aware of the value of traffic. The indicators for measuring the value of traffic are: visible exposure, CTR (click-through rate), CVR (conversion rate), etc. These three indicators correspond to the three most common sales models in commercial advertising: CPM, CPC, and CPL/CPS, so as long as you know the values ​​of these indicators, you can roughly calculate how much revenue these traffic can bring.

5. Is the current traffic distribution reasonable?

The last step in gaining a panoramic understanding of commercial traffic is to make a reasonable assessment of the current traffic distribution. However, the conclusion is not simply to draw a reasonable or unreasonable conclusion, but to find problems in the existing distribution method and improve efficiency.

There are usually two major categories of traffic distribution problems: the first is the uneven distribution of traffic among product lines within the platform . For example, the most typical situation is that product lines with low monetization efficiency occupy more traffic. The second type is that the traffic allocated to advertisers by the platform is uneven . The most common situation is that the traffic is concentrated in the hands of the top customers, and the budgets of the middle and tail customers cannot be consumed. In short, various problems will arise in the process of traffic distribution.

Through the above five questions, we can have a comprehensive and three-dimensional understanding of our own traffic. The purpose of such a comprehensive understanding is to discover the problems existing in the traffic and solve them with a systematic strategy.

2. Common Strategies in Traffic Distribution

As the saying goes, "What is worrying is not the lack of resources but the inequality of resources." Once it comes to the allocation of resources, there will be a lot of problems. Since the distribution of traffic among product lines within a media platform is often the result of management bargaining and needs to be discussed on a case-by-case basis, we will focus our discussion here on the issue of traffic distribution from the platform to its customers. Here is a summary of some common problems and corresponding strategies:

Problem 1: There is a lack of top-level traffic regulation logic, and traffic is allocated solely through bidding among brands/car series, resulting in ad slots being fixed in the top brands/car series. This is not conducive to budget consumption by mid-tier/tail-end advertisers, and is not conducive to activating healthy competition in the market.

1. Response strategy: Level market traffic allocation strategy

There are many ideas for building top-level traffic control logic. Here, based on the characteristics of the automotive industry, traffic can be regulated according to market levels. According to my country's definition of the automobile market, it can be divided into six levels: micro, small, compact, medium, medium-large, and large. If the popular categories that users are most concerned about nowadays, such as new energy, SUV, MPV, sports car, etc., are separately proposed, a complete classification system can be formed.

Based on the classification system of the level market, we can design a traffic allocation strategy. The calculation formula of the traffic allocation weight is as follows:

Traffic distribution weight = a * number of car models in the same market level + b * traffic in the same market level + c * user attention in the same market level

Among them, a, b, and c represent the weight values ​​of each factor respectively. We do not need to care about the specific weight values, which can be handled by the algorithm students. Let's focus on why these three factors were chosen, as shown in the following figure:

Because our goal is to reasonably distribute traffic in the dimension of level market, the size of the level market is a key factor to consider. Usually, the market size is described in units of money, but it is not suitable in the current business scenario. Therefore, we use the traffic in the level market to reflect the market size, and the number of car series in the level market to reflect the intensity of competition among the levels of the market and the maximum number of customers we may face (the maximum number of car series deployed in parallel).

Finally, during the data analysis process, we discovered a counterintuitive phenomenon. For example, the user attention to SUVs, which is the market segment with the largest traffic, is not as high as that to compact and new energy vehicles. The divergence between traffic and user attention also proves that there are problems in the current traffic distribution of C-end products, so the factor of user attention is also taken into account in the model.

Provide the factors to be considered in the business as logically as possible to the algorithm students. At this step, the work of the advertising strategy product manager is completed. What features to add next and how to adjust the parameters are issues that the algorithm students need to consider.

2. Strategy Implementation

In addition to providing business factors, advertising strategy product managers also need to clearly understand how their strategies are implemented.

The implementation of the market traffic allocation strategy at this level is completed through the sorting module in the advertising delivery system. The sorting module is usually divided into three stages: rough sorting, fine sorting, and re-sorting. Business strategies that are frequently adjusted are usually implemented in the re-sorting stage because re-sorting is an optional stage. If problems arise with business strategies during the re-sorting stage, the strategies can be taken offline immediately without affecting the normal operation of the entire advertising delivery system.

The specific implementation steps are shown in the figure below:

The implementation principle is very simple. After the refined ranking results come out, a real-time/offline data table is introduced. The data table records the weight values ​​that need to be increased for each level of the advertisement. When calculating the ranking in the re-ranking stage, the weight values ​​can be included in the calculation. The final ranking result is the display result of the advertisement.

The advantage of adopting this implementation method is that algorithm researchers only need to maintain a real-time/offline data table to implement the strategy. The online and offline operation of the strategy is decoupled from other functional modules and will not affect the overall operation of the system or other strategies. In addition, it is particularly important to emphasize that when implementing the strategy, you must not arbitrarily change the calculation logic of other sorting stages, otherwise it may cause terrible consequences, so be careful!

3. Evaluation of Strategy Effectiveness

Since strategies aim to achieve different goals, there is no unified indicator for evaluating their effectiveness. However, strategies are generated because of problems, and as long as the problems are effectively solved by the strategies, then it should be considered a good strategy.

It is not easy to measure the effectiveness of the level market traffic allocation strategy introduced this time. First of all, the top-level traffic control logic is in place, but it is still necessary to observe whether the problem of core advertising positions being solidified by top brand car series has been alleviated. The judgment criteria for alleviation need to be evaluated in combination with the budget consumption of various types of customers. In short, the effect of this strategy is not intuitive, and we need to continue to observe and optimize it together with algorithm students.

Finally, it is important to note that there is a high possibility that the strategy that has just been launched will not be effective or obvious, and continuous optimization is needed to make the effect apparent. However, the work of effect optimization is not just the job of algorithm students. Advertising strategy product managers must also participate in providing business logic and improving the interpretability of the model.

4. Strategy Generalization

Since the level market is a classification method unique to the automotive industry, and the vast majority of readers are advertising practitioners outside the automotive industry, let’s take a look at the issue of strategy generalization. In fact, if you look closely at the process of strategy construction, you will find that there is a fixed routine for solving this type of macro allocation problem:

Macro allocation problem—introducing classification method—screening important factors—building algorithm model—online testing

Therefore, if you are on a comprehensive media platform and need to regulate the traffic of various industries, you can also copy an effective distribution strategy by following the same pattern.

To sum up, as the first part of the traffic allocation strategy, this article first introduces how to fully and three-dimensionally understand your own traffic. Only based on a full understanding of your own traffic can you discover the problems that exist. Next, we introduced the first traffic allocation strategy: the level market traffic allocation strategy. In the next issue, we will introduce two more common traffic allocation strategies: KA customer traffic inclination strategy and new customer traffic support strategy . Stay tuned!

Author: Everything needs Jingsheng

Source: Everything needs Jingsheng