Why do we need to build an Apple bidding intelligent delivery platform?

It was somewhat difficult to adapt to the transition from the advertising industry to the client industry. The environment here is not as simple as that in an advertising company. There is some gap, but I was more shocked and moved. It is said that Internet companies change quickly. In order to adapt to market changes, some companies’ products may be killed in less than half a year, and even the company’s strategies have to be changed several times. Sometimes when a market demand emerges, the opportunity is fleeting. In this circle, most people constantly change themselves to cater to the rules of the game in this world, while others try to create standards and rules at all costs.

In 2016, the most talked about topic in the mobile Internet industry was the second half. The rules of the game in the second half have changed, and Apple's application distribution is changing. When chatting with CPs these days, they all said that it is becoming increasingly difficult to manipulate the ASO rankings . Apple's rules change so quickly that people are at a loss and complain a lot. In China, when people heard that Apple’s bidding ads were about to be launched, many companies started calling for services to run the ads on their behalf. Some companies have a large user base and a loud voice, which does give them an advantage in this game. However, if everyone studies ASM based on their existing ASO operation experience, they will soon become homogenized, and the effect may not be the best. Going with the flow is certainly not what we want.

After Apple's bidding ads were launched in North America, we began testing them. Based on our own big data advantages and ASO foundation, we used big data to study algorithms, research intelligent delivery, and find core points. During this process, Sister Xi suddenly understood what dividends are. It is not just about grabbing traffic or taking advantage of the bonuses. To stand out from the bonuses, one must devote oneself to research and dare to stick to one's core propositions. One must not blindly cater to the world. One may choose a more difficult path at all costs, but in the end, one can develop a truly valuable product with a kind of paranoia and ingenuity.

With a paranoid personality , we do things that others have never thought of or done , making our products unique and full of charm.

One particularly prominent feature of this team is quarreling. For example, at the beginning of development, some people felt that artificial intelligence was useless, and then the product would jump out to argue. Most of the time, in terms of products and operations , everyone insists on what they think is right and defends it with all their might. This process is actually very painful.

In the end, we constantly worked out the differences and conducted continuous research on intelligent delivery, and even brought in all the big data engineers from the marketing data service to study the algorithms together. There were constant conflicts during this period, but everyone poured their emotions into the product, and slowly, some important things emerged.

For example, the concept of one-click thousands of groups of intelligent delivery (this function will appear later). When submitting, a dialog box will pop up, asking whether to generate an intelligent delivery series group and the number of strategy groups to generate. The intelligent delivery function is needed not only by CPs but also by agencies. When operating dozens or hundreds of keywords, manual work can be done, but if there are more than 2,000 keyword strategy groups, believe me, manual work is time-consuming and inefficient, and the cost investment is not cost-effective. Smart delivery can greatly save time and manpower costs.

The paranoid personality of products and operations, and the insistence on intelligent delivery, are ultimately what allows this platform to exude its technological charm. To quote a sentence from the picture below: The truth that most people ignore, the truth that seems distant and unbelievable is ultimately where real value is created.

The spirit of craftsmanship , the depiction of details and the grasp of the essence are achieved to the extreme

Before the launch, the product team proposed the concept of granularity, which was essentially an infinite refinement of targeting. Later in the campaign, it was found that Apple would not advertise to fewer than 5,000 people under the targeted strategy. It was also found that after splitting the campaign into countless targeted groups, the actual results were not as expected. After repeated research, we tried hard to think about the core key point.

At the same time, we had in-depth discussions with multiple CPs and finally clarified some core needs. For example, when doing advertising, whether it is ASO or ASM, keywords are the core. Many people fail in word selection. If there are only 50 words, that is obviously not enough. In other words, your cognitive radius is limited to these 50 words. There are more words that may be very efficient, but you have no chance to know them, and you lose at the beginning.

The WNA system (words network analysis) is based on this and continues to expand more words, such as 2,000 words. This can break through the boundaries of human cognition and use machines and big data to tell you more words that can generate click conversions. These words cannot be expanded according to existing third-party word expansion tools and human cognition. These words may not be highly relevant to the APP, but the cost of acquiring customers is very low.

Smart delivery is reflected in word expansion by opening the door to see what the relevance of various words is to the APP in different dimensions, and what the relationship between relevance and price is? When the keyword opening is wide enough, there will be more opportunities for judgment and selection in the end.

With the keywords, the next thing CP is most concerned about is bidding. According to the traditional manual method, they often bid from the lowest price to the highest price, or try blindly on impulse. The time cost, learning cost and trial and error cost are relatively high. The operating mechanism of Apple search ads is not that you can get impressions by bidding high or low. For example, CP should bid 1 yuan to get the desired impression, but often do not know this value. If it is less than 1 yuan, it will not be displayed, and if it is higher than 1 yuan, the money will be wasted.

Therefore, our team gradually figured out some rules in multiple rounds of delivery and worked out a function system (ASM delivery calculator) that can predict bids based on the APP keywords given by the CP and the desired display volume. The principle is to conduct deep learning of historical sample data, build models through big data technology, and finally simulate the curve and trajectory of CPA and display volume to predict the CPA and display volume of a certain keyword tomorrow. Simplify the advertising process and make bidding and results predictable. Of course, this is mainly based on the CP's own delivery data, not on other people's data, and the results of the two are completely different.

Being able to specify keywords more specifically and predict bids demonstrates the ultimate capabilities of machine deep learning and artificial intelligence, while also reflecting the product manager's grasp of core details. The spirit of craftsmanship is not limited to handicrafts. In this era, attention resources are scarce. Any product must be polished to its own unique extreme point in order to penetrate into the hearts of users. Any perfunctory product will not survive for long.

We must go all out and not deviate from the belief that we should do our best to stick to.

Once upon a time, we were proud of Apple because Steve Jobs was determined to make Apple into a great product, but today, no one takes it as their responsibility to make great products. Although Apple’s mobile phone sales seem to be getting better and better today, the core of Apple is gone.

An engineer who had worked with Apple for 17 years wrote in a long blog post that this was the first time Apple did not go all out, and in many ways, Apple had deviated from its previous creed of doing its best. For the first time in its life, it was not in sync with itself. For example, Apple no longer understands its users, repeatedly misjudges user expectations, and its products are slow to update. The Apple that once pursued, focusing on key users and paying attention to details, and perfect design, no longer exists. Without craftsmanship, the reputation and influence that Apple has worked so hard to build is gradually weakening, although the sales data is still good.

If you can go all out, stick to the most core and important creeds, and create attractive and valuable products, users will definitely follow you closely, but the execution and persistence of details here are the most difficult to achieve. I hope we can keep doing it well.

Mobile application product promotion service: APP promotion service Qinggua Media advertising

The author of this article @溪姐 is compiled and published by (Qinggua Media). Please indicate the author information and source when reprinting!