Tips for optimizing information flow advertising promotion!

I suddenly realized that information flow may be different from all the bidding optimizations we have had in the past, especially from search bidding. There must be a very profound logic behind this. If we understand this logic, we may get the key to open the door to information flow advertising optimization.

This article does not talk about techniques, but only about logic.

Therefore, although this is not rigorous, the bidding for information flow ads is to a large extent a game with "machines".

In essence, information flow advertising is a closed RTB system, rather than a system ranked by keywords, which means it is destined to conform to the machine's algorithm and meet the machine's optimization. Rather than just considering price competition with other advertisers. To some extent, while price competition is still important, the methods and operations are quite different.

This is the core issue of optimizing information flow advertising .

In fact, for all RTB types of optimization, putting machine optimization first is the right way of thinking.

This is because human labels are too rough, too chaotic, and too imprecise, and human behavior is erratic and interests are changeable, and labels are difficult to respond and update in a timely manner.

Therefore, performance-based delivery must use machine learning, especially supervised learning methods.

Simply put, the machine constantly tries various methods to "fully" help you achieve conversion. If the machine decides to display to a certain person, but the attempt is unsuccessful (for example, the wrong person is chosen, or the delivery location is not good), it will change the person to display to until a successful conversion occurs. The machine then summarizes the choices and delivery methods of these successful and unsuccessful people, and thus "explores" a delivery strategy that it believes to be the most appropriate.

Or, you can describe this process with a simple noun: trial and error.

Since it is trial and error, we need to give the machine time and opportunities. "Giving machines time and opportunities" constitutes the core logic of information flow optimization.

4. In the initial stage of the campaign, do not do too detailed audience targeting.

5. Even if you don’t plan to spend a lot of money in a short period of time, you should increase your budget setting. It is even better to set it to the level of “no limit on the amount of investment”. When the money is almost spent, turn it off manually.

“Governing a big country is like cooking a small fish”, and the treatment of information flow accounts is also similar.

Other strategies, such as increasing prices in the early stages of a campaign, increasing the budget to “unlimited,” and not doing too much targeting in the early stages, are essentially strategic thinking that “gives machines a chance.”

Because machines need to use trial and error, it not only takes time, but also requires constant "trying". If the targeting is too specific, the price is too low, or the budget is too limited, this will limit the machine’s tendency to try. If the machine does not make enough attempts, it will be difficult to reach a relatively ideal strategy range, and thus either it will not give you enough traffic and conversions, or the prices will be outrageous.

The probability of investing in good information flow with only a few thousand dollars a month is actually very small, and this is the reason.

As for information flow, as mentioned before, the goal of optimization is to enable the machine to better understand you. Therefore, it is not about bidding with other accounts and focusing on CTR, but about conversion.

The reason is self-evident, because machine optimization of information flow is supervised learning, and the basis of its "supervision" is transformation.

Therefore, don’t think that if the CTR of one of your ads is particularly good, the performance of this ad will be good; and if the CTR of another ad is not good, don’t rush to remove this ad.

Don’t do this, everything still depends on conversion in the end.

For the purpose of conversion, deliberately using borderline elements (sex and money) to entice customers may not necessarily work (although it still works for some industries, such as games). This can greatly increase the click-through rate, but it may not have a positive impact on the final conversion.

In order to achieve conversion, the design of the landing page of the information flow is different from that of the search:

Search bidding, because it is people who actively have needs who search for keywords to reach your landing page, so the amount of information on the landing page should be larger, more detailed, and the emphasis should be on "guidance".

As for information flow, people don’t actively search for it. Instead, they accidentally discover your ad while reading various “gossips” and then realize that they may have a need, or they just click in out of curiosity. Landing pages are not suitable for heavy-handedness at the beginning, but should focus on arousing interest and demand, and then quickly triggering users’ active inquiries. Therefore, the emphasis is on "lure".

There is only one word difference between "guide" and "lure", and the design of the landing page will be significantly different.

In this sense, not optimizing is even better than optimizing frequently.

The correct approach is to create an environment that allows the machine to try things out, or even to deliberately “trick” it (such as intentionally increasing the budget to infinity) so that it can more actively try and fail. Only in this way can your account quickly reach a more ideal state.