Recently, I saw an optimizer complain in a group: "I think users are all hooligans!"
Because many users clicked on my ideas and even downloaded my APP, but they just didn’t activate, register or pay! This results in a sharp drop in ROI.
Isn’t it infuriating? !
Don’t panic. If you encounter this situation, you can use the “ double bidding method”. It can optimize deep costs while ensuring shallow costs.
Next, let’s talk in detail about the principles, usage scenarios, and operating skills of double bidding.
01 The principle of double bidding
The principle of double bidding is very simple. Advertisers set the target conversion bid and the paying user (deep conversion) bid at the same time. The difference between the two is the target paying rate. For details, please see the official explanation below: On the premise of ensuring the basic conversion cost, the system will automatically prioritize users with the highest estimated payment probability among a large number of users to recommend the advertiser's ads, so that the final payment rate approaches the target payment rate. First, the system will mine the user characteristics, advertising characteristics and media characteristics of the samples based on the advertiser's activation data (as negative samples) and payment data (as positive samples), with the help of the huge amount of data and label system accumulated by the Big Data Engine, and conduct cross-learning based on these three characteristics. The more sufficient the "learning materials" given in the early stage, the more the model understands you and the better the advertising effect. Based on the preliminary learning results, the model will estimate in real time the possibility of users paying advertisers during the delivery process, and screen out users who are more likely to pay for display. After the launch, the model will quickly correct and upgrade based on the acquired launch data. This means that the more launch data given to the model in the early stage of the launch, the faster the model can learn and the more accurate the judgment will be. Finally, the model completes learning and enters a stable period. It can help advertisers obtain traffic that meets payment requirements in a stable and fast manner within different advertising styles, models, and material ranges. Advertisers can increase their budgets with the help of the model, explore gameplay, and tap more high-quality users.
02 Special indicators for APP promotion
Because App promotion is different from our usual advertising evaluation indicators, it requires very in-depth evaluation indicators. Downloads alone are not enough, but also activation, how many people retain the app the next day, subsequent payment status, etc.Downloads: How many people downloaded the app;Activation: How many people registered and activated after downloading;Key behavior: Determined by the attributes of the app. For a game app, the key attribute could be whether to create a game character or the level of the game.Second retention: retention on the second day, whether the user continues to log in the next day;Payment: whether payment behavior occurs within the app; These five indicators can all be used as conversion goals, but with different levels of depth. The front-end bids for successful downloads through Toutiao, Douyin, Baidu and other platforms, and the back-end bids for deep conversions, while optimizing the conversion costs of both the front-end and back-end, which is called double bidding.Conversion target bid/depth conversion target bid = target depth conversion rate For example, when we were running a promotion in the GuangDianTong backend, we set two goals as oCPM bidding decision factors: optimization goal: activation 16 yuan; deep optimization goal: first payment (i.e. deep conversion) 300 yuan. According to the formula, the target depth conversion rate is 5.3%, that is, for every 100 activated users, 5.3 users will pay.
03 Double bidding usage scenarios
1. Focus on the deep conversion effect , but the deep data is relatively sparse and the return flow is slow. If it is directly used as an optimization target, it will not be able to increase the volume and the cost will fluctuate greatly. In this case, double bidding can be used. For example: if you want to retain customers, get them to pay, and place orders, but the direct delivery effect does not meet your expectations, you can use double bidding.2. If you are concerned about deep conversion effects, but also shallow effects , you want both activation and payment, but want a certain ratio between the two, you can use double bidding. For example: 01. When game customers optimize payment, under the premise of stable payment costs, the activation costs also need to be controlled within a certain range of the bid to reduce the risk of not meeting ROI targets. 02. When optimizing the next-day retention of online service, ecosystem, and casual game customers, the activation cost must also be consistent with the bid in order to meet the overall assessment requirements. If the activation cost bid is too low, then the volume of deep conversions will not be enough.
04 Double bidding placement skills
1. The deep conversion bid must be higher than the shallow target conversion bid;2. The bid adjustment frequency should not exceed once a day, and the adjustment range should be less than 10%;3. The activation bid and deep conversion bid need to be adjusted by the same amount, for example, increase or decrease by 5% at the same time; Generally, a ratio will be fixed. For example, if the shallow bid is 10 yuan and the deep bid is 100 yuan, then the goal is to have one paying user among 10 activated users. If you want to raise the price, you must also fix the ratio of 1:10. If the shallow bid is 20, the deep bid will need to be 200 yuan.4. When the deep conversion rate meets the target but the volume is insufficient, the activation bid and deep conversion bid can be appropriately increased in equal proportion; For example, if the current overall ratio is normal, then we can increase both bids to grab more volume.5. When the deep conversion rate meets the target, but the deep conversion cost is higher than the acceptable cost, the activation bid and deep conversion bid can be appropriately lowered in equal proportion; (Note: it will affect the running volume)6. If the activation cost meets the target and the deep conversion rate is stable, and you want to increase the deep conversion rate, you can keep the activation bid unchanged and lower the deep conversion bid until the deep conversion rate meets expectations; (Note: This will affect the overall volume) If you want a high activation cost, you can lower the activation bid. If you want a high deep conversion cost, you can lower the deep conversion bid. If you meet the target, there is no need to make too many adjustments.7. If the deep conversion rate continues to fail to meet the standard, the model will limit the running volume. The target deep conversion rate can be appropriately lowered to give the model more time to learn. According to historical experience, it will take about a week of learning to reach the standard and increase the volume. What does it mean to appropriately lower the target depth conversion rate? For example, if the previous target deep conversion rate is 10%, it can be reduced to 5%, and the previous deep bid of 100 yuan can be increased to 200 yuan.
05 Other similar scenes
In addition to APP promotion, it is also applicable to the education industry, as shown below:The goal is transformed into "successful form submission", which is deeply transformed into "customer customization", that is, payment behavior. Many education industries have started to use payment forms. For example, there is a “9 yuan course” where you first submit the form to start choosing courses and class times, and then pay. There are even cases where you don’t even have to submit a form and can just pay directly. This is also similar to double bidding. There may be more double bids in the future because the cost of traffic is too high now. You will find that the education industry is increasingly using paid and deep conversion goals because they want higher-intent customers, that is, paying users. For ordinary forms, the manual tracking cost may be relatively high. Now that paying users are directly screened out, the conversion cost will increase exponentially. For example, the “9 yuan course” and “49 yuan course” in the K12 education industry may cost up to 600 to 700 yuan. What supports the high cost is the subsequent large amount of payment. From elementary school, junior high school to high school, students may continue to pay for learning.So the question is, will your current industry invest in deeper transformations in the future? Why are there deeper conversions in industries like APP promotion and novels? Because users will have a continuous substitution process, even if the user only pays 1 yuan at the beginning, he may spend 9 yuan, 10 yuan or even hundreds of yuan to continue reading or playing.The type of industry that double bidding targets is the industry that can guide users to pay continuously. If you are just a one-time deal and don’t have a product to encourage users to continue, then you may always use the form submission conversion method and will not touch double bidding.
![Li Xingxing's practical editing training camp will end in January 2022 [HD quality with materials]](/upload/images/67cc10da025be.webp)
![Black Hat - Group Traffic Tik Tok Gameplay, can get 2 million group traffic in a year [paid article]](/upload/images/67cc031fb3061.webp)
![[Overseas Ukrainian Beauty Video Blessings] Short video production of foreign beauties holding signs, orders are placed all year round!](/upload/images/67cc5399e8711.webp)
