Does increasing the volume of information flow advertising really require piling up a large number of accounts and plans?

In the information flow advertising industry, who is responsible for the advertising effect? The first thing that comes to mind is definitely the optimizer! Because they are the most direct operators of advertising, the advertising effects produced by optimizers of different levels are also very different. The difference in the level of optimizers is reflected in the methods and approaches they use in their daily advertising work.

These daily tasks basically involve setting up accounts, making plans, thinking of materials, reviewing ideas, monitoring data, adjusting strategies, making reports, etc. They are complicated and a bit boring.

1. Daily work of information flow advertising optimizer

Only information flow advertising optimizers know how hard their lives are!

Many newbies who want to enter the industry, or other positions in the information flow advertising industry, are actually quite curious about how these optimizers work every day:

-> View data

The first thing I do when I get to work every day is to look at the data to see the consumption on the previous day, the conversion volume, whether the cost is normal, and whether the budget and balance are sufficient. Let’s take a look at how far today’s data is from achieving the KPI…

Some optimizers may take out their mobile phones or computers to check the data as soon as they wake up, as if they are anxious to see how much money they have earned today... Every day is full of hope or disappointment.

-> Daily News

Every morning or every evening, you need to download the delivery data and organize it into a daily report, which should also analyze the delivery situation and the next delivery strategy adjustment plan.

If you are a Party A optimizer, you should report to your leader; if you are a Party B optimizer or a media direct optimizer, you should report to Party A.

Sometimes when a project is being established, a project requires a large investment, or a project encounters difficult problems, meetings and communications between Party A and Party B or across departments are required.

-> New Arrivals

In order to ensure the scale and effectiveness of the delivery, it is generally necessary to set up multiple accounts, and new plans and new materials must be continuously updated every day. Some companies also evaluate the number of new plans that optimizers submit every day.

However, creating new material does not mean that it can be released. It still needs to be submitted to the platform for review, and then it can be released only after the review is passed. The review process is sometimes fast and sometimes slow.

If the review is rejected, you have to revise it again and submit it for review again, which may take a whole day to complete. Therefore, it usually has to be reviewed one or two days before the release to ensure that there is material available on the day of release.

-> Monitor account

Check whether the cost can be further reduced and the volume can be further increased, and always pay attention to whether the data is abnormal (such as a sudden drop in volume or a sudden increase in cost).

Previously, two group members shared their account rollover situations: one group member took a nap on Sunday afternoon, and when he woke up, he found that a new test plan built with new materials in his account had exceeded the cost. He quickly paused it after discovering it, but the volume had already exceeded the cost (please imagine the consequences yourself)...

Another group member went out for dinner on the evening of the Lantern Festival and found out that his plan to increase volume had been suspended for two hours after the second review. He quickly contacted the media contact person to find out the specific reasons for the review and had it re-submitted.

-> Adjust account

It’s nothing more than changing the bid, budget, targeting or material, and then waiting to see if the data shows any signs of improvement.

It should be noted that the adjustment does not take effect immediately. There is a certain delay in platform data synchronization and it takes time to adjust the algorithm. Therefore, the frequency of changes should not be too high, and the number of changes should not be too many. Of course, the frequency and number of changes are not absolute.

A friend from a heavy gaming company shared his experience in advertising. Sometimes he would adjust the bid, budget and targeting frequently, but he found that the ROI adjusted according to his own methods was generally good.

I discussed the reasons with him, and it was mainly because his adjustments were generally to adjust the consumption speed rather than the cost. In other words, his adjustments were to expand on the original algorithm model's exploration of the population.

For many optimizers who lower bids and adjust targeting, they need to see how big the intersection is between the population after the adjustment and the previous population. If the intersection is too small, it will cause great damage to the algorithm model, which is also the reason why the plan is easily ruined.

-> Off work

On the surface, the optimizer went home after get off work, but in reality he changed his work location and still needed to monitor the account after returning home...

Because the traffic pool and bidding market of information flow advertising are constantly changing dynamically, you may not be able to take a second to pay attention to them. Sometimes when a new product is launched, you may have to work all night.

This is probably something that the SEM optimizer circle cannot imagine. I saw a recruitment information for an SEM optimizer in the group before, which described that "the working hours are 7.5 hours, and there is basically no overtime." It can be seen that SEM advertising is still relatively stable.

The information flow advertising optimizers in the industry group were only envious and said they all wanted to have such a job. Of course, the salary of information flow optimizers is also enviable for SEMers. There are pros and cons!

Basically, the daily work of information flow advertising optimizers is a repetitive cycle. Most optimizers make a living by increasing volume every day.

When 0 o'clock comes, it is a new beginning. The fact that there was no volume yesterday does not mean that there will be no volume today, and the fact that there was volume yesterday does not mean that there will be volume today. Please note that this "amount" refers to the amount within an acceptable cost range.

2. Does increasing advertising volume really depend on piling up accounts and plans?

From the moment of setting up an account, information flow advertising optimizers have to worry about increasing the volume. I don’t know when it started, but piling up accounts and plans seems to have become the universal “volume-raising symbol” in the information flow advertising industry.

The following are two common operations for information flow advertising optimizers to increase volume:

-> What if the quantity is not enough? Stack accounts + stack plans!

Because the traffic distribution of the advertising platform has a certain degree of randomness (when everyone's ecpm is the same), if you use different plans, you might be able to get some "luck" back. From a probability perspective, there should be one or two plans that can achieve explosive growth.

-> What should I do if the amount is too little? Copy old plans + copy old materials!

In addition to seeing if there is anything that can be done to remedy the situation when the volume is reduced, the most convenient way is to quickly copy the old plans and materials that had good volume before. If this account goes out of business, then change to another account, move the plan there and start over again.

The phenomenon of piling up accounts and plans is most obvious in industries such as gaming, e-commerce, and online services. From my communication with the top advertisers in these industries and the actual results of their delivery, I can see that piling up accounts and plans in large numbers is feasible. Some of them are on Bytedance, some are on Tencent Advertising, or other platforms, even if they are different accounts opened by the same entity (same business license).

However, this method has a major premise, which is that the optimizer must control the bid, budget, targeting and creativity, etc., to ensure that the traffic between accounts/plans does not affect or the impact is not too large, otherwise they will be competing for traffic.

Therefore, it is beneficial to open multiple accounts and create multiple plans for advertising. Of course, there are some advertisers who are not suitable for stacking accounts and plans, especially those with high conversion costs or relatively narrow target audiences:

• For example, for a certain SLG game, the average conversion cost is 4,000, which means that if a new plan wants to pass the cold start, it must accumulate at least 20 conversions. In other words, a plan must cost at least 4,000*20=80,000 to learn and enter the mature stage. If too many plans are created, a lot of money will be spent on the cold start stage alone.
• Another example is a certain heavy game, which does not have a broad target audience like many small games, so it may not be suitable for some media platforms whose DMP population labels are not so matched and whose target audience is not so large. Building a large number of plans will result in a lot of advertising waste to non-target users. Moreover, if the target audience traffic pool is not large, building so many plans is just competing for traffic.

Each type of advertisement needs to selectively control the number of accounts and plans based on actual conditions. Stack accounts and plans scientifically within a reasonable range, rather than blindly and without section.

3. How to “stack accounts and stack plans” scientifically?

In fact, it can be seen from the official suggestions of the media platforms that it is necessary to have multiple accounts and multiple plans appropriately. The following are the daily new suggestions for game owners from the ByteDance Engine and Tencent Advertising Platform:

【Big Engine】

• It is recommended to ensure that the number of new plans created for a single account every day is at the level of 10+ , to promptly expand the budget for potential plans, and to clean up long-tail plans.
• It is recommended to have 5-8 creatives for each plan . Too many will increase the exploration cost. A reasonable number of creatives will help improve the estimated click-through rate/estimated conversion rate.
• The materials must be updated at a certain frequency and must keep up with the decay rate (for example, the e-commerce industry plans to reach a consumption peak on the second day of its life cycle, and the consumption on the fourth day is almost the same as on the first day).

【Tencent Advertising】

• Advertising accounts are divided into EQQ accounts and MP accounts. The number of accounts determines the initial testing speed . Having multiple accounts allows you to simultaneously start testing multiple products/multiple traffic pools/multiple test directions, which has a positive effect on rapid explosive growth. The specific number is recommended to be selected based on your actual situation.
• It is recommended that EQQ create at least 15-20 new ads per household every day . Testing one traffic system for each household will help the model accumulate quickly.
• It is recommended that MP create at least 20-25 new advertisements per household every day . The account can choose to test only the Moments or the Official Account, or mix and match.

In fact, the current situation of the entire information flow advertising industry is that many advertisers or optimizers only see that others have increased their sales by stacking accounts and plans, but they do not know the specific operational details.

They "copied the same thing" and found that the effect was not that obvious. For example, one account could invest 100,000 yuan, and they expected to invest 1 million yuan by setting up 10 accounts. In the end, the total investment in each account was still only 100,000 yuan or more. The more the scale did not change, the more he kept piling up accounts and plans, hoping to find a way out.

This is a kind of blind stacking. Especially with the support of advertisers/agents’ self-built or third-party advertising efficiency improvement tools, batch creation of new plans has become increasingly convenient, so the scale of plans has increased dramatically.

For media advertising platforms, piling up accounts and plans will also generate a large amount of junk data, which will cause certain troubles for data processing and algorithm models.

Therefore, Bytedance and Tencent Advertising have successively launched some policies for the advertising MKT API:

• On January 26, 2021, ByteDance launched a health mechanism: health-related factors include but are not limited to the total amount of invested plans, the total amount of newly created plans daily, etc., which may affect frequency control, permissions, value-added and other rights. It is recommended to promptly clean up invalid/outdated/unsustainable investment plans, allocate resources rationally, and allow health to develop in a benign manner.
• In order to improve the efficiency of Tencent's advertising delivery and management, starting from March 1, 2021, the platform will remove advertisements that meet both of the following conditions: (1) passed the review before September 1, 2020; (2) had zero consumption in February 2021. Please be aware that the advertisement can still be edited, copied, and re-submitted for review and release after it is taken offline.

In fact, multiple accounts and multiple plans are mainly used to serve multiple delivery strategies, and optimizers can use them to conduct AB testing to compare data.

However, most optimizers like to build repetitive systems, which are illogical and messy. It is impossible to analyze the data well, which is also a big problem for account and plan management.

Scientific pile accounts and pile plans should be "differentiated"!

At least there should be differentiation in materials, creative titles, targeting, and media selection. However, how many identical plans can be in an account, or how many plans the same material can be placed in, depends on how flexibly you can handle it based on the actual running results (collision probability results) and creative diversity.

The replication plan also needs to be scientific, and you can't just copy it randomly. What are the characteristics of a program that is worth copying? At least those that have passed the cold start learning period (with more than 20 conversions) and have seen a significant increase in consumption and cost advantages in the first few days after the new plan is launched. Of course, you can also use one-click inheritance to reuse the model data of the high-quality old plan delivery phase to reduce the cold start learning cost of the advertising platform.

Scientific pile accounts and pile plans should also be "structured"!

In the early stage, think clearly about the delivery strategy to be determined and the different influencing factors to be tested. Plan them first, then determine how many accounts to divide, how many plans to create, and how to set up the planning strategy.

If there are no chapters, when there is a problem with advertising delivery and you need to locate it, you can't remember which plan used which targeting or what creativity, so you have to check the positioning one by one.

The conventional structure uses classification dimensions such as product, targeting, ad space, creative material, and promotion goals, which can be combined with one or more.

The naming rules should also be configured accordingly to facilitate account monitoring process analysis and data retrieval. The account and plan building categories for different industries, different products, and different optimizers are also different, and the naming rules should be configured accordingly.

• Account structure: Look at the specific delivery needs of different industries and advertisers, such as different promotion purposes (attracting new customers, attracting active users, etc.), conversion target types (activation, payment, etc.), product categories (daily necessities, beauty products, etc.), and delivery traffic pool ends (TikTok, Toutiao or Pangolin Alliance of Bytedance; EQQ or MP of Tencent Advertising, etc.).
• Plan structure: This requires special attention. It is generally recommended to use it for AB testing, such as matching elements such as age, gender, region, population package, materials, copywriting, etc., in an organized and purposeful manner.

I believe that many optimizers are often not confident in their own operations:

• For example, when there are many accounts, you may worry that if you don’t pay enough attention to one of the accounts, something will go wrong, such as cost explosion or volume explosion, and you will not be able to keep track of it...
• For example, when there are a lot of plans, I will doubt whether the settings are set correctly, and worry about whether there will be mistakes in the rush...
• For example, when analyzing the data differences between different plans, I was worried that I would make the wrong conclusion. There were so many plans and so many settings that I couldn’t tell which one was which…

It can be seen that it is necessary to build accounts and plans in a structured manner. After the structure is determined, it can save a lot of time. At the same time, it is convenient to clearly compare the effects when looking at the data, and to discover and quickly find the plans that need to be adjusted in time.

Optimizers basically understand these principles, but not many of them can actually put them into practice. They tend to just do whatever they think of. Scientific investment needs to start with developing good habits of building accounts and plans.

4. How to scientifically “manage” multiple accounts and multiple plans?

When there are multiple accounts and multiple plans running, some of which even span multiple advertising platforms, how should optimizers elegantly handle multi-threading and multi-processing?

Advertising efficiency improvement tools are very popular in the industry. Some are built internally by advertisers or agencies, while those without a technical team will use external third-party tools (Chuangliang, Reyun, Mailiang Xiaofeiji, Youmi, Jiuzhilan, etc.).

Appropriate advertising tools are helpful for managing multiple accounts and plans. In addition, some tools also have APP or mini-program versions. It is a great blessing for optimizers to be able to view data anytime and anywhere, so they don’t have to hold a computer all the time.

Despite this, when using third-party delivery tools, you still cannot avoid the need to log in to the direct delivery interface such as the Massive Engine backend and Tencent Advertising backend to operate and compare data. After all, the data and functional interfaces of the MKT API are sometimes not very stable (for content about the MKT API, please refer to the previous article "Let's talk about the MKT API and RTA of the advertising platform from the shallow to the deep").

Therefore, when managing multiple accounts, optimizers, optimizer supervisors or project managers need to open multiple browsers to view multiple accounts, which is still very difficult to manage. To address this demand pain point, the industry has developed some browser plug-ins (Youmi Advertising Manager plug-in) and client operating platforms (Buying Volume Airplane) to solve this problem.

Most of these third-party tools on the market are charged for enterprise versions, so some optimizers may not be able to use them. There are probably three types of optimizers who read my article:

• My company has built its own delivery tools or has purchased third-party functions. Some functions are easy to use, some are not easy to use or some functions are not supported. However, from a personal perspective, I want to find a tool that can solve these needs. For example, someone in the group said that the company's tools do not support real-time export of reports.
• My company doesn't have a delivery tool, but I'm looking to buy one. I suggest you take a look at the companies I introduced above, and go to Baidu to check their official websites to find out which one is more suitable for your needs. If you need a matchmaker, I can also help you. Generally, there are free enterprise trials , and you can also apply for trials for all of them to compare.
• The company did not build the tool and does not plan to pay for it. Optimizers can also download the Youmi Delivery Manager plug-in or the Buy Volume Airplane mentioned earlier. Both are free personal versions and can be installed and used.

With the help of advertising efficiency improvement tools, the management of multiple accounts and multiple plans is relatively less difficult and messy than before.

It is recommended that optimizers must use advertising efficiency improvement tools scientifically. Remember that it is not used to create repetitive plans in batches, but should be used to assist in better building and managing differentiated and structured accounts and plans.

These tools are more about empowering optimizers, saving optimizers’ time on repetitive tasks so that they can do more refined work, including designing delivery strategies, thinking about creative copy, and analyzing data. If you encounter various metaphysical problems during the advertising process, you can also refer to the previous article "Scientific Analysis of the Metaphysics of Bytedance and Tencent Advertising".

Summarize

Before summarizing, let me share a complaint from an optimizer: "When salesmen from a certain large company came to the company to discuss business, the performance of several accounts was surprisingly good, especially when they came to sign the framework, the volume exploded, but after they left, the performance dropped. Is there any direct reason?"

This complaint is quite interesting, and I like to watch the fun, so I posted this to the industry group and @ed an algorithm designer from the factory, hoping to create a spiritual collision between the front-line "optimizers" and the back-end "production and research personnel."

• The factory staff stated that there is absolutely no such setting, and believed that most optimizers only look at the account perspective, not the customer perspective, and do not understand customers and their products . They advised optimizers not to always think of themselves as being persecuted and to stick to data-driven thinking. Whether you make money or lose money in the stock market, you will reflect on whether there is something wrong with your strategy. Why is it that when it comes to advertising, all the problems are with the platform? The drop in volume is more likely because the optimizer did not adjust the strategy in a timely manner based on the delivery results.
• At this time, an algorithmist from another large company also spoke up. He was much luckier and said that every time he went to a client to conduct research, his consumption would increase rapidly and he would keep thanking him. In fact, the number of customers increased before they had time to launch any changes. Advertising is just like buying stocks. It is easy to establish false causal relationships. Optimizers must have data thinking and a concept of confidence.

The phenomena mentioned above are all attributed to certain operations performed by personnel from large companies, except that one company was criticized for rapidly decreasing volume, while the other was thanked for rapidly increasing volume (I think to some extent it was the visit of algorithm personnel that gave advertisers a certain amount of confidence and belief).

Under the education of the algorithm personnel, the optimizers in the group had basically nothing to refute, and they all expressed their agreement with the statements of these algorithm personnel, believing that these were indeed "metaphysics" from the perspective of optimizers, and that their analysis of traffic changes was not scientific enough, but was attributed to some direct appearances.

This mental collision ended with the algorithm personnel crushing the optimizer personnel, and the algorithm personnel won! (To be honest, I have absolutely no intention of favoring research and development, haha~)

Therefore, please experience for yourself the causal relationship between piling up accounts, piling up plans and increasing volume. After all, there are many optimizers who rely on one or two plans to support 80% of the advertising budget...

Author: Advertising Practitioners Study Room

Source: Advertising Practitioners Study Room