How should the targeting of information flow advertising accounts be set?

The essence of information flow is advertising finding people, and targeting is to tell the advertising system what kind of people I (advertiser) want to find.

Generally speaking, targeting is divided into three categories: [basic targeting], [behavioral interest targeting], and the less commonly used [media device targeting].

1. Basic orientation

Region: The most commonly used method here is to select by [province or city], which generally excludes remote areas [ Hong Kong, Macao, Taiwan, Xinjiang, Tibet, Qinghai... ]

It can also be divided according to development:

It can also be divided according to districts and counties, accurate to the district:

It can even be divided according to business districts:

The business district can also add a location from the map and the surrounding radius (3-15km).

Gender: Generally no limit. However, there are special cases. For example, most of the education categories will choose women. This client is set according to the needs of advertisers.

Age: It is also set according to the needs of the advertiser. If there is no specific need, you can ask the advertiser for a population portrait or do a test based on your own experience.

Custom audience: generally used to target a certain group of people or exclude a certain group of people. In addition, the massive backend supports simultaneous targeting and exclusion of groups of people, provided that the two groups of people are not the same group of people.

To target a certain group of people, just select "targeting" after the group package. To exclude a certain group of people, just select "exclude".

2. Behavioral interests

Behavioral interest targeting includes behavioral interest and Tik Tok influencer targeting.

Behavioral interests are divided into unlimited, system recommended, and customized.

System recommendation: In simple terms, based on the big data calculations of Toutiao, we predict the conversion population that may meet the current plan and deliver the campaign to this group of people.

The system will optimize the population according to the different stages of the delivery plan. It is more dependent on the system and can be used for AB testing with customization.

Customization: It is divided into two parts: behavior and interest. Behavior refers to an action that a user has taken recently. Interest means that the user may be interested in something.

For example, in the promotion of K12 education, the keyword "advanced mathematics " is set in the behaviors and interests.

Behavior: The user has recently liked or commented on a high number of related ads or searched for a high number of related content.

Interest: The user may be interested in advanced mathematics or the system may think that he is the parent of the corresponding child, so there may be conversion.

Here is an example of the behavior part

Behaviors and interests support industry category word selection and custom keyword selection. Content can be added in batches, supporting up to 350 (category words + keywords).

Tik Tok Expert: This is particularly useful when activating and promoting a tool app. It means targeting people who have interactive behaviors with fans of this (type) account.

For example: For example, if I target the two-dimensional influencer category, then people who follow, like, comment, forward, and interact with the live broadcast room and products of the Douyin influencers in this category will see my ads, and the target audience will be relatively accurate.

It also supports expert classification and custom experts, up to 30 experts can be added (expert classification + TikTok experts)

3. Media device targeting

• Platform (ios/android)
• Device type (mobile phone/tablet)
• Network (wifi/4g/3g/2g)
• Installed User
• Filter converted users
• Filter time
• Android version
• Article Categories
• Operator (China Mobile/China Unicom/China Telecom)
• New users (corresponding to new users of the app)
• Phone brand
• Phone prices
• Occupational status

This targeted category is numerous and complex, but fortunately, except for the following targeted customers who have no specific requirements, we don’t have to worry about the defaults.

• Platform: Choose based on whether you are targeting iOS or Android
• Installed users: If you are targeting new users, select Filter. If you are targeting active users, select Targeting. For others, you can choose Unlimited.
• Filter converted users: Generally, you can select [Company Account] or [App] level.

02 How should orientation be used?

1. General voting + DMP crowd package

The advantage of general advertising is that it has a wide exposure, but it is uncontrollable as to what kind of people it reaches, and sometimes the quality is very poor.

As for the DMP population package, nothing is more convincing than the actual conversion. If you have registered for an educational course, then you have registered, and the population is more accurate.

Therefore, it is relatively better to use the general investment + crowd package (DMP) when the account has a certain amount of historical data accumulated .

Because your targeted population is expanded based on your historical accumulated data and has certain similarities, it can allow new plans to get through the cold start faster.

2. System recommendation

The system uses big data to try to reach the people who may be converted by the plan. This method can only rely on the system. But for beginners, trusting the system is not a bad thing.

For plans that do not have too many rigid restrictions on targeting selection, you can consider using system-recommended targeting. However, a certain amount of observation time is needed, after all, the system also needs a learning process.

In addition, the system recommendation targeting has already screened out precise groups of people, so other targeting at the planning level does not require excessive settings.

If the plan already uses a targeted population package, the system recommendation targeting function will not take effect. However, the system recommendation and exclusion population packages are compatible and can take effect.

3. Behavioral interests

Behavior represents what actions a person has taken, and interest represents what a person may be interested in and thus generate conversion.

Compared with general advertising, behavioral interest words can find a relatively accurate group of people. Generally, behavioral interest words can be set by setting industry categories + keywords and used cross-wise.

03 Some notes on orientation

The meaning of targeting is to help the system model find people. If you find people quickly and accurately, the account will grow very quickly.

There are generally three ways to select orienteering:

• First wide then narrow
• Narrow first then wide
• Always wide

Generally speaking, if a plan can be implemented in large quantities, then it can be said that its targeted implementation is likely to be successful. Why do we say it is likely to be successful? Because there are too many factors that affect the execution of a plan, such as creativity, time period, bid, etc.

Therefore, we can only assume that the running volume plan is relatively well targeted and can be observed in detail.

Orientation itself is a vague thing, and there are a few things to note when using it:

• Don't be too serious about the direction, as long as you can make sense to both customers and leaders.
• There is no best orientation, only relatively good ones. There is no absolute good or bad.
• It is recommended to use it with the background to generate a directional package, which is convenient when modifying without having to modify it one by one.
• Broad targeting can sometimes produce good data performance.

Author: Erbai

Source: Schrödinger's Information Flow (W77PPT)