Learning from past results: automating "targeting"
Posted: Sat Dec 21, 2024 3:56 am
Next, we will learn about automating "who = targeting" through Facebook Ads (hereinafter referred to as FB Ads).
Facebook ads have a function that uses the "Facebook Pixel" to analyze information on users who visited the destination (or took some action such as conversion) and automatically generates a " similar audience ," a user segment with similar qualities to the users analyzed . Acquiring this similar audience is one of the important steps in running Facebook ads.
Once you target this automatically generated lookalike audience, you won't have to worry about any other targeting settings.
Again, here is an example of how using lookalike audiences can perform:
The red box at the top shows the segment using similar audiences, and the blue box at the bottom shows the results of manually created segments based on attribute information (age, gender, hobbies, interests, etc.) that can be confirmed (or assumed) from the customer database using various settings in Facebook ads. The rightmost column is CPA.
In the past, I have had the opportunity to use "similar" segments through various DSPs, GDN, YDN, etc., but I don't think I have often felt so clearly that " lookalike audiences are reliable ."
*Of course, the results of manually created phone number database malaysia segments may vary greatly depending on the settings, but regardless of how well they are done, these results are based on as much manual consideration as possible, as mentioned above. For many other accounts, Facebook Ads' similar audiences have also produced good results, and we can see how accurate they are.
Required roles: "Design ability," "Creativity," and "Leadership"
In this era where the role and scope of automation is expanding and performance far exceeds that of human work, what is expected of operators? I will list some of the role changes that I have personally experienced and that are happening right before our eyes.
Design Ability
In light of the recent trend towards automation, a change in thinking is required when designing an account. This change means that while previously it was designed with a focus on "how to control" the focus will shift to " how to make it learn ."
As mentioned in " Criteo Senior Sales Representative Kono Masahiro explains: The essence of Criteo's operations is not 'bidding' but 'input' to the engine ", you need to be able to understand the basic specifications of advertising products and then design and configure them.
If you take convenient measures without understanding the importance of data learning, such as removing measurement tags to adjust delivery volume, arbitrarily changing budget settings, bid prices, target values, etc., or excluding some product records from the data feed, the automation function will not be able to perform to its full potential.
There are rules to the design, and it is necessary to create an environment where the automation functions of the advertising product can function at full capacity .
*I would like to delve deeper into this "design ability" further through this Ad JOURNAL.
Facebook ads have a function that uses the "Facebook Pixel" to analyze information on users who visited the destination (or took some action such as conversion) and automatically generates a " similar audience ," a user segment with similar qualities to the users analyzed . Acquiring this similar audience is one of the important steps in running Facebook ads.
Once you target this automatically generated lookalike audience, you won't have to worry about any other targeting settings.
Again, here is an example of how using lookalike audiences can perform:
The red box at the top shows the segment using similar audiences, and the blue box at the bottom shows the results of manually created segments based on attribute information (age, gender, hobbies, interests, etc.) that can be confirmed (or assumed) from the customer database using various settings in Facebook ads. The rightmost column is CPA.
In the past, I have had the opportunity to use "similar" segments through various DSPs, GDN, YDN, etc., but I don't think I have often felt so clearly that " lookalike audiences are reliable ."
*Of course, the results of manually created phone number database malaysia segments may vary greatly depending on the settings, but regardless of how well they are done, these results are based on as much manual consideration as possible, as mentioned above. For many other accounts, Facebook Ads' similar audiences have also produced good results, and we can see how accurate they are.
Required roles: "Design ability," "Creativity," and "Leadership"
In this era where the role and scope of automation is expanding and performance far exceeds that of human work, what is expected of operators? I will list some of the role changes that I have personally experienced and that are happening right before our eyes.
Design Ability
In light of the recent trend towards automation, a change in thinking is required when designing an account. This change means that while previously it was designed with a focus on "how to control" the focus will shift to " how to make it learn ."
As mentioned in " Criteo Senior Sales Representative Kono Masahiro explains: The essence of Criteo's operations is not 'bidding' but 'input' to the engine ", you need to be able to understand the basic specifications of advertising products and then design and configure them.
If you take convenient measures without understanding the importance of data learning, such as removing measurement tags to adjust delivery volume, arbitrarily changing budget settings, bid prices, target values, etc., or excluding some product records from the data feed, the automation function will not be able to perform to its full potential.
There are rules to the design, and it is necessary to create an environment where the automation functions of the advertising product can function at full capacity .
*I would like to delve deeper into this "design ability" further through this Ad JOURNAL.