Facebook Social Score. Factors that determine the quality of a Facebook Ads profile (Part 1).

This material is not a manual on ad account farming and is not intended for such purposes. The only purpose of these notes is the study and evaluation of the global ranking algorithms and the blockage of ad accounts/social profiles on Facebook Ads service.

The content of this text does not reveal the details of the interaction of a particular person with Facebook Inc. therefore it doesn't violate policies and terms of use of this system. Any use of the material with the intent to violate the above-mentioned rules and policies are at the discretion and responsibility of the person who performs such an action.

If you already know all the factors, you can check our 2nd Part "Variables that affect blockage before account creation."

Read Part 2

But we advise you to start from the beginning. 

The mechanics of ranking of profiles and ad accounts is a combination of complex algorithms that take into account a substantial list of various factors. Everyone agrees that each profile has its own Quality Score that affects advertising, the frequency of security trigger calls, number of impressions and even the quality of the audience. The importance of what is called Trust Score/Trust Rate/Quality Score cannot be overemphasized. However, no one can comprehensively articulate exactly how this indicator is calculated.

The basic and the most obvious tips on maintaining profiles can be found online: behave yourself like a real user, do not run aggressive advertising, do not run multiple profiles using the same devices (mobile/desktop).

 

First, these tips don't help at all without proper understanding of the basics of ranking algorithms. Second, these tips still don't get us any closer to understanding how the mechanics itself actually functions.

 

Over the last 3 years of studies of various aspects of Facebook Ads, we were able, in our opinion, to gather the most truthful and the most detailed description of algorithms and their variables. The published material is the result of work of several media buying teams, tests of thousands of hypotheses and hundreds of ruined profiles. At the same time, we studied in detail all public information that the network representatives publish on newsroom.fb.com and searched many open repositories with the code from the research.fb.com Open Directory.

 

Ranking mechanics is a combination of complex algorithms and formulas. Therefore, the gathered material is quite voluminous (today few people like longreads). Since the monetization of traffic from Facebook Ads is handled by affiliates of different levels, we will try to represent all the data in the most accessible way.

To ease the process of full assimilation, the content will be divided into three structural components:

How ad accounts are blocked;

Variables involved in rating formation:

- Before registration;

- At the moment of registration and before the creation of the ad account;

- After the creation of the account and before the launch of advertising;

- Once your ads are running;

- After the blockage;

The formation of the Quality Score profile in dynamics;

 

We'll start with the most simple things: understanding blockage. To understand what affects the deactivation of your account, make sure you know HOW it happens.

The principles of blocking of ad accounts:

 

Rob Goldman, vice president for ads at Facebook once inadvertently admitted that the main tool for capturing the infringing content is the sequential analysis of behavioral patterns.

 

According to the latest Facebook public report, about 6.5 million profiles are blocked daily for publishing, attempting to publish or intention of publishing violating or prohibited content. It's worth noting that 98.5% of the accounts were stopped before the violators received the first complaints from the users. Obviously, the vast majority of these blockages were on spammers but a significant share of the affiliates were also involved.

 

According to the same document, Facebook recognizes that the most important metric of success is the interception of content (including adware) before it is shown to the end user. Facebook nevertheless claims that about 3% of all published content violates the rules one way or another.

 

The analysis of behavioral patterns takes into account all connections with other accounts and also maps the sequence of actions inside the profile. We'll examine this issue in the third part of our presentation but for now it's worth remembering the following: the accounts that are linked to or associated with profiles that either violated or still violate the rules are not immediately blocked. Facebook usually takes time to study the activity of the profile before finally banning it. This is necessary in order to continuously run Machine Learning and Artificial Intelligence of the social platform. This circumstance is used by so-called blackhatters who launch phony accounts in order to spend as much money as possible before being blocked.

 

It is important to remember: like Google, Facebook aggregates a huge amount of data about each user and then keeps it for a long time. A lot of data are gathered even before the actual registration of each user's devices. The very existence of these data together with various matches and performed actions determine the basic Social Score of the profile

 

Advertising specialists recognize that behavioral analytics are imperfect. Violators are constantly coming up with new ways to mimic loyal users and honest advertisers. For this reason the learning process never stops and robots will sometimes miss violating adverts to scrutinize the abnormal or not quite habitual behavior of advertisers.

This is the first article from the cycle. The next topic is "Variables that affect blockage before account creation".

Read