The evolution of the Facebook algorithm

[tanjimajuha20]

Every day, millions of users share their thoughts, photos, videos, and other content on Facebook. But how does the social network decide which content to show first and which to hide? The answer lies in the Facebook algorithm.

Yes, it is constantly changing and improving, but its main function is to show users content that they are likely to find interesting.

2009. The news feed was 1000 cell phone number list formed in accordance with the popularity of publications.

2015. The "See First" feature was introduced. Users were given the opportunity to choose the pages that the algorithm should give priority to. In addition, the social network began to limit messages from pages that published a large number of advertising messages.

2016. Friends' messages become a priority. Also, informational and entertaining content gets to the top of the film.

2018. The top of the feed is made up of messages that cause discussion. The algorithm takes into account the number of shares, comments, and reactions. The time of engagement is also becoming important. However, preference is still given not just to the duration of viewing the content, but to the quality of interaction with it.

2019: The "Why am I seeing this post?" update is released.

2020. The platform begins tracking fakes and disinformation.

2021. The algorithm starts working based on machine learning and AI.

2022. Using three ranking signals. Who posted, what type of content, and what kind of engagement there was with the correspondence.

What does a typical Facebook feed look like?
According tocompany report, in the first quarter of 2023, 70.5% of views came from posts shared by friends or from groups the user had joined. 18.8% came from content from sources the user wasn’t a member of but the algorithm thought might be interesting (also called “unrelated posts”). The remaining 10.7% came from event announcements, company catalogs, and advertising.

What makes up a typical FB page?

To help people understand how the feed generation algorithm works, the companycreated14 system maps with explanations.

Read also : Got an idea? Marketer, be prepared to answer these 14 questions before pitching it to management
Key factors for content evaluation
Meta's AI model uses basic inputs to help select content that people are likely to be interested in. Yes, the system takes into account the unique features of a post, how others have interacted with the post, and how people have responded to similar content in the past.

For example, if many people have responded positively to a particular post or a similar one, the post will appear higher in the feed. However, if that interaction was negative, or if some of the content is considered objectionable based on community standards, the content is removed or placed lower in the feed.

What other signals influence the formation of the tape?
Facebook predicts how long you're likely to spend viewing a post . It does this by taking into account:

how many posts from this category have you viewed;
the ratio of the number of messages to the total time you spent.
Are you scrolling past a publication ? This is influenced by:

publication characteristics (e.g. video or photo format);
Have you viewed similar publications?
Likelihood of interaction with the page. Signals that influence this prediction include:

your previous activity on this page;
General activity - how often you like, comment, and visit the page;
the number of comments available for the publication;
number of comment branches for a publication.
What is the probability that you will open a photo or video in full screen ? To do this, consider:

how many of these clicks have there already been;
how often the content was distributed;
number of reactions.
How interesting are you to publications of a certain group. The following signals are used:

how much time has passed since the last visit to the group;
the level of activity in it;
diversity of publications.
In general, the system always tries to show you a variety of content in the feed . For example, it ranks content so that the user does not see several videos in a row.

Read also : Advertising with Ukrainian Instagram bloggers - all about the cost, formats of cooperation and benefits for the brand
Facebook is also trying to combat the spread of fake news . So posts that are likely to contain deceptive attempts will have a lower chance of appearing in your feed. In particular, the platformblocksfor click bait, spam links or links to low quality web pages. The system also checks photos, videos and text. In some countries, thisare attractedindependent experts certified by the International Fact-Checking Network. They manually check the authenticity of information in individual publications.

The system also takes into accounttime of publicationThe social network generally shows fresh content first so that users are always up to date.

How is the comment rating formed?
You may have noticed that comments below posts are not always displayed in chronological order, or that you can only read selected ones first.

How to form a comment rating on Facebook

There is an algorithm here too. Artificial intelligence ranks comments according to what it considers most suitable for each user. To do this, they take into account:

comment popularity;
publication time;
author.