In a meeting with the media, Julian Gutman, shared some details of his dynamics, and the main factors that come into play for the algorithm to offer a personalized feed, based on user behavior.
As they mention in TC, analyzing our previous activity on Instagram, machine learning will predict those publications that will arouse our interest, taking into account the content of the publication. Another factor that is taken into account is the date on which the publications were shared, since it will prioritize the newest ones.
And the third factor is the relationship or interaction we have with the creator of the publication. That is, if in the past we have interacted a lot (likes, comments, etc.) with the posts of a certain user, their content will have priority in our feed.
Taking into account all these factors, the algorithm organize our feed. And then, it goes on to a second stage, where it takes into account the frequency with which we use Instagram, the number of users we follow and the time we spend in the feed.
So, finally when we open Instagram, the algorithm will present the content as if it were an assistant… showing us the best posts since the last time we connected, from a varied group of users, taking into account the amount of time we spend.
This does not mean that Instagram hides posts, it simply personalizes the feed so that we see the most relevant content for us, but if we continue scrolling we will see the rest of the posts without problems.
And while the Instagram team knows that users still miss the chronological feed, they consider the algorithm a success. According to their data, the dynamics proposed by the algorithm have allowed users to view up to 90% of their friends’ publications.