Instagram finally reveals more about how its algorithm works

The Instagram algorithm is something many of us love to hate. It’s an invisible entity we’ve been trying to beat over and over again to make our content more visible. But the truth is – none of us really know how exactly the algorithm works. Well, the company has finally decided to become more transparent about it and tell us more about how Instagram picks content for its users.

Instagram CEO Adam Mosseri recently published a blog post unveiling the secrets behind the dreaded Instagram algorithm. According to him, this is only the first one in the series of posts, so we can expect to learn more about this topic in the future. So, how does it work?

First of all, “The Algorithm” is actually a misconception. There isn’t just one, but Instagram uses “a variety of algorithms, classifiers, and processes.” Each of them has its own use and purpose, so there isn’t just one algorithm we often refer to. “Each part of the app – Feed, Explore, Reels – uses its own algorithm tailored to how people use it,” Mosseri explains.

One of the main questions most of us have is why on Earth our Feed hasn’t stayed chronological. “As more people joined and more was shared, it became impossible for most people to see everything,” Mosseri writes, “let alone all the posts they cared about.”

“By 2016, people were missing 70% of all their posts in Feed, including almost half of posts from their close connections. So we developed and introduced a Feed that ranked posts based on what you care about most.”

The content you see in your Feed and Stories differs from what you’ll find in the Explore or Reels. This is because the first two are focused on the people and content you already follow, and the other two help you to discover something new. Here are the criteria for ranking posts in Feed and Stories:

• Information about the post. These are signals both about how popular a post is – think how many people have liked it – and more mundane information about the content itself, like when it was posted, how long it is if it’s a video, and what location, if any, was attached to it.

• Information about the person who posted. This helps us get a sense for how interesting the person might be to you, and includes signals like how many times people have interacted with that person in the past few weeks.

• Your activity. This helps us understand what you might be interested in and includes signals such as how many posts you’ve liked.

• Your history of interacting with someone. This gives us a sense of how interested you are generally in seeing posts from a particular person. An example is whether or not you comment on each other’s posts.

As for Explore, Instagram wants to help you discover new content that you may find relevant. It still does a terrible job in my case, but maybe it takes more time, I don’t know. Anyway, here is what’s included in the predictions of what you might like:

• Information about the post. Here we are looking at how popular a post seems to be. These are signals like how many and how quickly other people are liking, commenting, sharing, and saving a post. These signals matter much more in Explore than they do in Feed or in Stories.

• Your history of interacting with the person who posted. Most likely the post was shared by someone you’ve never heard of, but if you have interacted with them that gives us a sense of how interested you might be in what they shared.

• Your activity. These are signals like what posts you’ve liked, saved or commented on and how you’ve interacted with posts in Explore in the past.

• Information about the person who posted. These are signals like how many times people have interacted with that person in the past few weeks, to help find compelling content from a wide array of people.

Reels is one of the most recent features of Instagram, pretty much a copy of TikTok. “Reels is designed to entertain you,” Mosseri writes, and it also shows you stuff from accounts you don’t follow (yet). These are the most important signals Instagram takes to show you Reels it thinks are relevant:

• Your activity. We look at things like which reels you’ve liked, commented on, and engaged with recently. These signals help us to understand what content might be relevant to you.

• Your history of interacting with the person who posted. Like in Explore, it’s likely the video was made by someone you’ve never heard of, but if you have interacted with them that gives us a sense of how interested you might be in what they shared.

• Information about the reel. These are signals about the content within the video such as the audio track, video understanding based on pixels and whole frames, as well as popularity.

• Information about the person who posted. We consider popularity to help find compelling content from a wide array of people and give everyone a chance to find their audience.

Shadowbanning

“Shadowbanning” is an Instagram-related concept we’ve mentioned several times. Basically, if you’ve been shadowbanned, it means that the platform has limited visibility of your posts. There are even some apps and websites that help you discover if this has happened to you, but in 2019 Instagram denied that this was happening. According to Mosseri’s most recent blog post, the company still denies it.

“We also hear that people consider their posts getting fewer likes or comments as a form of “shadowbanning”. We can’t promise you that you’ll consistently reach the same amount of people when you post. The truth is most of your followers won’t see what you share, because most look at less than half of their Feed.”

However, Mosseri admits that Instagram hasn’t always done enough to explain why it takes down some content. The plan is to work on this some more and to become more transparent about it. I sure hope it will actually happen.

Finally, you can help to train the algorithms and make them work better for you. After all, artificial intelligence isn’t always that intelligent so it needs more input from us. Here’s what you can do to tailor your Feed and Explore to suit you:

• Pick your Close Friends. You can select your close friends for Stories. This was designed as a way to let you share with just the people closest to you, but we will also prioritize these friends in both Feed and Stories.

• Mute people you’re not interested You can mute an account if you’d like to stop seeing what they share, but are hesitant about unfollowing them entirely. This way, people don’t know you’ve muted them.

• Mark recommended posts as “Not Interested.” Whenever you see a recommendation, whether it’s in Explore or in Feed, you can indicate you are “not interested” in that post. We will do our best not to show you similar recommendations in the future.

As I mentioned, Mosseri promises that there will be more blog posts like this that will shed the light on how Instagram algorithms work. And when we learn more about it, does it mean we’ll finally win the race against “The Algorithm(s)?” What do you think?

[via The Verge]