A blog post written by Instagram CEO Adam Mosser explains how content is categorized for users in different sections of the app.

Mosseri explains how content is classified in the main feed, the research feed, the reel feed, and the carousel of stories.

Here is a summary of the main highlights of Mosser’s article.

## Instagram algorithm (s)

There is no single algorithm that would organize all the content on Instagram, Mosseri says.

Each part of the application – Input, Explore, Reels and Stories – has its own algorithm. Content is categorized in each of these sections according to how people use it.

## How feeds and stories algorithms work

Feed and story algorithms organize the latest messages shared by the people you follow.

Thousands of signals are used to determine the order in which the content is displayed.

Mosseri says these are the main signals in the main feed and in the carousel of stories:

• About the mail: This includes how popular it is, how many likes it has, when it was posted, where it was posted, and other general information.
• About the content providerA: This includes how much commitment the creator has received in recent weeks.
• User action: Contains signals about a user’s recent activity on Instagram, such as how many posts they’ve kept.
• The history of the user interacting with the content provider: An example is whether a user and a content provider have commented on each other’s posts.

From there, the algorithms try to predict how likely the user is to interact with the message.

The more likely a user is to commit to a message, the higher the message will rank.

For feed posts, Instagram looks at user probability:

• Spend a few seconds on the message
• Comment on the post
• Like the message
• Save the message
• Tap the profile picture associated with the message

In some cases, the algorithm takes into account other factors. For example, an Instagram feed avoids showing too many messages in a row from the same person.

## How the Instagram Explore Algorithm works

The Instagram Explore algorithm places content in a separate feed from accounts that the user is not yet tracking.

It works in the same way as the main input algorithm by collecting signals from messages previously held, saved, or commented on by the user.

Mosseri explains how these signals are used to search for relevant content in accounts to which the user is not directly connected:

Let’s say you’ve recently liked several photos of San Francisco chef Cathay Bi (@dumplingclubsf). Then we look at who else likes Cathay’s photos, and then what other accounts those people are interested in. Maybe people who like Cathay are also in the SF dim sum spot @dragonbeaux. In that case, the next time you open Explore, we may show you a picture or video from @dragonbeaux. In practice, this means that if you’re interested in dumplings, you might see posts on related topics like Gyoza and dim sum, without necessarily understanding what each post is about. “

When the Instagram Explore algorithm finds messages that the user may be interested in, it classifies them using the same signals as the main input.

In summary, these signals are:

• User activity on Instagram
• The history of the user interacting with the content provider

Lastly, the content selected for the Explore feed must clear a set of guidelines that are specific to the content recommendations.

This ends up in cases where content that is otherwise appropriate for Instagram cannot be considered suitable for the Explore feed.