Facebook-owned Instagram has now published a new one blog posts where they tell a bit about how they arrive at what should be displayed in Instagram user feeds and how this is ranked.
Instagram’s basic interpretation is that users in both the flow and stories prefer to see content from their friends, family and relatives. The flow design is then based on all new posts that have been published since the user last visited the app.
Each post then has thousands of “signals” attached to it. These can be things like the time they were posted, if they were posted from a mobile or the web or if you often like the person’s posts.
The most important signals that determine how posts are ranked in the feed are, at a roughly:
- How popular an entry is. How many people liked it. When it was posted. How long is it if it’s a video. What place the post is from and if a person is tagged.
- Information about the person who posted. How interesting the person is judged to be for you. How many times have you interacted in recent weeks.
- Your activity. What you liked and how many posts you liked.
- Your history with a person. How interested are you in general in seeing posts from a particular person. This can be affected, for example, by how often you comment on each other’s posts.
The more you interact with a post and the higher Instagram ranks the type of interaction with the post, the heavier these choices will weigh for what is shown in the future.
Instagram’s Explore mode works in a similar way, but with the difference that it is adapted to show new content based on what it thinks you are already interested in.
When Instagram was launched, all posts were in chronological order. According to Instagram itself, users in 2016 will miss 70 percent of all posts in the feed, which led them to introduce algorithms to rank them instead.