Skip to content

Foundations · module 1 of 7 · 7 min read

How Instagram decides who sees what

There is no single algorithm. So what is actually deciding your reach?

  • connected reach
  • unconnected reach
  • recommendations
  • ranking per surface
  • seed audience

There is no single Instagram algorithm, and believing there is one causes most of the bad advice in this space. Instagram has ranked surfaces — Feed, Reels, Stories, Explore, Search — and each ranks differently because each is doing a different job.

Connected versus unconnected reach

The most useful distinction in the whole platform.

Connected reach is people who already follow you. Ranking here is mostly about your relationship with each person: how often they interact with you, how recently, whether they seek you out.

Unconnected reach is people who do not follow you, shown your content because a system predicted they would like it. This is recommendation, and it is where growth comes from.

They behave differently, respond to different signals, and can move in opposite directions at the same time. Followers flat while views triple, or views collapsing while your regulars still see everything, are both normal — and the aggregate "reach" number hides which happened.

Instagram exposes this split in insights as followers versus non-followers. Look at it before drawing any conclusion.

Ranking is prediction

Each surface predicts how likely you are to take particular actions — watch to the end, send it to someone, comment, linger — and orders content by a weighted combination. The weights differ per surface, which is why the same post performs differently in Feed and in Reels.

Two consequences worth internalising:

  • Every viewer gets a different ranking. There is no "position" to occupy.
  • The prediction is about the viewer, not about you. A post is not good or bad in the abstract; it is a better or worse match for a particular person at a particular moment.

The seed and the expansion

Recommended content is typically shown to a small group first. If it performs, it expands. This is why reach is so lumpy: the difference between 2,000 and 200,000 views is often one early threshold, not a tenfold difference in quality.

It also means small accounts can reach large audiences — a genuine change from the follower-gated era — and that consistency matters more than any individual post.

What trips people up

  • Reading "reach" as one number when it is two.
  • Assuming a bad week means a penalty. Recommendation is volatile by design.
  • Optimising for followers on a platform that increasingly distributes to non-followers.

You have got this when

You can look at a drop in views and say whether it was follower reach or recommended reach that changed — and know that those need different responses.

Go to the source