What Instagram has actually said
Most articles about the Instagram algorithm repeat each other. The most reliable thing we can point to is what the head of Instagram has said himself. In a video reported by Social Media Today on January 22, 2025, Adam Mosseri said: "The top three signals that matter most for ranking are watch time, likes and sends. So when looking at your insights, pay close attention to average watch time, likes per reach, and sends per reach."
That is a short statement, and it is worth reading slowly, because it contains three signals and one instruction about where to look.
The three signals
Watch time
Watch time is how long people actually watch your video. The metric Mosseri points you to is average watch time, which is in your insights. A reel that people leave after a second tells Instagram something very different from one they watch to the end.
Likes per reach
Notice the wording: likes per reach, not total likes. Reach is the number of people who saw the reel. Dividing likes by reach turns a raw count into a rate, which lets you compare a reel that reached few people with one that reached many.
Sends per reach
Sends are the times people send your reel to someone else. Same idea: it is measured against reach. A send is a person deciding that a specific other person should see this, which is a different kind of signal from a like.
Followers and strangers are not weighted the same
Mosseri also said that "likes are slightly more important for connected content, and sends are slightly more important for unconnected content." In plain words: when your reel is shown to people who already follow you, likes count slightly more. When it is shown to people who do not follow you, sends count slightly more.
The article adds one caution that is worth keeping: there is some variance depending on where content is shown and who it is shown to. Nobody outside Instagram has the exact formula, and Instagram has not published one.
What is not on the list
Social Media Today points out that comments are not mentioned at all in Mosseri's overview. That is the article's observation, not a statement from Mosseri, so it would be a stretch to say comments do not matter. What you can say is that they were not one of the three he named.
You will also see claims online that sends are worth three or five times a like. Those numbers do not appear in what Mosseri said, so treat them as someone's estimate rather than a rule.
What to do with this
You cannot control the algorithm, but you can control what you measure.
- Open your insights and find the three numbers. Average watch time, likes per reach and sends per reach.
- Compare reels with each other. A rate is only useful next to another rate from your own account.
- Look at the first seconds. Watch time starts there. If people leave early, the opening is the first thing to test.
- Ask whether someone would send it. A reel that is useful, funny or specific enough to forward to a friend is built for the signal that matters most with strangers.
- Test with people who do not follow you. Since sends count more for unconnected content, that audience is where this signal shows up first.
Why a rate beats a count
Imagine two reels. One was seen by very few people and some of them sent it to friends. The other was seen by many people and almost none of them sent it. The second has the bigger totals, but the first is the one where people did something with it. Because Mosseri points to signals measured against reach, comparing reels this way is closer to how he suggests reading your insights than looking at the biggest total.
It also protects you from a trap: a reel that reached a huge audience by luck can look like a success even when most people left in the first seconds. Average watch time is the number that exposes that.
A simple weekly routine
You do not need a dashboard for this. Once a week, open the insights for your last handful of reels and write down three numbers for each: average watch time, likes per reach and sends per reach. Put the reels in a list from best to worst on each number. Then look at what the top ones share: the opening, the length, the topic, how specific the idea was. Make the next reels more like those and less like the bottom ones. After a few weeks you will have your own picture of what your audience watches and forwards, which is worth more than any generic rule.
Common mistakes
- Chasing total likes. The guidance is about rates against reach.
- Treating one reel as proof. Variance is real; look at several.
- Believing exact weights. Nobody outside Instagram knows them.
- Ignoring where the reel was shown. The same reel can be read differently for followers and for strangers.
The short version
Watch time, likes and sends are the three signals Instagram's head named as the most important for ranking. Look at average watch time, likes per reach and sends per reach, remember that sends lean more toward strangers, and do not trust anyone who gives you exact weights.