The Instagram Reels algorithm feels mysterious until you learn what it actually measures. Instagram's head, Adam Mosseri, has named the three signals that matter most for ranking: watch time, likes per reach, and sends per reach. That is the whole game in one sentence. Every editing choice, hook and caption you make should serve one of those three numbers, because the algorithm is simply counting how long people watch, how many like what they see, and how many send it to someone else.

Watch time is the total time people spend watching your Reel, including replays, and average watch time is that total divided by the number of initial views. The algorithm reads a high average watch time as proof that the Reel is worth showing to more people. Completion rate matters even more: a Reel that most viewers finish is far more likely to be pushed to non-followers than one that most viewers abandon halfway.

The practical rule is to make every Reel exactly as long as the idea needs and not a second longer. Cut pauses, dead air and slow intros, and put the payoff early enough that viewers have a reason to stay. Check your retention graph after every post: the flatter the line, the more engaged your audience was. Find the exact second where viewers drop off and fix that moment in your next Reel.

Likes per reach measures how many of the people who saw your Reel liked it, and it tells Instagram whether the content resonated or just passed by. A Reel seen by ten thousand people with a thousand likes performs better in the ranking than a Reel seen by fifty thousand people with the same thousand likes, because the first one delighted a bigger share of its audience. Reach without resonance is not growth.

To lift likes per reach, make Reels for a specific person with a specific problem, not for everyone. A Reel that speaks directly to new parents, beginner photographers or small bakery owners will earn likes from the people it was made for, while a generic Reel earns a shrug from everybody. Narrow targeting is not limiting your audience; it is concentrating your likes.

Sends, or shares by direct message, are the strongest signal of all, because sending a Reel to a friend is a personal recommendation. Mosseri ties sends directly to reaching new audiences, and for good reason: a Reel people send is a Reel people vouch for. Content that teaches something useful, says something funny about a shared experience, or solves a problem a friend has is the content that gets sent.

Design for sends deliberately. End tutorials with a line like send this to someone who needs it, and make relatable Reels about experiences your audience shares with their friends. A Reel about the struggles of freelancing gets sent between freelancers. A Reel about a cooking shortcut gets sent between home cooks. If nobody would send your Reel to a friend, the algorithm will notice.

Before any of the three signals matter, your Reel must be eligible for recommendation. That means original content, no watermarks from other apps, no reposted clips, and a length under three minutes if you want to reach non-followers. Reels longer than three minutes can still perform for your existing audience, but they will not be recommended to new people.

Think of eligibility as the door and the three signals as the engine. A Reel that passes the door but scores weakly on watch time, likes and sends will still stall. A Reel that scores strongly on all three but carries a watermark may never get through the door at all. Check the basics first, then optimise the signals, and review your Insights weekly to see which of the three needs work.