For the last three years the like count has been the worst possible signal for whether a Reel or a Short is going to spread. Creators still chase it because the heart icon is loud and visible, but Meta and Google have both moved their ranking systems past surface engagement. The platforms now optimize for retention and for shareability, and they reward content that buys time and earns redistribution. If you are guessing whether a clip will pop based on the first 200 hearts you see, you are reading the wrong dial. The actual dashboard is buried two screens deeper, and it tells a different story. There are three numbers that decide whether a reel goes anywhere, and they are the ones you should be watching.
The first number is average watch percentage. This is the share of the clip a typical viewer consumes before swiping away. A 30 second video that holds people for 24 seconds has an 80 percent average watch, and that single fact will outrank a 90 second video that holds 35 seconds at 39 percent. Meta has been transparent that retention is the dominant ranking signal, and creators who study their own analytics tab can see it directly. The goal is not to make the video shorter for the sake of it. The goal is to fit the length to the actual amount of substance you have. If your idea takes 22 seconds to land, do not stretch it to 45 to look more produced. The math punishes you for filler every single time.
The second number is send rate. That is the percentage of viewers who tap the paper plane icon and forward the clip to a friend or to a group thread. Sends are the strongest social proof the algorithm can read, because someone is willing to put their personal relationship on the line to share what they just watched. Anything above 1 percent is unusually strong for Reels. Above 2 percent and the platform will start pushing the content out to non followers within hours. Sends rise when a clip says something the viewer wishes they could explain to someone in their own life. It rises when a clip names a feeling, calls out a pattern, or delivers a useful fact in under 20 seconds. The clip becomes a tool the viewer uses to talk to someone else, and the algorithm cannot resist that.
The third number is replay rate. This is the percentage of viewers who watch a clip more than once in a single session. Replays are common on visual oddities, satisfying loops, dance choreography people want to learn, and short scripts where a punchline lands late. Anything above 8 percent on a Reel is real, and anything above 15 percent is a viral candidate. The clean way to design for replays is to end where you started. Loop the visual so the cut from the last frame to the first frame is invisible. Use a hook that only makes sense once you have seen the whole video, so people scroll back to check. Replays double your watch time without doubling your production time, which is the reason creators who optimize for them outscale creators who do not.
Once you start tracking these three numbers, the work shifts. You stop writing scripts to sound clever and start writing them to keep someone watching. You stop ending on a generic call to action and start ending with a line that gives the viewer something to say to a friend. You stop padding clips and start cutting frames. The first 50 reels you make this way will probably underperform what you used to post, because the format demands more discipline. The next 50 will start to bend the algorithm in your direction. The creators who get pulled up the platform are not the ones with the best ideas. They are the ones who learned to feed the three numbers the system is actually counting.
There is a simple weekly cadence that makes this practical. Every Friday, open the analytics for your last 10 posts and write down only those three figures. Look for the outliers in both directions. Ask what was different about the top clip and what was wrong with the bottom clip. Make one specific change for next week, like cutting two seconds off your average length or moving the hook earlier. Do that for eight weeks and you will know more about your own audience than any guide can teach you. The platforms publish the data. Most creators just never look at it.
