How to Read Your TikTok, Reels and Shorts Analytics (What Actually Matters)

Open your analytics and you get hit with a wall of numbers. Views, likes, reach, profile visits, average watch time, completion rate, shares, saves, follows, on and on.

Most of it doesn't matter. A few numbers matter a lot. The problem is nobody tells you which is which, so people stare at likes and ignore the stuff that actually drives the algorithm.

Here's the short list.

The number that matters most: watch time

The single most important number is how long people watch. Call it average watch time, average view duration, whatever your app calls it.

Here's why it beats everything else. The app decides whether to push your video based on whether people keep watching. If they stick around, it shows it to more people. If they bail, it stops. Likes and comments are nice, but watch time is the thing the algorithm actually rewards.

So when you look at a video, the first question is: did people watch most of it, or did they drop off fast?

Completion rate: are they finishing it?

Close behind is completion rate — the percent of people who watch all the way to the end.

For short videos, you want this high. Short form generally sees 60–90% completion, and a good target for anything under 30 seconds is 70% or higher (Social Insider, Virvid). On TikTok specifically, people often cite 75%+ completion as the point where the algorithm really starts pushing a video out.

If your completion rate is low, your video is too long for what it delivers, or it loses people in the middle. Cut it down or tighten the middle.

The retention graph: where exactly people leave

This is the most useful screen most creators never open. The retention graph shows you, second by second, where people drop off.

Look for the cliff. There's almost always one spot where a bunch of people leave at once. That spot is your problem. Maybe it's a slow intro, maybe it's where you answered the question too early, maybe it's a boring stretch. Find the cliff, fix that exact moment, and your next video holds better.

A quick benchmark to aim for: TikTok viewers tend to watch around 78% of a video on average, YouTube Shorts around 73%, and Reels around 65% (Social Insider). If you're well under that, the issue is usually the first few seconds.

Shares and saves beat likes

If you're going to look at engagement, look at shares and saves before likes.

A like is cheap. A share means someone thought it was worth sending to a friend. A save means someone wants to come back to it. Both are stronger signals that your content actually landed, and both tend to travel further with the algorithm. A video with fewer likes but lots of shares is often the better video.

What you can mostly ignore

• Raw view count on its own. It depends on your size and the algorithm's mood. Use it next to your own average, not as a trophy.
• Follower count as a daily metric. It's a slow result of everything else. Watching it every day just makes you anxious.
• Likes in isolation. Easiest to get, weakest signal.

Put it in order

When you check a video, go in this order:

1. Watch time / average view duration — did they stay?
2. Completion rate — did they finish? (aim 70%+)
3. Retention graph — where's the cliff?
4. Shares and saves — did it spread?
5. Everything else — context, not the main event.

Do that and you'll learn more from one video than most people learn in a month of watching their like count.

The annoying part is each app buries these numbers in a different place and shows them differently, so comparing across TikTok, Reels and Shorts is a hassle. That's something Creaswipe pulls into one place for you — the same key numbers for all your videos, side by side — so you're reading your performance instead of digging for it.