TikTok Analytics tells you the average watch time and the percentage who watched the full video. It does not tell you where people left or why. The TikTok Retention Analyzer reads your script or a description of the video, predicts the drop-off points, scores the loop, and gives you three edits ordered by impact.
If you paste the numbers from Analytics as well, the report compares its prediction against your actual result and diagnoses the gap. This guide explains how to read it.
First-second stay rate
The report starts with an estimate of how many viewers survive the first second, based on what is on screen and said at second zero. This is the single biggest lever. A video with a 60% stay rate and average content beats a video with a 35% stay rate and great content, because the second one never gets the chance.
Drop-off points and reasons
Each predicted exit has a timestamp and a cause: a slow beat, a repeated idea, a sign-off that signals the end, a promise the video has not paid yet. The reasons matter more than the timestamps. If the cause is "second example adds nothing", the fix is a cut, not a re-shoot.
Loop quality and the re-cut plan
The loop score measures whether the ending flows back into the hook. Low scores usually come from an explicit ending. The re-cut plan lists a new beat order and trim points you can apply in the editor in ten minutes, and is usually worth trying on the existing footage before filming anything new.
Tips
- Paste both the script and the Analytics numbers when you have them. The comparison is the most useful part.
- Apply the three edits in order. The first one alone often moves average watch time by a second or more.
- Run the tool on a video that did well, too. Knowing why it held viewers is how you repeat it.
Common mistakes
- Analysing only flops — The reasons a video worked are as specific as the reasons one failed.
- Re-shooting instead of re-cutting — Most drop-off fixes are trims. Try the re-cut plan first.
- Adding more content to fix retention — Longer is rarely the answer. Tighter is.
Worked example
Input: Script pasted (35s). Analytics: average watch time 11.2s, 18% watched full video.
What the report returned: Stay rate estimate 45% — the first frame is the creator talking, not the shot pouring. Predicted exits at 6s (second sentence repeats the hook) and 22s ("so basically" signals a wrap-up). Loop 30/100. Edits: open on the pour, cut 5–8s, remove the last 4s and end on the tamp.
Frequently asked questions
Does it need a link to the TikTok?
No. It works from the script or a description plus optional Analytics numbers; TikTok does not offer per-video data to third parties.
What is a good average watch time?
It depends on length. As a rule of thumb, 50% or more of the video's duration is healthy; the report benchmarks against the format you describe.
Can it analyse a video I have not posted?
Yes. Paste the script; the prediction is most useful before you post.