The Two Numbers Are Not Rivals
Watch time and retention get discussed as though a creator has to pick a side, and the framing itself is the problem. They are not competing measures of the same thing. They are a quantity and a ratio, and a ratio is not interpretable without the quantity it came from.
Retention is the proportion of your video the average viewer got through. Watch time is the amount of attention you captured, either per viewer or in total. One is a verdict on the edit; the other is closer to a measure of the value you produced for the platform, which has to fill a feed with something people will stay for.
The reason this matters practically is that they can move in opposite directions from the same change. Shorten a video and the ratio usually improves while the quantity usually falls. Extend a video that has something to say and the reverse happens. Any decision judged on only one of them is being judged on half the evidence.
What the Platforms Say They Use
YouTube is specific about Shorts. It states that "Our systems use the signals for % of viewers who chose to view, avg. view duration and avg. % viewed to inform ranking." Average view duration is a quantity. Average percentage viewed is a ratio. Both are named, separately, in the same sentence.
TikTok describes it differently but points the same way, stating that "For most users, user interactions, which may include the time spent watching a video, are generally weighted more heavily than others."
Neither platform describes one as a substitute for the other, and neither publishes a formula for how they are combined. What they do make clear is that time held is not a secondary concern, which means the common advice to relentlessly shorten everything is at best incomplete.
The Four Combinations, and What Each Is Telling You
High on both is the straightforward success: the video held people, and it was the right length for what it had to say. There is nothing to diagnose.
High watch time with low retention means the material was strong enough to hold real attention but the runtime outran it. The fix is almost always in the last third, where a video that has already delivered its payoff keeps going. Look for the point where the curve stops falling gently and starts falling steeply, then ask what the video was still doing after it.
High retention with low watch time is the flattering failure, and it is the one most likely to be celebrated by mistake. A very short video can hold a high proportion of its audience while capturing very little attention. If this is your pattern across a whole channel, the question is not how to improve the ratio but whether you are leaving value on the table by never attempting anything longer.
Low on both is an opening problem in the overwhelming majority of cases. When people leave in the first seconds, everything downstream is measured on an audience that was never really there, and no amount of improving the middle will show up in the numbers. The four cases, in short.
| Watch time | Retention | What it means |
|---|---|---|
| High | High | Right length, and it held people |
| High | Low | Strong material, runtime outran it |
| Low | High | Short video, flattering ratio |
| Low | Low | An opening problem, not an ending one |
Which One Should Drive the Edit
Use retention to decide what to cut and watch time to decide whether cutting was right. That ordering resolves most of the tension between them. The curve shows you which stretch is losing people; the total tells you whether removing it cost you more than it saved.
The underlying principle is simpler than either metric: every section has to earn the next few seconds of somebody's time. A video that passes that test has a healthy ratio as a side effect, because nothing in it was dead weight. A video that fails it can be rescued by cutting the dead sections or by giving them a job, and the numbers will tell you which worked.
What the numbers will not tell you is which section is the dead one, because an average has already blended every viewer's journey into a single figure. Retensis reads the video against its own curve and names the moment, which turns the choice between these two metrics into a question you can actually act on. If you want the arithmetic side first, converting retention into seconds is the fastest way to stop misreading either one.
Frequently asked questions
They answer different questions, so the honest answer is that you need both and they fail differently. Retention tells you whether the video earned its own runtime, which is a verdict on your editing. Watch time tells you how much attention you actually captured, which is closer to what the platforms are weighing when they decide where to show it next. A video can be excellent on one and poor on the other, and knowing which is which changes what you fix.
Constantly, and short videos are the usual cause. A well-made 10 second clip might hold 80% of its audience, which sounds superb, and still capture only 8 seconds of attention per viewer. A 60 second video at 40% holds people for 24 seconds. The shorter video has the better ratio and the longer one held people three times as long, which is why quoting retention without the runtime beside it is close to meaningless.
Both appear in what the platforms describe. YouTube names average view duration and average percentage viewed as separate signals that inform how Shorts are ranked, and TikTok says time spent watching sits inside the group of signals that generally carries the most weight. Neither describes one as replacing the other, which is a reasonable hint that optimizing a single number at the expense of the other is not the strategy it appears to be.
Only if the video was longer than its material justified. Cutting a dragging video improves both numbers at once and is simply good editing. Cutting a video that was holding people improves the percentage while throwing away the attention you had earned, which reads as an improvement in one column and a loss in the one that matters more. Decide from where the curve actually falls, not from a rule about length.
See what Retensis Vision finds in your next video
Upload a video or paste a YouTube URL and get a full multimodal breakdown of your hook, pacing, audio, delivery, and predicted retention, in about 90 seconds. Free to start.
Analyze your video free →