What's a Good Retention Rate? Benchmarks for YouTube, TikTok, and Reels

Almost everyone who looks at a retention graph asks the same question: is this number good? The honest answer is that "good" isn't a fixed line — it depends heavily on the platform, the format, and even the length of the video. A number that would be alarming on one platform is unremarkable on another.

Why there's no single benchmark

Retention rate measures the percentage of a video an average viewer watches. That percentage behaves very differently depending on what someone signed up for when they pressed play. A viewer who tapped a fifteen-second TikTok has made almost no commitment — they can bail at any point with zero cost. A viewer who clicked into a twenty-minute YouTube video has already opted into a longer format, so a moderate decline through the middle isn't automatically a red flag the way it would be in the opening seconds.

This is why comparing your retention number to a single "industry average" is close to meaningless without knowing what that average was measured against.

Rough ballpark ranges by platform

Treat these as general reference points, not precise targets — actual "good" varies by niche, audience, and video length.

YouTube (standard videos): Retention in the 40–60% range for an average view duration is commonly considered solid for videos in the 8–15 minute range; anything meaningfully above that is strong. What matters more than the aggregate number is the shape of the curve — a steady decline reads very differently than a sharp cliff in the first 15 seconds.

YouTube Shorts, TikTok, Reels: Short-form lives and dies on completion rate — the share of viewers who watch to the end. Creators and platforms generally treat anything near or above 100% average watch time (meaning many viewers rewatch part of the loop) as a strong signal the algorithm rewards. Below roughly 50% completion on a 15–30 second video is usually a sign the middle or the payoff isn't landing.

Longer-form and podcasts: Retention curves here tend to be flatter and more forgiving — viewers who commit to a 40-minute video have already filtered themselves for interest, so a gentler decline across the runtime is normal and not necessarily a problem to fix.

Where benchmarks mislead you

The biggest mistake is treating a benchmark as a pass/fail test instead of a comparison point against your own history. Your own channel's baseline — how a new video's retention compares to your last ten uploads in a similar format — is almost always more useful than an external number, because it controls for your specific audience and niche. A finance channel and a comedy channel will never share the same "normal" curve, and neither should be judged against the other's.

The second mistake is chasing the aggregate number at the expense of the shape. A video with 55% average retention and a hard cliff at second five has a very different, and more fixable, problem than a video with 55% retention and a smooth, steady decline. The average alone hides which one you're looking at — you have to look at the curve to know.

Using benchmarks before you publish, not just after

Retention benchmarks are useful for evaluating a video that's already live, but by definition they can only tell you where you landed relative to a range — not why, and not until real viewers have already watched it. If your last few uploads have been landing below where you'd like, the more useful question isn't "what's the industry benchmark" but "where specifically is this cut likely to lose people, and can that be fixed before it's live."

That's the problem Axony is built to solve: instead of waiting for a published video's retention curve to compare against a benchmark, it analyzes your edit before you publish and produces a predicted, second-by-second attention and retention curve, flagging exactly where a viewer is statistically likely to drop off. It won't tell you what "good" means for your specific niche — your own published history does that — but it will tell you, before you post, whether the cut you're about to publish is likely to clear your own bar or fall short of it.