The belief ignores that YouTube has explicitly and repeatedly said there is no universal ideal video length, and that two related but distinct metrics, average percentage viewed and absolute watch time, can pull in opposite directions depending on how a video is made. A longer video can generate more total minutes watched even if a smaller percentage of viewers finish it, while a shorter video can post a much higher completion percentage but contribute fewer absolute minutes to a viewer’s session. Since both metrics feed into how the recommendation system evaluates a video’s session value, “longer is always better” collapses one input into a false universal rule when the actual relationship between the two metrics is what determines outcomes, and that relationship is content-dependent, not length-dependent.
The core mechanism: two different metrics, not one
Average percentage viewed (sometimes called audience retention percentage) measures what portion of a video the typical viewer watches, expressed as a percentage of total runtime. Absolute watch time (or average view duration in raw minutes/seconds, multiplied across all views) measures the total minutes a video accumulates. These are mathematically related but not interchangeable: a 4-minute video with 80% average retention delivers roughly 3.2 minutes of watch time per view, while a 40-minute video with 30% average retention delivers roughly 12 minutes of watch time per view, four times as much absolute watch time despite a far lower completion percentage.
YouTube has been direct in Creator-facing documentation and public creator communications that it does not have an ideal video length and has pushed back specifically on the idea that simply making videos longer is a reliable optimization tactic. The reasoning given is that stretching a video’s runtime without a corresponding increase in the value delivered per minute tends to reduce the percentage of the video viewers actually watch, and a video that loses viewers early or mid-way isn’t necessarily earning more total watch time just because its runtime is longer; it may earn less, if viewers churn out well before the end. Conversely, YouTube has also acknowledged that longer videos have the mathematical opportunity to accumulate more total watch time from viewers who do stay engaged, which is presumably part of why watch-time-heavy formats (podcasts, long-form commentary, tutorials) can perform well in recommendation surfaces when they hold attention. Both statements are true simultaneously because they describe different metrics; the belief in a “longer is universally better” rule flattens that nuance into a single directional claim that doesn’t hold once you separate the two numbers.
The practical result is that neither metric alone should be read as “the” ranking driver. A high percentage-retention video that’s very short may still contribute relatively little absolute watch time to a viewing session, and a long video with mediocre percentage retention may still contribute a lot of absolute watch time, if enough viewers stick around long enough in raw-minute terms. What actually seems to matter, based on YouTube’s own framing of session-based value, is whether a video keeps people watching (in whichever combination of length and retention achieves that) relative to the alternative of them going to do something else, including watching a different video.
Why the universal-length belief persists despite this
Part of why “longer is better” spread as a rule of thumb is that, in aggregate across the platform, longer videos that succeed do tend to accumulate large absolute watch-time numbers, and creators observing top-performing long-form channels can mistake correlation (successful long channels have a lot of watch time) for causation (length itself causes the watch time). But this ignores survivorship: the long videos that get held up as examples are the ones that also achieved strong retention; the many long videos that don’t hold attention and post weak percentage-completion numbers don’t get cited as evidence because they didn’t perform well, which biases the visible sample toward “long and good” rather than “long” as an independent success factor. YouTube’s own guidance has consistently pointed creators toward making videos exactly as long as the content warrants and no longer, explicitly discouraging padding runtime as a growth tactic, which only makes sense as guidance if length itself isn’t the lever; retention-per-minute is.
What this means practically for content decisions
The actionable implication is to stop treating runtime as a lever to pull independent of content value, and instead evaluate any length decision against both metrics together:
Before extending a video’s length (adding a segment, an intro, additional examples), ask whether that addition earns proportionally more absolute watch time than it costs in percentage retention. If an addition is likely to cause a meaningful chunk of the audience to drop off before reaching it, it may reduce total watch time even though it increases runtime.
Look at a channel’s own retention graphs (available in YouTube Analytics) for existing long-form content to see where viewers are actually dropping off, and treat those drop points as evidence about what the audience will and won’t sit through, rather than assuming a longer format will automatically be tolerated because “the algorithm likes long videos.”
When testing longer formats, track both average percentage viewed and estimated total watch time for the video (views multiplied by average view duration) rather than optimizing for only one. A win on one metric and a loss on the other isn’t a clear signal either way; the goal is genuine improvement in total minutes delivered without a collapse in the percentage of the video people actually watch.
Resist using industry rules of thumb about a specific “ideal” runtime (in minutes) as a target, since YouTube has explicitly said no such universal number exists; the right length is whatever length lets the specific video’s content sustain the highest combination of retention and total watch time for that specific audience and topic.
