One in five measured videos was not worth the time it asked for

Of the 3,243 videos we have measured, 659 were not worth the runtime they asked for — 20.3%, or one in five. The other four were.

That is the number the whole product turns on, so it is worth being precise about what it is not. It is not a claim that a fifth of YouTube is bad. It is a count of the times our measurement said: whatever this contains, it does not justify sitting through it.

A note on the numbers. Unlike the rest of this site, the figures on this page are frozen at the date above rather than recalculated on each visit. A report people cite should not move under them. When we run this census again it will be published as a new edition with a new date.

One video in five, but only one minute in seven

The 659 videos we would skip come to 145 hours of runtime — six full days. Against the library’s total of about 1,046 hours, that is 13.9%.

So the headline understates one thing and overstates another. One in five videos is a skip, but only about one minute in seven, because the skips are the shorter videos: a median of 9:08 against 14:18 for everything else.

The line that matters in that table is padding. A video we would skip carries 56.9% padding against 37.5% for one we would not — well over half its runtime restating, recapping or setting up rather than adding. That is not a video with nothing in it. It is a video with something in it, stretched past the point where waiting is worth it.

The useful part starts at almost exactly the same moment in both piles — 1:00 against 1:04 — which rules out the obvious theory. The videos we skip are not the ones with long introductions. They are the ones that never get dense after the introduction ends.

The subject you are in changes the odds by a factor of thirteen

Twenty subjects, and the share we would skip runs from 5.0% to 64.1%.

Travel is the outlier and it is not close: 41 of 64 measured travel videos were not worth their runtime. Sixty-four videos is a small sample and the figure should be read as an illustration rather than an estimate, but the direction is not subtle, and it is consistent with everything else we measure about that subject — travel also has the worst title accuracy of the twenty and the second-highest padding share.

At the other end, science and engineering sit at around one video in twenty. Both are subjects where the material chains: the next minute depends on the previous one, so there is nowhere for a video to go sideways without the explanation visibly falling apart.

The subjects near the top share a property. Design, productivity, business and tech news are all built around ideas that can be stated in a sentence, and a video that has committed to twenty minutes on a one-sentence idea has to find nineteen minutes somewhere.

Short videos are where the waste is

The single most useful thing in this measurement contradicts the way most people choose between two search results.

30.2% of videos under ten minutes were not worth their runtime. None of the fifty videos over an hour was. The share drops with every band and then flattens out around one in eight.

This is the same effect we found in the padding census, seen from the other side. A subject that genuinely takes an hour fills an hour. A subject that takes four minutes and is being delivered in eleven has to pad, and the padded eleven-minute video is the one that ends up in this bucket.

Which makes the habit of preferring the shorter of two results a coin weighted against you. A short video is cheap to watch, and that is a different property from being worth watching.

The zero in the last band is real, and it is also fifty videos. Do not read it as a rule; read it as the direction the whole chart is already pointing in.

The title is the tell, and you can see it before you click

We score how accurately a video’s title describes what is inside it. Grouped by that score, the verdict separates harder than it does by subject, by length, or by anything else we measure.

Where the title describes essentially a different video, 72.1% were not worth the runtime. Where the title is accurate, 5.6% were. That is a thirteen-fold difference on a signal that is visible from the results page, before anything is played.

We are not claiming the title causes the outcome. The reasonable reading is that both are downstream of the same decision. A video optimised for the click is optimised for the click throughout, and the title is simply the part of that you can see in advance.

The practical version: the promise in a title is a sample of the video’s judgement about your time. A title that is careful about what it claims usually belongs to a video that is careful about what it spends.

Getting to the point immediately is not a good sign either

We also measure the second at which a video starts delivering, and the relationship between that and the verdict is not the straight line you would expect.

Videos that begin paying off within the first thirty seconds were skipped 26.4% of the time, against 14.1% for the ones that took between thirty seconds and a minute. Past two minutes the rate climbs again, to 35.6% for the videos that need more than five minutes to arrive.

So the curve has two bad ends. A long wait is the obvious failure. The instant start is the less obvious one: a video that has nothing to set up is often a video with nothing behind the opening either, and the thirty-to-sixty-second band — long enough to say what this is and who it assumes you are — is where the material that respects your time actually lives.

The same shape appears in the level of the material. Videos pitched at beginners were skipped 28.8% of the time; the hundred pitched above intermediate, 6.0%. We have written that measurement up separately in the advanced census, and the two findings are the same finding: the crowded end of the shelf is where the losses are.

What the verdict does not mean

It is not a review. It is a measurement of one exchange — this runtime against this content — and the same video can be a skip for one person and the right thing for another. Somebody who wants the introduction is well served by a video that restates the basics for eleven minutes.

It is not about the channel. Every figure here is per video. A channel with a skip in our library has one video we would not spend that time on, and nothing follows from that about the next one.

It is not a rating. We do not publish stars, we do not rank channels against each other, and the number on a video is our own reading of that video, not YouTube’s and not the audience’s.

How this was measured, and what it does not say

We measure explanatory YouTube videos on several axes and reduce them to one decision: is this worth the runtime it asks for. That decision is ours. Method and its limits are in what value density means.

The sample is not YouTube. It is 3,243 English-language explanatory videos in twenty subjects, assembled to build a recommendation library rather than to be representative of the platform. It skews towards material somebody thought was worth measuring, which should make the skip share too low rather than too high.

Some of the subject bars rest on small samples. Travel, gaming, comedy and DIY each carry fewer than eighty measured videos. Read the top and the bottom of that chart as an ordering; the middle is closer than the bars suggest.

Padding is measured on explanatory material only, so that row of the table rests on 3,002 of the 3,243 videos. The verdict, the title accuracy and the runtimes cover all 3,243.

The figures are frozen at 25 August 2026. Scores are recalculated as the underlying measurement improves, so the live numbers on the shelves will already have moved slightly from these. This edition will not be edited; the next census will be published separately.