[0:00] How does the YouTube Shorts algorithm
[0:02] exactly work? Why do Shorts get a spike,
[0:04] but then the graph becomes flat? How
[0:06] will the algorithm decide whether your
[0:08] Shorts will get views in millions or
[0:10] thousands? The answer to all these
[0:12] questions will be found in this video.
[0:14] Let's begin. So, we know that big
[0:16] YouTubers keep getting views on their
[0:18] Shorts, but in your case, the views are
[0:20] 100 or 1,000. But after that, the graph
[0:23] becomes completely flat. So, let's
[0:25] understand why this happens. Todd
[0:27] Sherman, who is the product lead of
[0:29] YouTube Shorts, which obviously means
[0:31] that he can tell us a lot about the
[0:32] Shorts algorithm. In an interview, he
[0:35] told that whenever you upload a Short,
[0:37] the algorithm finds a random seed
[0:39] audience for a Short. This seed audience
[0:41] is basically a small group of viewers
[0:44] with whom the algorithm wants to test
[0:45] your Shorts. Because the algorithm
[0:47] doesn't know much about your channel and
[0:49] its content yet. The word random is
[0:51] important here, because it's possible
[0:53] that the seed audience isn't your target
[0:55] audience. The algorithm randomly
[0:57] selected these audiences to test your
[0:59] Short. So, let's say that seed audience
[1:01] doesn't engage properly with your Short.
[1:03] So, after a while, YouTube stops pushing
[1:05] your content to audience, which means
[1:07] there will be almost no views. Let's
[1:10] assume that by chance that seed audience
[1:12] is your actual target audience. And if
[1:15] your content is good, then naturally
[1:18] viewers will engage well with your
[1:19] Shorts. This is the positive signal for
[1:21] the algorithm. After this, the algorithm
[1:24] finds more viewers that match the seed
[1:26] audience. If this new group of viewers
[1:27] also engage properly with your Short,
[1:30] then your Short will be pushed even
[1:31] more, and this will keep on repeating.
[1:33] So, now you must be understanding how
[1:35] the algorithm of Shorts exactly works.
[1:38] Almost every Short creator has a
[1:40] question in their mind that if our Short
[1:42] was performing well in the initial
[1:43] phase, then why did the algorithm stop
[1:45] its views? You have to understand that
[1:47] the algorithm of long video and short
[1:49] video is not the same and can never be
[1:52] the same. In case of long-form video, a
[1:54] view means that the viewer is
[1:55] intentionally choosing to watch the
[1:57] video. However, this is not the case
[1:59] with shorts because here there is no
[2:01] concept of thumbnail. Here the viewer
[2:02] watches the video through the short
[2:04] feed. This means that if the algorithm
[2:06] wants to know if your video is good or
[2:08] not, then the algorithm has only one
[2:09] option. That is testing your videos to
[2:11] seed audience. The next question that
[2:13] will come to your mind will be what are
[2:15] the important matrices which algorithm
[2:18] decided that your short has performed
[2:20] well and has been shown to other people
[2:22] besides the seed audience. The first
[2:24] metric, how many choose to view? Just
[2:26] like in long video, we have CTR.
[2:29] In the shorts, we have how many choose
[2:31] to view metric, also known as viewed
[2:34] versus swiped away. This metric tells
[2:36] you how many percentage of viewers
[2:38] intentionally stopped to watch to your
[2:39] short and how many swiped it without
[2:42] watching the video. The percentage of
[2:43] viewers who stopped intentionally to
[2:45] watch your video, only their views will
[2:47] count. Your next question will be how
[2:49] much should this viewed versus swiped
[2:51] away percentage be? So, you must know
[2:53] Paddy Galloway. He's a YouTube
[2:55] strategist and he has worked with a lot
[2:57] of big creators including Mr. Beast. So,
[3:00] last year Paddy and his team conducted a
[3:02] study where they studied more than 5,000
[3:05] shorts across 33 YouTube channels in
[3:07] different niches. According to his
[3:09] study, shots whose viewed percentage was
[3:12] 60% and less than that, those shots
[3:14] didn't perform as well. The best
[3:17] performing shorts which went viral,
[3:19] their viewed percentage was on an
[3:20] average between 70 to 90%. Again, this
[3:24] is just a study. So, I recommend you to
[3:26] not obsess over these numbers. But yes,
[3:28] you can take a rough idea from this that
[3:31] if your viewed percentage is 80% or more
[3:33] than that, then there's a high chance
[3:35] that your short will go viral. Even I
[3:37] personally, all the viral shots I've
[3:39] seen, their viewed percentage is above
[3:41] 75%.
[3:43] Now, I know that the maximum creators
[3:45] who are watching this video, their
[3:46] viewed versus swiped away ratio is not
[3:48] that good. So, now let's understand how
[3:51] to improve this number.
[3:53] So, as I said, in the case of long
[3:54] videos, we have CTR. Similarly, in the
[3:57] case of shorts, CTR is equivalent to how
[4:00] many choose to view. If we want to
[4:02] improve CTR, we work on the title and
[4:04] thumbnail of the video. That is, the
[4:06] thing that the viewers see first. So, if
[4:09] we talk about shorts, then which is the
[4:11] thing that the viewers see first? You're
[4:13] right, the first 3 seconds. So, it's not
[4:16] wrong to say that the first few seconds
[4:18] of your short, that is, its hook, plays
[4:20] the same role as in the case of long
[4:22] videos of the title and thumbnail. Now,
[4:24] in a long video, we have 30 seconds to
[4:26] hook the viewer. In shorts, we have only
[4:29] 3 seconds. This is why in both these
[4:32] formats, creating a hook is very
[4:34] different. But even though long-form and
[4:36] shorts need different types of hooks,
[4:38] the foundation of both is the same. That
[4:41] is, you have to spark curiosity in the
[4:43] viewer's mind in some form.
[4:45] You can create this curiosity with a lot
[4:47] of things. For example, depending on the
[4:50] type of video you're creating, you can
[4:51] say some lines in your hook, which will
[4:53] immediately stop any viewer who is
[4:55] mindlessly scrolling. For example, in
[4:57] the start of the short, you can say,
[4:59] "I'm sure you don't know about ABC."
[5:02] Here, simply replace ABC with something
[5:04] about which you have a video. Example,
[5:07] "I'm sure you don't know about this
[5:09] YouTube setting. I'm sure you don't know
[5:10] about this recipe. I'm sure you don't
[5:12] know about this new Minecraft update."
[5:14] And this will work as a very good hook.
[5:17] But it's not necessary that you say
[5:19] something in your hook. Because the more
[5:21] powerful a line is to grab the viewer's
[5:23] attention, the more powerful and
[5:25] intriguing visual. Let's take a look on
[5:27] Mr. Beast's shorts. Can you slice a
[5:29] bullet with a katana? Yes, sir. In this
[5:31] short, Mr. Beast shows the visual of the
[5:33] first 3 second, in which bullet almost
[5:36] hit the katana.
[5:37] But just before the climax, that scene
[5:39] has been cut so that viewers can watch
[5:41] the video till the end. But if you
[5:43] really want to make your shorts viral,
[5:45] then just creating a hook is not enough.
[5:47] You also need a good audience retention.
[5:50] When we make long videos, the focus of
[5:52] YouTube is on watch time. Because the
[5:54] algorithm wants your content to keep the
[5:56] viewer on the platform for as long as
[5:58] possible. That's why if a video is 1
[6:00] hour long and an average viewer is
[6:02] watching it for just 15 minutes, even if
[6:05] its retention is bad, the algorithm will
[6:07] promote that video a lot because it
[6:09] keeps the viewer on YouTube for 15
[6:11] minutes.
[6:12] But in the case of shorts, it's
[6:13] different. Here, YouTube doesn't care
[6:16] about AVD. Instead, the algorithm looks
[6:19] at audience retention. If the audience
[6:21] retention is 25%, then it won't work.
[6:24] The best-performing shorts have a 90% or
[6:27] 100% retention, or even more if the
[6:30] viewer watches the short again. But it's
[6:32] not like the shorts algorithm doesn't
[6:34] care about AVD. According to a study by
[6:36] Paddy Galloway, shorts with more AVD
[6:39] performed better as compared to others
[6:41] with lower AVD.
[6:43] So, keep this in mind if you're thinking
[6:45] of making 5-second shorts to get 100%
[6:48] retention and go viral. I'm sure after
[6:50] hearing all this, you're confused as to
[6:52] what's the best length for shorts.
[6:54] Should we make 10- to 15-second shorts
[6:56] to get good retention? Or according to
[6:58] Paddy's study, should we make long
[7:00] shorts? I'll give you the answer that
[7:02] YouTube employees gave, and that is make
[7:05] a short that aligns with your story.
[7:08] Some niches will have quick, instantly
[7:10] rewarding shorts, while in other niches,
[7:12] you can use storytelling to build a
[7:14] narrative arc with a beginning, middle,
[7:16] and end, and you can make longer shorts.
[7:19] So, yeah, the best length for YouTube
[7:21] shorts is in which you can perfectly
[7:23] deliver your content.