"How DOES The YouTube Algorithm Work?" — Creator Myths Busted!

so if you don't like this video are you just going to not recommend it to anybody uh it's not my decision hello insiders I am Renee Richie your friendly neighborhood Creator liaison and today we are talking to Todd opre Todd what is it that you do at YouTube so I lead the growth and Discovery team so there's a couple outliers that I often hear from creators and one of them is they have a channel they're doing a certain kind of video and then they just decide to do another kind of video and it takes off

it it blows up it becomes viral they get so many views from it they start to worry that it's going to distort how the algorithm sees Their audience yeah so I think it's important to understand that for the most part the algorithm for Discovery is focus more on individual videos than like trying to come up with some comprehensive average performance of a channel similar to how you don't want your videos to be shown to the wrong viewers such that they won't engage and it'll hurt your performance we don't want to be showing the wrong videos

to the wrong viewers either and so we try as much as possible to understand that video's audience obviously if there's high demand for that topic yeah um and you choose not to make videos on that topic more then maybe you're you won't continue to have that outside performance um that's something you just need to decide um based on what your interests are and and what the audience interests are and find the best match uh for you so I don't need to go through and delete all those videos no I wouldn't recommend deleting those videos because

there will be a subset of that audience who will be interested in your other content yeah generally I wouldn't advise deleting videos um because I think you're just removing opportunities at that point because I was afraid of this at first too that if I do one bad video it begins like the downward spiral of my channel but it sounds like you're not going like based on channel and the Creator but more on those video by video basis yeah if if your last video wasn't so great and your next video is great we want to we

want to realize the potential of each video yeah we wouldn't want the system to just decide oh one bad video this this channel is not able to create anything because that's not the reality next video could be the best video in the history of YouTube it could one of the other things that you said to me that really really changed my perspective on YouTube was that YouTube does not push videos for creators YouTube pulls videos for viewers could you share that a lot of creators their kind of uh interaction with YouTube is they make a

video and then they upload it and and then out from the system they get impressions and Views and things like that so I think it's natural for creators to think of like oh well I give YouTube my video and then YouTube pushes my video out to a bunch of people but it's actually uh more the reverse so yes you upload your video to YouTube but what triggers a viewers seeing an impression of your video isn't some you know YouTube plan that says oh look here's a new video from Renee I'm going to I'm going to

give this to 10,000 people and then see what happens that's not how it work for T ISRO all the videos yeah instead what happens is we don't we don't calculate recommendations until a viewer actually uh appears on YouTube they open up the app or they open up the YouTube homepage it's at that moment that the recommendation engine is engaged and the challenge that the recommendation engine is given is given this viewer who's coming to YouTube from this type of device at this time of day with this history of of videos they've watched before what are

the best videos that we could show them right now and so we are very centered on that viewer and finding the best videos for that viewer that also ties into something else you told me which I really loved and that is the algorithm doesn't give up on video so creators shouldn't give up on them either yeah I see creators focusing so much often on their real time Analytics imediate yeah and I get it like you spend a lot of effort to like make a video you wanted to you know reach as many people as possible

and you're looking at it and you know I'm always hearing from creators talk about oh this video was a nine out of 10 and yeah you know and they're just kind of down about it what I often try to say to creators to reassure them is like how many times have you seen where a video started out as maybe a nine out of 10 nine out of your last 10 but then you know a few weeks later ended up being a 5 out of 10 or a four out of 10 so while I understand the

focus on wanting to get the immediate uh excitement from the audience when you upload it one of the great things about YouTube is we don't limit recommendations to just videos that have been uploaded this week for example and so over time there's always going to be more opportunities for your video to get you know connected again if if there's like renewed interest or just kind of a different Trend flows through the ecosystem sure look at your early dat data maybe uh if something's underport forming to your expectations consider whether you might package it better the

better thumbnail that conveys the value of the video or better title But realize that the ultimate success of that video isn't always determined in the first 24 hours there's this other concept that creators have sometimes about the algorithm being punitive like your channel starts to get fewer views or you take a break that suddenly you're put in the recommendation Penalty Box so to speak I'm sure that's something that you hear often how do you address that when creators bring it up we want to get every video out to the audience that would be most interested

in it we don't want to be putting in logic in the recommendation system that that draws the wrong conclusions about a video that just came out and so we want to be led by the audience and so we aim to not overemphasize historical data if that data isn't particularly predictive of each video and so that's why we tend to lean more on the per video you know performance and feedback uh than try to like come up with these like penalty boxes for channels we don't want there to be a a player in the Penalty Box

when they could have been scoring goals yeah right and so uh so we really try not to to add those sorts of rules and and enable videos to freely find their audiences I've heard you talk about the algorithm follows the audience but do creators have to think about that as well like reinvention and updating and trying new formats do creators have to follow the audience as well yeah I think that's really important I think shorts is a interesting case study in this space a lot of uh creators they may say like oh well YouTube is

pushing shorts on viewers and YouTube wants shorts to be more successful than long form that's not why YouTube has focused on Shorts YouTube is focused on shorts because audiences have let us there they've told us that they want more videos that are more efficient and get to the point faster they want more entertaining participatory content and we see them you know audience is voting with their behavior of the apps they're using to get that content and we feel obligated to offer that to the audiences and we want audiences that want shorts to be able to

find them but also if an audience member is happy with long form we want them you know to find that easily as well so um that's an example of you YouTube following the audience as the competitive landscape changes as a Creator I would recognize that there is a lot of attention some of which is shifting from long form to short form some of which is new uh opportunities to recognize that as part of their content strategy and then within a format there are also Dynamics there's there's you know creators who might say hey you know

I've been doing this the same content for a couple years now and it's not performing as well anymore and sometimes I'll look into you know work with them and encourage them to look at their audience Tab and see well what are some other channels your audience is watching and we'll find that oh there's a new channel that came out in this that that's also serving the audience and you know that channel wasn't around two years ago and and now you have to recognize that there's a bit of supply and demand going on and when there's

you know more Supply you know with the same amount of demand you may get less views and then you got to figure out well where is there unserved demand that you can maybe tap into um or where can you do deliver better on the demand that you started with if creators looked at their content strategy similar to like any small business or or business that's trying to figure out well you know how do I maximize my Revenue I I think about things like market share and competition and things like that so those Dynamics are are

very uh relevant in the YouTube ecosystem as well I make a video and then I see someone else has made a video that to me feels really similar but they get 10 times the views that I get or maybe like the topic I see like gets tons and tons of views but I'm not getting them when you hear about those kinds of things what are some things you look at to understand performance one of the things that I'll look at and encourage creators to look at on their videos is how did the uh viewers who

saw and impressions of the video in the subscriptions feed react because that feed it doesn't have any like special recommendations algorithms it's pretty much always the people going into the subscriptions feed you could go into traffic sources browse features and then subscriptions and most of the time when I do that um I find that videos that underperform also underperform when they're shown to subscribers in that in that feed which suggests to me that it's some issue with the content or the way the content was packaged you could also look at the the average V view

duration for that audience that subscriptions feed offers a bit of a control over other variables that can lead click-through rates to for example go down with a broader audience I think sometimes too like when I look at those videos I see that they're topically or structurally the same but the feeling I get from them are very different and that's not always as easy to measure yeah it's hard to measure and uh you know viewers can be finicky and and quirky and it's sometimes hard to uh to understand why some things are successful when similar things

are not I think this is something that's always been the case with with media uh even before YouTube DOD thank you so much if you have any comments if you have any questions please leave them down below and then keep it real