What’s up Insiders, it’s Rachel here and I’m a product manager on YouTube’s recommendation system. This is a series where I’m talking about how our recommendation system works, and today, I am going to cover 10 frequently asked questions about our recommendation system, as well as some tools and analytics that you can use to analyze your performance. Well, let’s get to it. So these are 10 of the most frequently asked questions that I received from creators about our discovery system or the algorithm. I often get this question that if one of your videos under-performs, will that hurt your channel or your channel’s future performance?
This is fairly unlikely because what our systems really learn from and care about is how each video and how it performs when it’s recommended to viewers. Our system relies much more on these video-level signals because we want to match viewers to videos, and so what our systems learn from is how videos perform when they are recommended. And so, one video underperforming, very unlikely to probably hurt the future trajectory of your channel.
Another important point is that like, videos would never be able to go viral if this wasn’t the case. If we only looked at a channel’s previous performance to indicate future potential, we would never have viral videos and that would be a shame.
And so individual video performance is what’s most important. Okay. This one I’ve covered before, but does changing a video’s title or thumbnail re-rank a given video in the algorithm? Yes, that actually might happen. However, it’s not because our systems are identifying that you’ve changed your title or thumbnail, it’s because they’re learning from how viewers interact with a video in recommendations.
So if you change the way that your video looks, viewers will interact with it differently. And if you change your title and thumbnail so that more viewers click to watch, your video could rank higher and receive more impressions.
If you change the way your video looks and fewer viewers choose to watch, it may rank lower. So yes, it may change your video’s rank if you change the way your video looks, like your title or your thumbnail, but the act of changing it probably won’t make any difference. Okay, this one I know has been controversial, we’ve spoken about this before, but if you take a break in uploading, will that hurt your channel performance?
We absolutely encourage creators to take breaks because we found there’s no correlation between break length, the days between uploads and the change in views after that break. What that means is, there’s no consistent drop in viewership for channels if they take a break. That’s what we’re really looking for when we did this study, like, do we have a bias in our systems that could potentially mean that channels will miss out if they take a break? What we found is that there is no correlation.
So this is not a guarantee that a channel might not drop after you take a break.
Channels may increase or decrease, and what really influences whether or not a channel takes off after they take a break is one, what’s happening with their audience in between that time span, do their interests change, do they start watching more of other stuff? And then the second factor is how good your content is when you come back. Sometimes it takes viewers a little bit of time to get back into the habit of watching something. If you think of your favorite TV shows that have gone off for a season and a new season comes out, sometimes it might take you a few weeks to start watching it again because you were watching something else. And so when you take a break, there is a lot of things happening with your audience, but most creators just cannot keep producing the same level they can without feeling burnout at some point, and so breaks are absolutely essential for your own wellbeing and maintaining your creativity.
Your viewers will be happy if you’re happy too.
Okay, this one is, my video views or impressions are low, but it looks like viewers are engaging with my video just the same. My click-through rate and average view duration are totally typical, what’s going on? Now, the most common scenario here is when your video is being outranked by other videos outside your channel on YouTube. What’s happening when you’re comparing your own performance within analytics is you’re comparing against yourself.
But the way our recommendation system works is we’re ranking videos relative to all other videos on YouTube that a viewer might watch. So, if this ever does happen, which is fairly rare, most of the time, it’ll line up with your own performance.
But if your click-through rate and average viewer duration look high and your views and impressions are still low, the majority of the time what’s happening is other videos outside your channel are ranking even higher for those viewers. Does monetization impact video discovery? No, our recommendation system does not know which videos are monetized and which are not.
However, both our advertising and discovery systems pay attention to maturity of content.
So does your video contain racy or sexual content, profanity, or graphic violence? Those signals are often used to inform, is a video relevant for a given viewer and appropriate for them? And so, that is used by your system, monetization is not. Our recommendation system is just focused on the long-term and recommending videos that viewers find satisfying.
This is another common question. If I upload my video as unlisted and later flip it public, would that hurt your performance? No, it won’t. Our recommendation system learns from audience behavior on public videos when they’re recommended. So totally use unlisted feature.
What matters for our recommendation system is how your video performs when it is offered to viewers in recommendations. Okay, publish time. Is it better to post earlier in the day, late at night, like when should I publish my videos? The publish time is not known to impact a video’s long-term performance. The way your recommendation system is designed is to match viewers with the videos they’re most likely to watch, regardless of when that was uploaded.
Now, it can very much matter for live streams and premiers when you need your audience to be on YouTube to watch it because it has a limited timeframe. And for that, we recommend using the when is your audience online card under the audience tab of YouTube analytics. If you go to your homepage today, you will notice that many of the videos may be weeks, months, sometimes years old, really, we’re looking for the most relevant, best recommendations regardless of upload time. Tags. We’ve spoken about this a lot.
Are they important? Not really.
What’s most important is what viewers use to decide whether or not they wanna watch a video, which is the title and thumbnail. So if I’m gonna say all the things that creators should spend time on, you know, it’s the things that viewers make decisions based on on what to watch, title and thumbnail, then your description, minimal or no time on tags.
We mostly use those to correct for common misspellings.
Channel location, like which location you set your channel in. That is not used to inform recommendations. So use whatever country you want. That’s not gonna inform which viewers are recommended your videos. Upload frequency.
So how much do I need to upload? You shouldn’t feel pressured to upload daily or weekly. Again, we did lots of analyses and found that views growth is not correlated with the time between those uploads. So upload how many videos works for you. Different audiences have different types of demand for different quantities of content, but you do not need to upload daily or weekly to be recommended.
All right, so that’s it for FAQ. If you learned anything from this section, focus on your audience and the algorithm or our recommendation system will follow. We have a viewer-focused system. That’s what it’s designed to do. So the features in YouTube analytics that I wanted to call out that will help you focus on your audience are the audience interest card of course, so what videos and channels your audience is watching.
This can help you identify title and thumbnail ideas, collaboration opportunities, figure out like what types of videos your audience likes and how their interests change over time.
Now, I know a lot of creators don’t have enough data to be here, and I wanted to clarify the eligibility criteria, even if it’s not that actionable. If you have a channel that makes content for kids, you’re not gonna see any results because that data is protected. And then if you do not have over 50 overlapping viewers with enough videos or channels, you may not be shown in some of these cards. We wanna make sure that we have enough significant data to share so creators can act on this.
So if you don’t yet have access to these cards, I would use generally like searching for similar content that you produce on YouTube, looking at what content is receiving more viewership and seeing if you can learn from your peer channels or similar creators.
New and returning viewers. So these are the two new metrics we’ve launched that tell you, do you have an audience of similar viewers coming back to your channel to watch more, or do you have a more casual audience where viewers are coming in and out of your channel and then watching some of your videos? And I wanted to clarify the kind of content strategies you can use based on this data. So, the first one in purple across these are returning viewers.
These are viewers that are not new to a channel. They’ve watched the channel before and the turquoise line with new viewers. So viewers who are watching a channel for the first time at least in the past year. So if your channel looks like this first example where it’s receiving most of its viewership from returning viewers and few viewership from new viewers, in order for this channel to grow, they’re going to have to pivot their content to appeal to a new audience outside of their core.
So some ways that you can do that are trying to do collaboration or experimenting with a new topic or format.
For example, if I’m a gardening channel, I could try some adjacent topics, like experiment with houseplants or beekeeping because that’s slightly similar to, I think my existing audience would be interested in it, but I think that branching out into those new topics might help me build a larger audience. In the middle, this is if you have a decline in returning viewers. So that purple line you can see dropped a little bit at the beginning of the chart.
This often happens when you change a favorite host that may be presenting on your channel, like you have a personality that people grow to love and they watch them all the time and then a new face shows up. Sometimes it takes viewers a little bit of time to adjust.
It also happens when a series might end. So you had like a popular series that’s going on and then you release the last episode. Sometimes again, viewers might’ve only been watching you for that one series and then they will move on. Now if this happens, you want to focus on retaining that existing audience or reengaging people who were watching you in the past. So think about why these viewers started watching your channel in the first place and what you could make to keep them coming back to watch more.
So maybe it is a new series, but it is about a similar topic but it’s then a new format that excites you. The last one is the most actionable and the most exciting to me, I think. So this are channels that, in this case, this channel has a fairly casual audience, they’re getting most of their viewership from new viewers, that turquoise line is on top. So people are probably going in watching these channels, like a few videos, but they’re not coming back. But if you get a spike in new viewers, like all of a sudden, lots of new viewers are discovering your channel, that’s the most actionable opportunity that you could really capture in what video you make next.
So let’s say lots of new viewers are discovering you because you released a popular video about European history, then you wanna make a series about European history. Let’s say you had a viral video about the Renaissance, then make three more videos about the Renaissance if you want those viewers to return to watch more. So if your channel looks like this where you get an increase in new viewers, really focus on that retention of your new audience. These are some channel comparison features that we’ve also launched, which are in the deep dive of analytics. Also did a whole video on this, I’ll link it in the description if you’d like to watch more.
All of these dots are videos at the exact same timeframe. So in analytics, you can go in and compare all your videos at the same age and make fair comparison. So comparing videos at the first 24 hours, in this case all videos are 24 hours old, you can also choose for seven days or first 28 days. And this is really to give you an idea of what content is working on your channel, if it’s really, broadly appealing, make more of that and then analyze the groups of videos that aren’t receiving as many views and impressions and asking yourself, are these just more niche interest, is they’re like a smaller audience for these? Or maybe there’s something in common about their title and thumbnail styles that you could improve on.
So, comparison tools like this are really good to get an idea of what’s working, what can you improve on over time? So, if you’ve watched all of the episodes in this series about how the recommendation system works, I really just want you to leave knowing three things, focus on your audience, that is the key to success on YouTube.
If you know what viewers like and you’re adjusting your content along with their interests, recommendations and impressions will follow. The second is, it’s always crucial to keep experimenting over time.
If you make the same thing forever, at some point viewers interest will change.
So constant experimentation is essential to sustain your audience’s interest over time. The last one is really what we just covered about new and returning viewers. It’s that if you want to build an audience on YouTube, you really have to do two things, you have to attract new viewers and retain the ones you have. Okay, we covered a whole lot in this series, but if there are any of your questions that I did not address our answer in these videos, please drop them below and I’ll try and incorporate them in future videos. So, as always, keep it real.
https://dash.sparkloop.app/signup?origin=newsletter_radar&aff=290086c5
Tok Mate Launch Phase 2 / Weekend
https://jvz2.com/c/1399091/411066/
Discover more from Making Money Is Easy
Subscribe to get the latest posts sent to your email.
