The YouTube algorithm decides what people see on YouTube 70% of the time. And according to the Pew Research Center, 81% of American YouTube users say they regularly watch videos recommended by the algorithm.
If you are a developer working on getting more YouTube views or a brand developing your YouTube marketing strategy, the platform's recommendation algorithm is very important. How do you optimize your channel and videos to work with, not against?
YouTube isn't usually known for being very transparent to developers or advertisers about how the proverbial sausage is made. In this article, we'll take a look at the history of YouTube's priorities when it comes to helping viewers discover new videos. We will explain how the algorithm works, as well as the latest changes to the YouTube algorithm for 2020.
Bonus: Download a free guide detailing the exact steps One creator received 23,000,000+ views on YouTube with no budget or expensive equipment.
A brief history of the YouTube algorithm
The first video from YouTube was uploaded in 2005. Fifteen years later, users are uploading 500 hours of video to the platform every minute.
How do 2 billion users find what they want to see? The short answer is that it has changed over the years. But here's the long answer too:
2005-2012: Show number (a.k.a. clicks)
For the first seven years, YouTube rewarded videos that got clicks, not those that kept users busy.
Obviously this system had a tendency to show people a lot of clickbait: misleading titles and thumbnails multiplied. Users would click but then feel cheated, probably a little annoyed, and then partially abandon videos. Eventually YouTube realized their user experience was going down the drain and switched tacks.
2012: time (a.k.a. view duration)
In 2012, the platform announced an update to the recognition system that will be used to identify the videos that will actually be viewed. By prioritizing videos that grab attention (and the overall time a user spends on the platform), YouTube has been able to reassure advertisers that they are delivering a valuable, high-quality experience for people.
Meanwhile, YouTube also encouraged developers to stop bothering with algorithm optimization (i.e. making videos shorter for a higher retention rate, or making them longer for more playback time).
Instead, as is still the case today, YouTube encouraged people to only "make videos that people want to see".
2016: machine learning (a.k.a. the algorithm)
In 2016, YouTube released a whitepaper that made some waves. In it, product engineers described the role of deep neural networks and machine learning in the platform's recommendation system.
(Source: Deep Neural Networks for YouTube Recommendations, 2016)
Despite all the impressive technical terminology, this white paper was of course not a panacea. You can read it, but even if you understand it (or get your smart friend to explain it to you), it isn't Coca-Cola's secret recipe. (It's more like Coca-Cola announced the drink was so tasty because it goes through a carbonation process and it also contains sugar.)
We don't currently know how many details are under the hood of the YouTube algorithm. However, we do know that it measures perceived viewer satisfaction to create an addicting, personalized stream of recommendations.
2016-2020: Borderline content, demonstration and brand safety
In the past few years, YouTube has had a lot of questions about the type of videos its algorithm shows up and promotes (or does not promote).
YouTube CEO Susan Wojcicki says YouTube takes its responsibilities seriously and tries to reconcile a broad, fair range of opinions to ensure that downright dangerous information does not spread. For example, YouTube says that changes to the algorithm in early 2019 resulted in 70% less watt time for "borderline" content. (Borderline content is defined as content that does not entirely violate the platform's community guidelines, but is harmful or misleading.)
It's a complicated problem as it touches on every problem: from white supremacy to coronavirus. For example, YouTube developers said in March 2020 that the platform had demonstrated videos that even hinted at the existence of the coronavirus. YouTube's position, meanwhile, is that it wants to support a wide variety of opinions (i.e., how governments should respond to the coronavirus) but not the dangerous ones (i.e., videos that say the virus is a hoax or that drinking hand sanitizer is will heal). Wojcicki stated that "when people search for coronavirus topics on YouTube, an average of 94% of the videos they see in the top 10 results are from high-profile channels."
Regardless of where you are, developments are still ongoing. Therefore, this is an important discussion that both creators and advertisers need to be informed about.
If you're a creator, remember that just because the algorithm rewards the content you create with high visibility and ad revenue, YouTube won't flip and demonize your channel or video if your content crosses the line to make advertisers find objectionable .
In the meantime, advertisers need to know that their sneaker ads aren't funding anti-Vaxxers or conspiracy theorists. The YouTube algorithm in its current form is used to demonstrate borderline content, mainly to protect brands. At the same time, YouTube says that it may never be able to guarantee 100% brand safety.
How will the YouTube algorithm work in 2020?
According to YouTube, the algorithm is essentially a "real-time feedback loop that adapts videos to the different interests of each viewer". It decides which videos are suggested to individual users.
The algorithm has two goals: Find the right video for each viewer and let viewers look further. Therefore, the algorithm monitors user behavior as closely as it does video performance.
The two most important places that the algorithm affects are search results and recommendation streams.
How the YouTube algorithm affects search results
Unsurprisingly, the videos you get when you search for "carnivorous houseplants" are different from the videos I get when you search for "carnivorous houseplants". Search results are based on factors such as:
- The metadata of your video (title, description, keywords) and how well it matches the user's query
- The engagement of your video (likes, comments, time)
How the YouTube algorithm affects the recommended videos
The recommendation stream is a two-fold process for the algorithm.
First, videos are rated by assigning them a rating based on performance analysis data. (Scroll down for our list of all known factors.)
Second, videos are compared to people based on their watch history and similar people.
The idea is not to identify “good” videos, but to match viewers with videos they want to watch. The end goal is for them to spend as much time on the platform as possible (and therefore see as many ads as possible).
There are three other places where the algorithm has a major influence on the recording:
- Your YouTube homepage
- Trending videos
- Your subscriptions
- Your notifications
How YouTube determines the algorithm
While we're not at Google, here is a list of all of the factors YouTube has mentioned in its various public discussions about the algorithm over the years.
When a video is rated, the algorithm checks its performance:
- Whether people click on a video (a.k.a. Impressions vs. Views: Thumbnails and titles are important here)
- How much time do people spend watching a video (time or retention)?
- How many likes, dislikes, comments or approvals a video receives (a.k.a. engagement)
- How quickly a video becomes popular with snowballs or not (this is called the viewing speed, growth rate).
- How new a video is (new videos may receive extra attention to give them a chance to play snowball).
- How often does a channel upload new videos?
- How much time do people spend on the platform after watching a video (session time)?
When a video is assigned to a potential viewer, the algorithm examines the personalization:
- What channels and topics have you seen in the past?
- What have you been doing in the past?
- How much time do you spend watching?
- How many times has this video appeared for this person?
- What do you not see?
7 tips to improve your organic reach on YouTube
Here is our list of the bound and true methods to play well with the algorithm.
1. Optimize your video description text
Contrary to popular belief, that block of text under your video isn't just a place to link to your social networks (although you should definitely be doing it, too). It also helps the algorithm show your video when users search for your topic. So make sure you preload the first sentence with a clear, keyword-oriented description of your video.
As in the examples above, make sure you:
- Use natural language, not keyword salad
- Focus on a keyword or two and repeat them in both your description and your title
For more information, check out our complete guide to SEO Optimized YouTube Descriptions, including tips on how to identify your keywords.
2. If something works, flush and repeat
To increase that leverage on YouTube, as these five unexpectedly interesting YouTube channels learned about, you need to pay attention to what your audience wants. That means paying attention to your analysis, but also your gut.
The YouTube algorithm wants to offer users more of what they have liked in the past. Experiment cleverly, take feedback from your audience, and give everyone time to adjust.
(Source: Dad, how am I?)
For example, this local dad launched a channel during the pandemic lockdown, and its premise – answering questions people usually ask their dad when they don't have one like him – has attracted 2.4 million subscribers in two months. It's a unique, serious, and emotional value proposition, and it's especially impressive because this channel has been successful in a vertical content industry (i.e., how-to-DIY videos) that seemed pretty saturated.
Also note that he reads picture books once a month, which leads us to conclude that the algorithm rewards those who make their viewers cry.)
3. Publish often
The number of videos and the frequency of uploads are an important factor for the algorithm and especially for the YouTube home screen. (This personalized list of new and interesting videos is similar to Instagram's Explore page.)
If you can increase the quantity without losing quality, go for it. The more videos you post, the greater the chance that you'll hit the right nerve. Maybe you can turn that one big hit into a series. Or, with little effort, you can introduce a new weekly feature that fits into your brand's established niche. like a reaction video on Tuesday or a Wednesday study with me or a Thursday Twitch stream.
4. Make your videos public when your audience sees them
Timeliness is an important ranking factor for any social media algorithm we can name (the Instagram algorithm, the Twitter algorithm, the Facebook algorithm), and YouTube is no exception.
YouTube's notification feature will ping your subscribers when you upload a video. This is definitely most effective when done when you are looking for something new to look at.
Overall, though, we encourage you to take a look at your YouTube analytics to determine the optimal time of day or week to release your latest masterpiece. In many cases, it also means planning your YouTube videos in advance.
5. Keep viewers busy throughout the video
Another important performance metric for the algorithm is display time. You may see advice advocating making your videos shorter or longer, but just make them as interesting and fun as possible.
For example, this six minute video of a brutal raven talking to her best friend is solid across the board. We assume that not only the clickability, but also the retention (a.k.a. viewing time) contributed to the fact that the views of this video skyrocketed. (This was the broadcaster's breakout video, which hit 4 million views when their average is typically well below a million.)
Once you've got people checking through to the end, you can use end cards and / or playlists (see # 6 on our list of ways to get more YouTube views) to suggest that they watch your next video . Because nobody needs a recommendation algorithm if people trust your recommendations, right? Correct.
6. Get involved with your community
We'll never stop saying that. Reply to your comments. Talk to your people. Remember, the algorithm "knows" whether you are having meaningful conversations or just paying lip service to improve your vanity data.
If you are able to have too many people to reply to, you can always make an appreciation video. Like in this video where this illiterate person hears all the compliments people give him.
If no one pays you tens of thousands of compliments every week about the cute noises you make, that's fine, too. With Hootsuite, you can skip the video and manage conversations for your channel. How so:
7. Turn viewers into subscribers
According to YouTube, your channel's subscribers provide a number of important initial signals that determine the success of your video. In other words, these fans are the test field – if they love it, the algorithm is more likely to show the video to new eyeballs.
For more tips on how to get free YouTube subscribers, check out our list.
Grow your YouTube audience faster with Hootsuite. Schedule videos and moderate comments in the same place that you manage all other social networks. Try it for free today.