We Analyzed 1.3 Million YouTube Videos. Here’s What We Learned About YouTube SEO

youtube ranking factors banner

We analyzed 1.3 million YouTube videos to better understand how YouTube’s search engine works.

Specifically, we looked at the correlation between ranking factors — like views, comments and shares — with YouTube rankings.

We learned a lot about YouTube SEO. And I’m sure you will too.

Here is a Summary of What We Discovered:

1. Comments appear to be an influential ranking factor. We found that a video’s comment count strongly correlates with higher rankings.

2. Longer videos significantly outperform shorter videos. The average length of a first page YouTube video is 14 minutes, 50 seconds.

3. We discovered that video views have a significant correlation with rankings.

4. The number of shares a video generates is strongly tied to first page YouTube rankings.

5. There’s a moderate correlation between a channel’s subscriber size and rankings. This means that even small channels have a chance to rank their videos in YouTube.

6. Video likes are significantly correlated with higher rankings.

7. “Subscriptions driven” has reasonably strong correlation with rankings. Therefore, videos that result in new subscribers have an advantage in YouTube search.

8. We found a very small relationship between keyword-rich video tags and rankings. This could represent the fact that YouTube can now understand video content without the help of metadata.

9. Videos that contain an exact match keyword in their video title appear to have a slight edge over videos that don’t.Β This means that including a keyword in your title may improve your rankings by a slim margin.

10. We found zero correlation between keyword-optimized video descriptions and rankings.

11. HD videos dominate YouTube’s search results. 68.2% of videos on the first page of YouTube are in HD.

I have detailed data and information of our findings below.

Video Comments Have a Very Strong Correlation With Rankings

YouTube encourages creators to publish videos that maximize engagement. Needless to say, comments are a strong indicator that people are engaging with your video.

But does YouTube use comments as a ranking signal?

Our data suggests that they do:

youtube comments chart

As you can see in the chart above, the more comments a video has, the higher it tends to rank. Considering YouTube’s emphasis on user engagement, this result isn’t a big surprise.

Key Takeaway: Videos with lots of comments tend to rank best in YouTube.

Longer Videos Outrank Short Videos

When it comes to video SEO, should you create short videos? Or are you better off with longer videos that cover a topic in-depth?

We analyzed our data to find out.

Our data shows that longer videos tend to significantly outrank short videos.

video length chart

In fact, the average length of a video ranking on the first page of YouTube is 14 minutes, 50 seconds.

What’s happening here?

YouTube has publicly confirmed that a video’s total watch time is a key ranking signal.

youtube-creator-blog-screenshot

Also, in 2015, Google was granted a patent for an algorithm that uses “watch time” as a ranking signal.

In short, YouTube wants to promote videos that keep people on YouTube for a long period of time. Longer videos accomplish this best, hence the preference for longer video content.

Another theory is that longer videos provide more overall value in a single video. This is true for “how-to” videos as well as for content designed to entertain. The value that longer videos provide may encourage more interaction signals (including comments and likes) that ultimately impact rankings.

In fact, if you do a cursory search of popular keywords, you’d be hard pressed to find a short video (< 3 minutes) ranking highly in the search results.

youtube search results 2

Key Takeaway: Longer videos perform best in YouTube search. The average video on the first page of YouTube’s search results is 14 minutes, 50 seconds long.

Video Shares Are Strongly Tied to Higher Rankings

Google has consistently denied the fact that social signals play a role in their algorithm.

However, YouTube’s algorithm works independently of Google. So there’s a possibility that YouTube uses shares from social media networks like Facebook, Twitter and LinkedIn as a ranking factor.

In fact, we did find that shares have a strong correlation with higher rankings in YouTube:

total video shares

It’s important to note that we used YouTube’s public share report for this analysis.

video shares

Why is this important?

One of the major issues of using social shares as a ranking signal is that they’re easily gamed. Anyone can can hand someone a few dollars in exchange for sharing a piece of content 100 times on Facebook.

This isn’t the case on YouTube.

Unlike sharing content using a webpage’s social sharing icons, YouTube knows which users share video content…and where they share.

youtube social share icons

This tracking makes this signal much harder to game.

Combine that with the fact that YouTube encourages publishers to create highly-shareable content (and that YouTube reports shares in YouTube Analytics), and you have a strong possibility that the relationship between shares and rankings is more than a chance correlation.

Key Takeaway: Highly-shared videos outrank videos with fewer shares.

A Video’s View Count Is Significantly Correlated With Rankings

Video views used to be YouTube’s #1 ranking factor.

The thought was that: lots of views=popular video=quality video.

However, YouTube discovered that views often serve as a poor indicator of video quality.

So they changed their algorithm to emphasize factors like audience retention and engagement:

youtube emphasize watch time

However, we discovered that a video’s total view count continues to have a significant correlation with rankings.

video views rankings chart 2

It appears that you still need a critical mass of views to rank in YouTube. (In fact a YouTube engineer stated that, while views aren’t as important as they once were, YouTube still uses them).

That’s because, without views, your video can’t generate the other signals that YouTube uses to evaluate your video’s quality (like total watch time and comments).

But at a certain point, views have diminishing returns.

That’s why you often see high-quality videos rank above lower-quality videos (even when the lower-quality video has significantly more views).

example of low video view video ranking

Key Takeaway: Video views are significantly correlated with higher YouTube rankings.

A Channel’s Subscriber Count Is Moderately Correlated With Rankings

We found a moderate correlation between a channel’s total subscribers and rankings:

channel subscribers chart

This is good news if you run a small or new channel.

Unlike Google, which seems to have a preference for big brands, YouTube is more likely to rank content from “the little guy”.

For example, for this popular keyword, videos from two small channels outrank a video from a channel that has over 2 million subscribers:

ranking with low subs

This type of result isn’t uncommon on YouTube.

(Of course, channels with millions of subscribers have an edge. But our data shows that this advantage isn’t as significant as you may think).

Key Takeaway: Channels with lots of subscribers have an advantage in YouTube. However, videos from smaller channels consistently outrank videos from popular channels.

Videos With Lots of Likes Outrank Videos With Fewer Likes

It’s no secret that YouTube prefers videos that engage their audience.

And video “likes” serve as a powerful engagement signal. After all, likes are a crowdsourced way of evaluating how the YouTube community feels about your video.

That’s the theory. But what does the data say?

Our study revealed a significant correlation between likes and video rankings:

youtube likes and rankings chart

This suggests that YouTube may use likes as a ranking signal.

However, as you know, correlation doesn’t always mean causation.

Videos with lots of likes are also likely to be high-quality. And high-quality videos generate other ranking signals (like audience retention) that YouTube values.

Key Takeaway: YouTube may use likes as a direct ranking factor. Or it could be that heavily-liked videos generate other signals that YouTube truly cares about.

Videos That Result in New Channel Subscribers Rank Higher Than Videos That Don’t Generate Subscribers

If someone really enjoys a video on YouTube, what are they likely to do? Subscribe to that channel so they can see that channel’s future videos.

In other words, a video that encourages lots of new subscribers is a sure sign of quality.

Not only that, but getting new subscribers is an extremely hard metric to game at scale.

Sure, you can get a few people (or bots) to subscribe to your channel after watching a video. But it’s much more difficult than generating thousands of fake views or likes.

Knowing that, its likely that YouTube uses “subscriptions driven” as a ranking factor.

Our data did indeed show a significant correlation between “subscriptions driven” and higher video rankings.

subscriptions driven rankings

As they do with shares, YouTube displays the number of subscriptions driven underneath each video:

subscriptions driven

(Publishers can choose not to show this information publicly).

Like with most metrics, you can boost the number of subscribers your videos generate by creating world-class video content.

However, you can also ask viewers to subscribe:

subscribe CTA

I’ve found that a clear call-to-action to subscribe significantly boosts my “subscriptions driven” on each video.

Key Takeaway: “Subscriptions driven” has a reasonably strong correlation with higher YouTube rankings.

Keyword-Rich Tags Have a Weak Correlation With YouTube Rankings

In the early days of online video, platforms like YouTube relied on metadata to understand your video’s topic.

For example, YouTube would analyze your video’s title, description, tags…even your video’s filename. Essentially, the more text you could attach to your video, the better.

Today, YouTube can “listen” to every word of your video (without needing you to upload a transcription):

youtube automatic transcription2

Knowing that, does YouTube still use video tag metadata?

We found a weak correlation between keyword-rich video tags and rankings:

keyword in tag chart

While tags don’t appear to be as important as they once were, our data shows that they still make a small dent. So it makes sense to use them.

(Also, YouTube recommends that you use descriptive tags. This suggests that they still use tags to understand the content and context of your video).

Key Takeaway: Including your target keyword as a tag may help with rankings. But the overall impact of tags appears to be small.

Keyword-Optimized Titles Are Slightly Correlated With Rankings

Traditionally, you video’s title was piece of metadata that YouTube put a lot of emphasis on.

However, we found that including an exact keyword in your video title only has a slight potential impact on rankings:

exact match title

These findings could mean a few things:

It could be that YouTube has de-emphasized the importance of video titles. However, this seems unlikely as YouTube has stated that: “Titles contain valuable information to help viewers find your videos in search results.”

What’s more likely is that YouTube has developed a deeper understanding of a title’s meaning (beyond simple keyword matching).

In other words, they may use a less-sophisticated version of Google’s semantic search. If so, YouTube wouldn’t need to see a specific keyword in your title to rank you for that query. A synonym would do the job.

In fact, its common to see videos ranking well in YouTube for popular keywords…even when they don’t contain the exact term in their title.

youtube search results example

Key Takeaway: Using your target keyword in your title may help you rank for that term. However, the relationship between keyword-rich video titles and rankings is very weak.

There’s No Correlation Between Keyword-Optimized Descriptions and Rankings For That Term

Does including a keyword in your video description help you rank for that term?

According to our data, keyword-optimized descriptions don’t have any impact on rankings:

keyword in description

This finding contradicts a common “best practice” of video optimization: keyword-rich descriptions.

There are a few possible explanations for this finding:

First, like with titles, YouTube may not require an exact keyword in your description to understand what your video is about. For example, let’s say that your target keyword is: “how to grow tomatoes”. Using terms in your description like “growing tomatoes” and “the best way to grow tomatoes” may work just as well.

Second, there’s the possibility that YouTube uses “keyword appears in a video’s description” as a ranking signal, but it’s so small that we weren’t able to measure it. In fact, we found several videos with no description at all ranking highly on the first page. This implies that your video description isn’t nearly as important as user-generated signals (including views and “subscriptions driven”).

Third, it could be that YouTube now ignores video descriptions as a ranking factor. This is unlikely as YouTube states that: “Well-written descriptions with the right keywords can boost views and watch time because they help your video show up in search results.”.

Despite this finding, I still recommend writing keyword-rich descriptions.

Why?

An optimized description helps you show up in the suggested videos sidebar, which is a significant source of views for most channels.

Key Takeaway:Β There’s no correlation between keyword-optimized titles and rankings for that term. However, I still recommend writing keyword-rich descriptions as they can help your video rank for related terms (and appear as a “suggested video”).

HD Videos Dominate The First Page of YouTube’s Search Results

Do high-definition or standard-definition videos perform best in YouTube search?

We discovered that HD videos appear significantly more often than SD videos on YouTube’s first page:

HD videos

This data can be interpreted in two ways:

First, it could be that YouTubers that create the best video content also tend to record in HD. Therefore, this is an instance of correlation only telling part of the story.

Second, there’s the possibility that YouTube has an inherent preference for HD video content.

It’s difficult to determine the full impact of HD vs. SD from our correlation data alone.

Regardless, the vast majority of videos that rank well in YouTube are in HD. In fact, 68.2% of all videos on YouTube’s first page are in HD.

Key Takeaway: HD videos are significantly more common than SD videos on the first page of YouTube’s search results.

Summary and Conclusion

I’d like to give a big “thank you” to Zach Russell for collecting and analyzing the data we used to make this study possible. I’d also like to thank Qi Zhao for helping us with statistical analysis.

And if you’re curious about how we conducted this study, here’s a link to our methods.

Now I’d like to hear from you:

Which result was most surprising to you?

Or maybe you have a question.

Either way, leave a quick comment below right now.

    1. You’re welcome, Pravash. Yup, this is a brand new study. We collected and analyzed the data over the last 2 weeks so this is what’s important for video SEO in 2017.

  1. This is great research, Brian! Something to be proud of for sure. Although YouTube has been around for a long time now we STILL don’t see too many companies leveraging this great media for ranking. Who doesn’t like INFO_taining your customers? Even if they do create a small biz video, they might not know how to rank it for keywords. That is where great guides like this one come in to help succeed online.

    1. Well said, Chris. I think the underlying issue is that, with video, there’s a barrier to entry that’s not there with a post. But as you pointed out, the upside of video marketing/YouTube is HUGE.

      1. Yes Dean, for SURE! I wonder sometimes if people realize that Google Owns YouTube. That being said, why are you not including videos into your marketing strategy? Huge Ranking Opportunities πŸ™‚

      1. Our short videos rank really well, many on page one on Google SERPS. My clients who are mostly in the real estate space- their videos always rank high for very short videos. It does depend on the niche you are in. It makes sense that “how to” videos are expected to be instructional and therefore 14 minutes is perfect. However, no one is going to get great audience retention on a tour of a house for sale that goes on and on for 14 minutes. Buyers want to see the photos of the house, inside of the house, they are not there to learn how to buy a house, they are there to buy a house, totally different niche. I think that Google knows the difference because we have a lot of videos on page one and page 2 that are less than 2 minutes in length. The other thing we do for retention is to link videos from one video right into a specific point in another video. This helps the buyers stay on the video longer because of curiosity.

        1. Noah Kagan is in the process of validating his channel on Youtube and many of his vids are short. Curious to know what Brian thinks about this. Might a certain niche, like Katerina’s one, choose to go with short from the get go?

    1. Yes..that’s true right now. But don’t you think that it depends on the audience and the points? I mean – if you’re trying to learn something, one very short video may not be very helpful.

  2. Hi Brian, another great post as I’m looking to start doing video this year!

    When it comes to comments, shares, and likes, have you seen any evidence that these aren’t just the RESULT of the high rankings, not the CAUSE of the high rankings? On your own videos, have you noticed that when these factors come first, those videos rank better than others that didn’t get that early engagement?

    1. Thanks Tom. This was a correlation study. So we only looked at how different metrics correlated with rankings. We’re not able to tease out cause and effect
      from our data. That said, from my own experience, I’ve definitely noticed that lots of comments tend to precede a high-ranking video.

    1. You’re welcome, Azzam. Yup, there’s a lot of misinformation out there about YouTube SEO. I hope our results can start a trend where more people use a data-driven approach.

  3. How confident are you in the results given the use of spearman correlation (thanks for sharing methods btw). You do supply some anecdotal evidence based on reports out of YouTube, but I often found myself saying things like, “well of course the comment count is highest on the higher ranked videos, it’s much easier to get comments when you rank well.” Any additional thoughts on this? Are there results above for which you advise greater caution, are there others you are more confident about? Would be interested to hear more.

    1. Caleb, you raise a great point. As with any correlation data, caution is warranted, correlation doesn’t mean causation, etc. And our study is no different.

      In terms of results I’m confident on, I use a combination of the strength of the correlation, statements from YouTube about what they value, and my own experience. And my experience is pretty much in-line with the data. The only one I was surprised to see was the effect of views as I see videos with fewer views outrank videos with significantly more.

      1. Hmm, very interesting, and I assumed there was definitely interpretation on your own experience. Thanks for sharing! Had you considered a multiple regression for analysis? Or did this data, and it’s collection, not lend itself to such analysis? This might offer greater insight into causal relationships if it is possible.

        1. We actually did consider multiple regression to try to tease out the individual effects of each ranking factor. It’s something we may pursue in a future study/update of this one.

          1. I had the same thought. The more views a video has, the more comments, likest, etc. it is likely to have. We need to look at the RATIO between views, and all other factors to see if those other factors have much influence.

      2. I’m not a stats person but I guess one way to look at the impact of comments on ranking is the timing of the comments. Often a video will get a flurry of comments when it is published but not so many after that – even though the video continues to accrue views and watch time. So in that scenario it would be the comments come first then the video is ranked highly as a result. What are your thoughts?

        1. Katie, that’s definitely a possibility. One of the weaknesses of a correlation study like ours is that it’s hard to tease out cause and effect.

      1. I think it could also be interesting to see what categories the videos fall into, like the channel’s main focus for example, marketing, entertainment, food, exercise, gaming ect. and if that has an impact on how YouTube ranks different video categories.

        1. Amie, that would be interesting. That said, our sample included hundreds of different categories. So they were represented in the data set.

  4. Hello Brian, the thing is that in the past the most popular videos were the short ones. Now… it’s exactly the oposite thing… Everyone wants to keep the users on their websites, decreasing the bounce rate. They’re not doing it so much for the users but to improve their profit.
    πŸ™‚

    1. You’re welcome, Jake. According to our data, a series won’t rank as well. Both from a “We’re YouTube and we want to keep people on our platform” angle and from a user experience point of view. It’s much easier to watch a single video than get the same info from 10 different vids.

  5. Fab article, thanks Brian.

    To me YouTube ranking is an enigma. I can upload one and it gets views, likes etc but upload another and it’s somewhere in the abyss. Part due to competition I am sure but I cannot find consistency. Your guide helps.

    One question actually… YouTube keyword research. Do you have any tips on finding those YouTube searches that carry search volume? All I use is the predicted search bar in YouTube.

    Finally, YouTube rankings in organic search. Any data you’d possibly share about this?

    Great thanks.
    Marcus

  6. Hey Brian,

    Awesome article with some surprising insight! Definitely going to take action based on these.

    The one that really caught my attention, it’s the correlation between keywords-rich description and ranking. Although I’ll keep using optimized descriptions (As you said), still, as a “little perfectionist” it’s great to know that πŸ˜‰

    Quick question, which tools did you use for this kind of analysis?

    Stay awesome,
    Sariel

    1. Thanks Sariel. That one caught me eye too. We had quite a few videos without any description at all rank well. So that was a sign that descriptions
      aren’t as important as they once were.

      In terms of tools, I’d have to ask my co-author (Zach) what he used. But from what I gather it was a Swiss Army Knife of different tools. This wasn’t easy to pull off.

    2. Hello Sariel, I figured I can chime in about the tools we used. To do the extraction, we used a combination of the Python library Scrapy and their companion headless web browser, Splash, to render JavaScript (its essentially a scriptable version of Google Chrome). Once we collected all of the data, we put it into a database and normalized the data. We then used the Python Library SciPy to extract the data and run the correlation algorithms. If you have any more specific questions i’d be more than happy to answer them.

    1. Hi Shad, there is definitely hope. I imagine other videos in that space also have no talking. So you’re all in the same boat

  7. Wow, just a great guide and free! The future of “SEO” (we’ll have to come up with a new name soon) seem to be outside of Google Search now, having so many other places to lookup content (Youtube in this case, Reddit, Pinterest… you name it).

    Thanks again!

    1. You’re welcome, Miguel. It’s true that other platforms steal a few % of searches from Google. But Google is still the dominant force in search by a mile

  8. Hi Brian,

    Thank you for sharing such a epic youtube ranking factors.

    From your every post i learns something new & intersting.

    Thanks again Brain for this awesome post!

  9. Thank you for this article, I am an SEO Analyst and I tend to do a lot of searching online to keep up-to-date with the SEO world and ever-evolving techniques. And I have to say, you’re site is one of the best out there offering top-quality content, and lots of it. Love your YouTube posts too, keep them coming! πŸ™‚

  10. A great study for brands looking for increased brand awareness or traffic to their YouTube channel.

    However, did you look to see if it’s still true that click through from YouTube to websites is low? If that’s still the case, it would seem that using YouTube videos as a method to drive website visits would not be recommended. What are your thoughts?

  11. Thank you for this detailed guide about YouTube ranking factors.
    It’s surprising to see your analysis of keyword-optimized descriptions not having any impact on rankings. I find it really hard to agree with this, based on my experience.

  12. Good timing! I just suggested to a friend who is traveling through Africa by the seat of his pants that he should start a video log. I’ll send him the link. Thanks.

  13. Brian, Great Post and Awesome! Content My team and I have been pushing video/SEO for almost 9 years now and Brian it changes constantly however, your study will be utilized to the fullest. I’ve followed you around for a while also and You are the MAN! keep up the great work and god bless you for sharing your study.

  14. So based on these findings would you recommend driving as much traffic as you can to new videos to increase views, shares, subscribes, and view time? Or is it best to let them rank organically and see if they sink or swim naturally based on the merits of the content.

    If so, what is the best strategy if you have a small audience?

    1. Jacob, it’s both. I recommend promoting your videos to your current audience (it’s great content anyway. So it’s a win for them). That can give your video an initial boost that helps a lot.

  15. Hey Brian. Thanks for this study. For me there’s something missing. What about CTR? Don’t you think that in 2017 Youtube use that like Google? ++ πŸ™‚

    1. You’re welcome, Bruno. I’m confident that YouTube uses CTR. Unfortunately that’s not public so it’s hard to analyze data on that.

  16. Hi Brian,

    Bookmarked it and will use it for sure!

    Got a little question for you:
    Do you think there is a correlation between subtitles and rankings? Does Youtube use it to understand the subject of a video and therefore make it rank higher?

    Thanks!

    1. Dany, great question. I actually don’t think they use them that much. They can transcribe videos with pretty good accuracy without them.

  17. Amazing stuff and I have bookmarked as well. Quick question: as video length appears to correlate with higher ranking, how do we weigh the significance of that vs. the data that shows view duration dropping off significantly after 2 min or so? Won’t the longer videos decrease % of views all the way through? Thanks

  18. Hi Brian, this is a great article. I only pay attention to facts with case studies. So thank you. I’m going to be in touch with you in the near future as I’m working on a new platform that literally does every single thing you mentioned and going to be recording my own case studies. I plan on sharing with you. So be on the lookout to hear from in the near future. Thanks again for your great article.

  19. Hey Brian,

    How about “latest uploads”?

    I’ve searched for “TVR1” (in Romania) and top 1-3 videos are uploaded within last 24 hours. They have only 2k views, 4-5 minutes length, few likes, etc.

    On 5th place is an 430 days old video, over 40 minutes in length, some likes and comments, etc.

    The first video (ID: 1Ap6uI-Nx5A) is ranking without having the keyword in title (TVR1).

    Some ideas?
    Thanks,

    1. Vali, good question. We actually looked a bit at that but didn’t find anything. New videos only get a temp boost. After that, older videos rank better because they have more comments, likes, watch time etc.

      1. I see… I’ll watch that video to see when will lose his rankings.

        1-2 years ago was a simple method to rank #1
        Step 1: upload a video on UNLISTED
        Step 2: boost with some views, likes, comments
        Step 3: make that video PUBLIC
        Once you made it public, it ranked on top.

        Any new tricks like that? πŸ˜€

  20. Are you sure about the duration of the video, that is 14 Min and 30 sec? As in many niches, it’s impossible to increase video length; then one need to add additional video minutes. It might reduce the interest of the viewer.

      1. Great but if you can check hairstyle, cooking type of niches the average time is around 6-8 minutes. My wife has a channel which has 70 recipes that fetched 130k views in 10 months in just three months work with 0 work and without further updates. There I see high duration videos receiving fewer view’s while less duration one’s getting more. But I agree with most information based niches this period is standard. I look forward to updating you if I can gather any information on it.

        1. Omkar, that makes sense. The data doesn’t necessarily mean “longer videos=better”. The #1 goal is to create an amazing video. And it sounds like your wife is doing that. I would keep going with what’s working for you.

  21. Excellent analysis of youtube SEO and video ranking factors.
    Just started our youtube channel, the information provided in this post gonna help me a lot.

    1. I also started implementing techniques described in this article on my youtube videos. I made some serious changes on my channel aswell on a few videos with low number of views. I will share my results as soon as I notice some changes.

  22. Awesome research! As a very small channel, I am testing the same theories as far as search and rank is concerned. I have 15 Subs, almost all of which were derived from one video that has viewer retention and comments! I set out to try and replicate the same results in future videos, and I landed right in between 2 videos of major YouTubers!! This video only had 8 views!! I am fascinated with how YouTube ranks and will continue to read your posts! Great work, sir!

    1. Thanks for sharing your experience, Joseph. Hopefully this study can help you re-create the success you had with that popular video.

  23. What a great article, Brian! Thanks, only a few people write such a quality article on such a topic

    I have a simple question, what should be the minimum length for writing a great description? I am told by so many people that your description should be of 500 words. Did you find any correlation between the videos that have a long description and the videos that have a short description?

    1. Aamir, good question. We didn’t find a correlation between description length and ranking. But long descriptions may help you show up for more “suggested videos”, so I prefer to write 250+ word descriptions.

  24. Found your e-mail in my Junk folder and almost deleted it before reading as I didn’t recognize “the sender”. Glad I didn’t as what a “gem”! Answered a ton of my questions about video marketing with Youtube. In fact I have paid for content with less valuable information about marketing with Youtube than what you provided for free in your report! Awesome content! Many thanks, Brian!

  25. Brian, great job! Have you figured out what weights certain factors have on suggested videos vs youtube search. It seem to be a different weights? You can promote video to the search or to the recommender. I noticed that certain strategies would get one but not another. What your thoughts and insights on that?

    1. Oleks, great point. “Suggested video” works different than YouTube search. So that would be interesting to look at.

      1. Brian, I hope you’ll find some time for an deeper look into this and share results with the community. It’s hard to understand one without the other. Thanks again for the report.

  26. Nice summary you provided here. I’m just not sure how did you know which videos of those 1.3M are videos with real views and real comments? Conclusion is always right if you make it from a proper sample. Otherwise results are not adequate. Anyways thank you for your research.

    1. You’re welcome, Ostin. We weren’t able to distinguish between fake and real views (even YouTube struggles with that).

      1. Oustin, Brian, this is in fact interesting point. From the one hand Youtube has up to 80% of unidentified visitors – so they cannot track it. On the other hand, Youtube won’t care that much about fake views. It’s in the interest of Youtube so that additional liquidity is maintained. It is still a problem for Youtube, but it’s not violating Youtube’s operating philosophy. The advertisers are paying anyway. Can you think of any strategy/metrics to figure it out on the mass scale? I would say disproportional likes/views or comments/views ratio might point out to the fake views video.

  27. Great article! Awesome to have something that is worth while to show exactly where to focus based on stats. Any stats about publishing consistently? This is something that has been recommended in various places too. Perhaps it’s just to keep your subscribers engaged in your videos.

    1. Thanks Amy. We didn’t look at publishing frequency as a ranking factor. But from my experience it doesn’t matter nearly as much as putting out amazing content.

  28. Great post Brian, but I echo some of the same sentiments shared by other comments:

    e.g. Of course top ranking videos have more likes, comments etc

    Would love to see the data analyzed in other ways in a future post.

  29. Wow Brian! Thanks for creating this amazing guide.

    Your insights and stats are super useful and I’ll definitely keep them in mind as I create more videos on Youtube.

    One quick question/comment, in your post you mentioned, “Google changed their algorithm to emphasize factors like audience retention and engagement.”

    Do you think that when people replay sections of your videos will play into the “retention and engagement” of those videos?

    Thanks man for creating awesome stuff time and time again.

    1. You’re welcome, Jared. I’m honestly not 100% about that. Replaying sections could mean that you weren’t clear or that section of your video is amazing. So I’m not sure how YouTube treats that.

    2. Jared, you might think of the following thing: video gets boosted each time when people are closing the popup ads or skipping the video ads. I haven’t figured out the way to use it though. Perhaps better placement of the ads might be done to facilitate this action.

      1. Thanks Brian and Oleks,

        I don’t use any Youtube ads. I was just looking at some of my video metrics and saw that there were certain parts of my videos that were rewatched multiple times by people. I’m guessing that they were trying to really follow what I was teaching in that section.

        I noticed that these two videos get more views and show up higher in organic Youtube results, so I thought the “rewatching” might play a factor in “retention and engagement.”

        Thanks so much for replying.

  30. Another quality article So surprising to read the description research. Perhaps as Google’s AI team get better at extracting what is being said in the video, then the video is the key to the description. It sort of makes sense as I rarely click on ‘more’ to see what has been written in the description section. Google’s click tracking would show them that too. Just a guess on my behalf.

    1. Ashley, it’s a good guess. As YouTube gets better at “listening” to videos, the description gets less important.

  31. Brian, great to see you applying your analytical approach, and step by step break down to another area. This is good timing for me (and I think quite a few others) as we start to look more seriously at Youtube as part of the mix. Great quality as always.

    1. James, you’re not alone: more and more people are jumping into video marketing (and YouTube) this year. So hopefully this helps.

  32. Brian – also a suggestion: could you look at adding LinkedIn as a share option on your articles? I tend to keep facebook for personal stuff, don’t use Twitter much, but I do share the occasional good article with my network on LinkedIn.

    1. Hey James, if you sign up for the newsletter you’ll get notified every time I publish something new πŸ™‚

  33. Great article Brian!
    Something worth noting, having multiple tags and keywords helps a video to rank for multiple keywords. I like to be found for a few phrases and mixing it up in the title has achieved that for me (and the description). What’s your research say? Still works, or not any more?

    1. Tim, we didn’t find that tags help you rank for specific keywords. But more tags can help you show up as a “suggested video” in YouTube, which is a massive
      source of views for most people.

  34. I have discovered a few days ago that “Audience retention” is the most important factor in my Youtube channel.
    The videos with higher retention percentage are the ones that get more views on my channel. Not only for higher rankings on search pages… mainly because these videos tend to get more views as suggested video on related videos.
    Thanks for this great post!

    1. Gaston, I’ve 100% seen the same thing. Audience retention and total watch time are huge. Optimizing your videos for terms people
      search for on YouTube is the other 50% of the equation.

  35. AMAZING research and intel on the YouTube algo! How did you examine and manage data for 1.3 million YouTube videos?

  36. I completely agree with Brian – creating a call-to action for subscribers really work. When we watch a video song of a popular artist, at the end that celebrity asks us to subscribe to their channel. Overall this is an awesome article. Cheers!

  37. Curious at the breakdowns for search/video types. Ecom vs Gamer vs Beauty vs How To vs Etc. I saw someone else mention how things are different depending on your vertical. Much to the sadness to YT Trolls… I wonder if they know that their comment wars and thus engagement only help videos and not really hurt them #oops

    Lastly – interesting data and awesome job digging in!

    1. Thanks Dave. Every industry is different to a certain extent. But I don’t think the algo treats videos in different verticals differently. So I’d say the findings apply
      to all industries.

  38. Although the 14 minute video length average is an interesting statistic, it in itself does not say much, there is no clear trend in the data there (maybe a slight upwards trend in length, but it is very faint) so I’m not sure we can draw a conclusion that it is a good video length. I would like to see the standard deviation added to that same graph. For example, is the average 14 mins but 4 mins and 24 mins also do well? You probably need a histogram as I suspect the distribution will not be symmetrical. In that case rather add percentiles (e.g. 10 25, 75 and 90).

    1. Jurgen, that’s a good point. If you check out our methods you’ll see we found a significant relationship between video length and rankings.
      That said, we may do a deeper-dive into the data in a future update.

  39. I like the fact that my new channel have opportunity to rank higher.
    Awesome content, I liked it soo much that I am going to share it in my WhatsApp group….

  40. Really amazing………………your article is really informative and helpful for me and others too because as you have described everything about YouTube SEO in descriptive, simply yet effective too. You’ve done a great job, doing analysis of 1.3 million videos are not an easy task.
    Thanks to share this precious article with us.

    1. Alex, we only found a correlation between HD and rankings. I’m not 100% sure it’s a ranking factor. That said, HD videos look better, and 4k looks great, so they may perform better.

  41. This was a nice read. It’ll be of a great help if you could also write about how to practically achieve the ideal scenario of at least a few of these YouTube SEO factors. And I’m surprised about longer videos ranking higher. It makes sense for YouTube, but sometimes shorter videos are better for users, aren’t they? eg. how-to videos

  42. Hi! Great study.

    Have you tried to use the median instead of the mean/average?

    Some outlier values, like really long videos, can have a great influence on the average.

    1. Thanks Diego. We did correlation as well as average. And we found a correlation between longer videos and higher rankings. The average isn’t so much a hard-and-fast rule as a way
      of showing that longer videos tend to do best.

  43. Hi Brian,
    Great Post, Keep up the good work.

    Just a question. How much time does it take to Rank a Youtube video?

    I have applied all your tactics, but the video is stuck at the 2nd Page.

    Anyways, Thanks for the awesome content.

  44. Hi Brian, another great post! Going to bookmark it!
    I want to ask you a quick question – Is any there any way that thumbnail can increase CTR on videos?

    Anyways, Big Thanks for this detailed research guide πŸ™‚

  45. OK. Great study, first of all! Hats off, Brian for putting so much time into this. This looks like a bit different from the textual content’s SEO factors. For example, I see you have pointed out in another comment above that ‘frequency’ of posting doesn’t seem to correlate with engagement. Or is it not measured?

    Also, I believe that the video cover thumbnails play a significant role to improve the CTR, in addition to adding a link to your other videos throughout the ongoing video, without distracting the viewer (and making them disable the annotations if it’s all over the face).

    Does it help to have a co-existing blog/a website with a domain name that’s same as the YouTube channel name and then also direct some traffic to the YouTube channel/vice-versa by ‘repurposing’ the content (text, images, slides)?

    Great job, again!

    Regards,
    Arun Sarathy

    1. Thanks Arun. We didn’t look at publishing frequency, so I was speaking from personal experience there. Yes, re-purposing content like that is a smart move.

    1. Hi Nick, we didn’t look at the influence of backlinks. My take: YouTube doesn’t use them as a ranking factor.

  46. I really appreciate you making these long guides for us . It really give us a competitive edge over out competition for you tube and in SEO in general . I used to send a lot of time in keywords and trying to make my videos short . But come to think of it my longer videos perform way better then shorter videos . So Ill be making longer videos for sure . Thanks again Brian .

  47. Hey Brain,

    Thanks again for another epic SEO resource. I believe Google is flat out lying when they say social sharing doesn’t correlate with rankings. Why wouldn’t it? It’s an excellent measure — people are “voting for” a piece of content by sharing it. I love how Google and Youtube are measuring engagement factors over overt techniques to optimize. We have to get smarter right along with the algorithms.

    Off to share!

    1. You’re welcome, Ayodeji. I agree that social signals have merits (that’s why search engines may use them). But the downside is that they’re easily gamed.

  48. Hello Brian

    Sorry for my bad english

    I write from Spain, I have seen your tutorials and articles and I think they are great.

    I am thinking of passing my articles to videos for Youtube

    What platform or software do you recommend?

    1. Pascual, what do you mean “passing my articles”? Do you mean repurposing your articles into video content?

  49. Grear guide brien i loved it. But i have one question, how to reach people to watch our video, then they will decide to share, comment, like our video as ranking factor. If i made excellent quality video but they is no views on it how it will rank and people will decide it is good or bad video? Thank u

  50. Hello Brian! I like your content, but i have to say that your graphics do not seem to be legit.

    With 1.3 million videos analyzed your graphics should be smooth and not have any spikes. What is your method for building graphics?

    1. Thanks Giovanni. Those spikes happen in any data set (even large ones). Most are smooth but spikes happen due to random chance.

  51. how if a long videos with short watch time ( 15 minutes video with 1-2 minutes average watch time ). is that wasted!? it could be bad for rank!?

  52. The backlinko research is extremely useful! I think the most positive take away from this being that YouTube still lets smaller subscriber accounts rank well and hasn’t caught big brand fever…

  53. Awesome work Brian!

    My wife has a cookery channel on YT and we have slowly grown the subscriber base over time. We started off doing them daily (which is alot of work!) I knew about watch time but had no idea about HD (we recently switched from 720p to 1080p) and the engagement makes perfect sense so I’ve asked her to specifically ask for feedback/comments when she does future videos.

    Thanks again!

  54. Hey Brian,

    Willing to read your informative post and I totally agree with your amazing points. You have scattered amazing tips regarding YouTube and really have high impact on our online businesses.

    YouTube is higher up in the sales funnel, so gearing keywords related to the discovery stage works best and it can be great choice for afiliate marketing. Ranking in YouTube is great, but ranking your video in YouTube and Google is even better. Eventually, thanks for sharing your valuable ideology with us.

    With best wishes,

    Amar kumar

  55. Hello Brian and first of all I have to say, that I’m a big fan of your work and that you are really one of the best content marketer I know (And the best SEO expert too!).

    I love this study you made, because I’m very intersting on this and do a lot of searching on this for me, manely with the help of Vidiq.

    But I’m a little disapointed with some point of your study, because like you said, user engagement is a big signal and I personaly think, that it would make a lot more sense to analyse it differently.

    You analysed the total of like, comment, share and new suscribtion but not the ratio of like pro view, comment pro view etc.

    And I think this is much more important for youtube than the total number of like, comment, share etc because they want to show the best video to the people, and if for example the total number of like was more important than the ration like/view, then the search result will show only old video.

    And secondly, same a bad video will get a lot of like etc if it rank first, because of the lot of view. In this case, a good video will outrank it because having better like/view ratio etc, same without having a lot of view. But yes, not allways and it can take a lot of time for this to happen and because of this, would love it, if you extend your post to show the correlation between ranking and like/view ration, comment/view ration etc.

    Would love to see the correlation with video age too.

    Still, love your post and all the works you do! Thanks a lot for all those things,

    Antoine

    1. Hi Antoine, you raise some excellent points here. YouTube may in fact use a ratio of likes, comments etc. But I’m not sure as we didn’t measure it. Maybe in the 2.0 version of our analysis

  56. Hi Brian,

    It’s been a long time i visited your site but today i returned again just because of your awesome article on youtube ranking factors.

    These are awesome factors that we should keep in mind while creating and posting videos on youtube.

    Thanks

  57. “10. We found zero correlation between keyword-optimized video descriptions and rankings.”

    This is not surprising at all! This is how it works in my eyes… word gets around that having keywords in your description helps with ranking… every marketer abuses it for a couple of years, then Google pulls it out of the ranking factor bucket… Sort of like meta keywords.

    Because we abuse things, they become invalid lol..

    This is such a great blog post, I very much appreciate the time put into the case study. Lots of great takeaways.

    1. Elijah, that’s a fair point. The other thing is that if everyone is optimizing it’s not a competitive advantage anymore

  58. My company sells spy products, as spy cameras, we have few videos and a lot of keywords to rank, i was asking myself if is a good strategy publish more than one time the same video, but with different titles, because some of our keywords are not close related, so, when we publish with some titles, we do not rank well for others keywords. What you think about it?

  59. Brian, thank you so much for this great research!

    The only thing i am wondering about is if my channel correlates with the ranking of my videos.
    For example:
    Does the number of uploaded videos on my channel take an influence on my ranking?
    Does my upload activity?
    Or maybe my general activity, like writing comments, watching videos a lot, creating playlists with foreign videos, etc.?

    1. Kai, we didn’t measure that. But in my experience most of the critical factors are related to the video itself.

  60. Hey Brian,
    Amazing research here. One question I have is did you see any correlation between YouTube title lengths and rankings or engagement? That would be something really valuable to know. Thanks!

    1. Good question. We didn’t look at title lengths. Do you find long titles get better or worse engagement?

  61. This is very interesting Brian. Two questions: Did you find that Page Authority mattered in a video’s SERO position? Is there an advantage to having backlinks directed at YouTube video URLs?

  62. Good article but I think the rivalry between Microsoft and Google goes further.
    Las week I published a video and using Tubebuddy published in FB, great so far, but as it was published in FB I did no get any video views in Youtube. This week did basically the same but instead of using Tube buddy to do the publish in FB, I did it myself and place a link to my video in Youtube. No matter how I boosted this new post, it never reached the expected amount of people.
    Please help me understand how to tackle this issue.

  63. Fantastic post, Brian!

    I’m a huge fan of the way you tackled the topic. You show valuable data that proves your points. Surprised to how big of a role interaction plays in the equation.
    Will definitely have this in mind every time I market on YouTube. Thanks for sharing!

    Rasmus

    1. You’re welcome, Rasmus. I’ve been doing a ton on YouTube lately and noticed that engagement (and watch time) was key

  64. You nailed it per usual, Brian and team!

    Based on your findings, what actions did you take to change your Youtube marketing strategy? What were you doing before that you’re now doing differently?

    1. Good question, Amanda. Kind of a lot to outline in a comment. But let’s just say I’ve put more weight into getting engagement and boosting my watch time with longer videos.

  65. Much of this is useful information but there is also some that is not really all that useful. The reason I say this is because all you show is a correlation between some of the sets of data, but the causation is not discussed. High ranking videos have a lot of shares and comments, but I would argue that this is probably BECAUSE they are high ranked, not that they are highly ranked because of the shares and comments – People find them more and therefore share and comment more on them.

    It might have been more useful to look at likes/shares/comments per view or per minute of watch time.

  66. Excellent post Brian! Thanks for sharing this with the community. I was wondering does the sentiments (Positive or Negative) analysis of the Video comments has any correlation with the rankings. Do we have any data or analysis around video comments sentiment analysis. It would be interesting to see any correlation of positive comments and higher rankings.

    Cheers!
    Nikhil

  67. Hello brian, really great stuff to hear but before I thought shorter video will rank better in search engine. Now, I know but how people are staying in specific video for long time? they are always busy ? then how this is possible? for ex: when we have a content with long words which describing the services and something else the user might get boring?

  68. Great article – and have bookmarked it.
    However quite depressing for a new start up who has good content but won’t have the subscribers or comments ever to get noticed. I feel I am locked out by the algorithms!

    Is it still possible for the little guy (57) with good content to be able to start up a channel, get noticed and succeed?
    Would be interested in your views.

    1. Colin, we didn’t analyze that. I haven’t seen embeds impact YT rankings directly. But they can help you rank in Google (embed=backlink).

  69. @Brian … If all the spearman correlation data is taken into account … all the factors mentioned accounts for 71.49% …. what about the balance 28.51%? What ranking factors could this 28.51% include?

  70. A mind-boggling research Brian! 1.3 million, just wow πŸ™‚

    I do have a question, do you have the slightest idea how automated playlists work? For example, I play a video (let’s say a song) and in the upper right corner, I choose a mix that goes along that video so I would have a flow that follows along the lines of the video I’m currently playing. Now after several videos, 20 or so, I get a song that seemingly has nothing to do with the first one that was played (I played Sting, and the song No. 21 was Justin Bieber). I thought about why could this be happening, is it my geolocation, do people listen to both Sting and JB so they serve me with the similar playlist or something third?
    What are your thoughts on that?
    Blake

  71. Thanks Brian! Very in depth and well put together. These ranking factors make sense. I wonder if it’s industry specific, too — for example I am in the English teaching industry and my videos tend to be short…like 5 minutes, because English learners might get burnt out listening to long English videos. But come to think of it, a lot of ESL videos rank high that are over an hour long. Anyway thanks for sharing these findings! Ryan

    1. Hey Ryan, with 1.3 million videos we ultimately represented every industry. That said, as you pointed out, every industry is a bit unique and different.

    1. Frederik, I have noticed the “huge” YT placement. They seem to use it mostly on long tails where the search is a 1:1 match for the video.

  72. Bookmarked this one! As always Brian your post was incredibly deep, well-researched, data-driven and easy to read. You truly are the master of amazing content. I got some great value out of this. One thing though – I din’t see you mention links to the videos themselves. I would think that the number of backlinks to the specific video would also play an important role, especially from related industry sites?

  73. Hi Brian,

    Great to see you broadening your reach and I’m sure it helps keeps interest levels high.

    I’m not personally into YT ranking at the moment but will rely on this incredible resource in the near future.

    Slight deviation, but one thing I think will add a tremendous amount of value to backlinko in the near future is: Amazon Search. Now they have even implemented local services similar to Google Guaranteed. Amazon Search optimizers are going to be (if not already, forgive me knowledge gap) in extremely high demand. Maybe a great idea for your next case study.

    Anyways, thanks again for sharing all this great knowledge.

    Best,

    Alan

  74. Great article, thanks! One question though: does Youtube change its search algorithm so no and then? Most of my views come from, like you stated, recommended videos (90%+). In the past few weeks the views have been dropping to 50% of what they more or less were before. This is due to the fact that Youtube isn’t recommending my videos that much anymore (I can see that also at the amount of recommended videos from my channel when I play one of my own videos). Since I haven’t changed the description and tags, it should be the Youtube algorithm that has been changed. Do you know this happens?

  75. Great article Brian. On a somewhat related tangent, when embedding YouTube videos into pages I have found it can hurt the loadspeed of the pages. Obviously this is poor for SEO. What are your thoughts on this? I have taken to using a plugin that embeds a still from the video with the play button on it as an image. If the user clicks it then loads the YouTube video. It has done wonders for loadspeed.

  76. Brian! The blog post is just amazing, and your research is outstanding. From my personal experience, I came across the longer videos rank well and the video which got a large number of likes and share also play and important role.

    But beside this I learned many new things useful in your post and would implement in my video campaign.

  77. Hi Brian, how much effort do you put in to get comments on your videos? Would it be a smart strategy to send an email out to your list with a link to a video, and then ask them to post their comments and questions on it? Or have them find the video on search and treat it like an Easter egg hunt. They have to post something in the comments when they find it, and they get a lollipop or something. If YT is looking for engagement, that’s easy enough to manufacture if you have a list. I value your thoughts on the topic, thanks!

    1. Matt, comments are pretty important. In my experience, the best way to get more comments is to ask your video viewers to comment on something specific. That way they don’t have to think. They just need to answer your simple question.

  78. Brian,

    Just wondering if you have these details from the million videos analyzed… What percentage of videos on the first page of YouTube are in 4k/2160p or 2k/1440p?

    Thanks either way!

  79. Yet another wonderful post on YouTube. Thanks, Brian.
    I would like to know one point. Is there any way to identify at which part of my video viewers drop-out the most? I feel this will help me to avoid such mistakes in next videos.

  80. I think subscriber base and number of views have a correlation,once you build a good following,any time you drop a video it gets lots of views almost instantly through notification and emailing outreach.that is why you see a video released just a few days ranking on first page already.

Leave a Comment

Your email address will not be published. Required fields are marked *