Uncovering YouTube Growth Signals and Trends - My Framework

This article is based on sections of a 40-page internal report written in 2023 for my startup’s team, offering in-depth insights and analysis of my YouTube channels.

If you’re running a YouTube channel, it’s important to evaluate your channel’s health and data for each video that you release, using such insights to continue improving and growing your content. However, the graphs and numbers you see only tell half of the story, and without proper context, could generate false implications on success/failure. These are some steps I take when doing a deeper analysis of the performance of my videos (with examples!), and how I turn these findings into actionable items.

Timeframes

Before diving into the analysis, we have to consider timelines. Typically, it’s good to watch your analytics in the first few days of an upload, as you can still make changes that could improve the video’s success. However, these changes are only “cover” changes, like new titles, thumbnails, and descriptions. Furthermore, the data you see may not be fully representative of the long-term success of the video, so it’s better to perform deeper analysis on videos that have been uploaded for at least 2 weeks. This way, you deal with less uncertainty and more data to make, ultimately, more informed decisions. Finally, check in on videos that have been out for a few months from time to time to understand long-term performance. In this article, I will be demonstrating the evaluation framework with two examples: A successful video and a flop, in the early-middle stage of its lifespan (around 2-3 weeks after uploading).

I will be demonstrating the evaluation framework with two examples: A successful video and a flop.

Video 1: Success

This video was about a piece of music gear, where I showcased some capabilities and short songs I made with it. It had a thought-provoking title for those that were interested in the instrument, resulting in a high click-through rate (CTR).

Video 2: Flop

This video actually took much longer to plan and make. It was supposed to chase two trends at the time: Christmas season music and super hard piano songs. While sharing a similar title and thumbnail to other successful videos at the time, it didn’t go anywhere. This video came out two months after the successful one.

1. Identify the Stage of Your Channel.

Before calling a video you made a week ago a “fail” solely based on view count, consider where you are in your YouTube journey. Just like in long-term investing, it’s important to zoom out and look at the wider picture. If on average you are getting more views on each video, then you are making progress! This is especially important to do in the early stages of a channel, as successes and failures can be widely ranged and an emotional rollercoaster. In the example, I had only made about 10 videos of varying quality and topic at that point, with views ranging from 100 to 4,000. Keep this perspective in mind when you are analyzing purely on numerical amounts.

2. Identify your Channel and Video Type.

Another important factor to keep in mind is the type of video and genre of content you’re creating. My channel was very niche, with successful videos sometimes far surpassing the top creators’ averages, just after a few months on YouTube. That being said, the flops also hit pretty hard, with very few people in the niche interested in new types of content I was doing. On the other hand, I hit over 500,000 views on a low-effort gaming video due to a wider audience.

Additionally, what kind of content is it? Are you making videos specifically to chase trends, evergreen content that will continue to be searched, or personality driven videos in hopes of slowly building a loyal fanbase?

Once these two things are considered, we can finally start looking deeper into the actual data for each specific video.

3. Identify the Data Trend.

There are common, recurring trends that happen with each video. These visual cues give us a high end view on the success of a video, but should only be one step in the analysis process. Here, we look at the reach these two videos got in the first two weeks using YouTube Analytics Advanced Mode, as we eventually want to focus on individual traffic sources.

Data: The first two weeks of a successful, review-style video.

Data: The first two weeks of an unsuccessful, entertainment-style video.

Right away, we see two clear differences: Only one has consistent gains on views each day, along with a very high CTR compared to the other.

The first video shows a consistent, mid-level trend pattern (not a quick to grow, quick to die trend), bringing in mostly new viewers (via suggested and search). This makes sense, as the channel was just starting to grow, and I didn’t have much of a fanbase yet. From this, we already have a pretty good understanding of how well the video is doing, and can explore the next step to see how it could be even better.

The second video shows signs of the “Insta-Death” pattern, where YouTube pushes out the video to viewers, but after seeing poor performance and other factors, stops recommending the video. For channel stage context, while these videos were only uploaded 2 months apart, I had 3 major video successes in between, but were not the same type of content as this video. Thus, we move to the next step.

4. Find and Review Other Engagement Metrics.

YouTube Analytics has a ton of information, including demographics, common watching times and days, and even device types. However, it is up to us to choose relevant data that can explain our video performance.

For the first video, we already know that people are interested in the topic and are interested in the video (15% in Browse Features means for every 100 people that get the video on their homepage, 15 people click on it instead of other videos, which is insanely high). The next thing to focus on then is content.

Viewership drops off quite a bit, with only around 20% of viewers finishing the video. Given the context of the video and audience, this is not too bad, but can be improved. In my next video, I might add more things that pop-up on the screen, change scenes once in a while, and maybe shorten some song demos to keep people more engaged. Another thing we notice is that subscribed viewers typically watch a bit longer, which is good! This means we have more loyal subscribers that actually enjoy our content. We can explore the last step to see if a change in content is justifiable or not.

Now going to the second video. We see that our CTR is really low, but why? Based on how well the channel was doing overall, this may have been caused by a lack of interest from return viewers. This is evident when we compare New vs Returning viewers.

We can take 2 things away from this: Despite the rapid growth of the channel overall, most viewers from previous videos aren’t watching this one, and even with YouTube’s algorithm pushing it out, it’s not attracting as many people as we’ve hoped.

Looking at viewer retention, we also notice that on average, 20% of the viewers watch all the way through. This seems alright at first, but with similar content, this retention is potentially below average. Why? Because the form of content was different. While most people clicking on such a video expect just a song, I included a creation process and the story behind it, reducing retention from the very start. Perhaps the thumbnail and title were misleading in a sense.

With that, I still expected the video to do better. We also still don’t know why the CTR is so low. Additionally, I’ve had videos where both retention and CTR are all super high, yet the video failed to take off. To understand these situations, we turn to the final step of the research process.

5. Compare to Similar Videos and Topics.

Think about it: the algorithm behind YouTube is meticulously engineered to keep people on the site for as long as possible. This not only means serving videos that appear to be promising, but also sticking to safer, relevant topics with an established viewership. If your new video was on a topic relevant in 2017 with no reference to nostalgia or history, it probably won’t get far, despite your friends watching it all the way through (yeah, this happened to me). Similarly, if you miss a trend but still try making content on it, most viewers will be fatigued from an overdose of the trend, hurting your chance of success.

Besides the changes suggested in the previous steps, our first video looks good overall! At this point in time with over 30,000 views and still growing, it is a highly successful video, all things considered. Thus, it’s time to revisit the second.

While still just theories, the two reasons why I think this video did not do well was due to trending content fatigue, combined with a seasonal topic that was not only suffering from content fatigue, but potentially trying to target the wrong audience, general music listeners who didn’t care about the creation process or how hard the song was.

With many topics, you can use Google Trends to understand the relevance of a topic. For ones that are harder to search, it’s always helpful to have videos on your own recommended in a similar niche. I distinctly remember a large amount of these videos appearing in my feed, but dwindling as time passed, and were almost non-existent when I published my video.

If after all of that, you still don’t understand a video’s poor performance, there might have just been videos that just did better than yours. YouTube can’t put too many videos in front of one person, so it needs to select the content strategically. At times, this might just be luck, but in others, someone else’s content may have just barely done better.

With all of these things in mind, we can finally come to a conclusion and go back to update thumbnails, titles, cut irrelevant sections of videos, and create new content accordingly. By meticulously analyzing the data in YouTube Analytics, we can really understand the success factors of our videos, helping guide adjustments and leading to long-term growth. Good luck out there!