Do Virality Scores and "Proven Ideas" Work?
How do YOU validate your content ideas?
The basic assumption in YouTube circles is that if you imitate things that work, your things will also work.
This is literally the intro to the excellent industry-standard newsletter and tool - Creator Hooks. “I just model what works”
The assumption is that breakout videos (or outlier videos) are more worth imitating because they work better.
This makes sense, but how do you know who the winners are?
And, out of those, who is the winner-est?
The View Problem
The absolute first thing people see about a video is the number of views.
Big view numbers are impressive, but they don’t actually tell you stuff you need to know to figure out what video to imitate. There are 4 specific problems with just looking at big views:
The Mr. Beast Effect (bad practices are imitated blindly)
The Recency Issue (old bangers aren’t bangers no more)
Supply / Demand Problem (is the market saturated?)
The Matthew Principle (Elon Musk can post a literal period and get 69M impressions)
Because just looking at views doesn’t give us enough information, we look at virality scores and heuristics.
So, how do the industry-leading tools solve this?
Creator Hooks
Outlier Views / Average of 10 video views (5 before and 5 after) x100
This makes a ton of sense, you want want to know not just that there are more views, but by how much. This is measured in standard deviations. If you’re looking at a highly variable channel (where 1 video gets 100 views, and another gets 10,000) you’re going to judge an outlier not if it gets 13,000 views, but if it gets 130,000.
Variability is important.
Creator Hooks is focused on titles and thumbnails and provides a database of outlier titles and thumbs with the Pro service.
1of10.com
This service seems to take a similar approach - looking at view averages for the channel (not sure how they calculate it - if it’s a limited series like Creator Hooks or the full channel history).
Note that this approach and Creator Hooks look at view totals - a useful measure for sure, but one I look at only in the context of another measure that I’ll explain below.
Here’s why: a video that went big at launch may be a bad video to imitate if the topic is no longer relevant. In other words, the the recency problem.
For example, this result from 1of10 is 3 years old. It sure performed well, but … is it worth imitating now?
Maybe the Maldives trend is now dead, or maybe this is a great example to imitate and get a huge video.
It’s hard to tell.
Ultimately, you have to rely on other indicators showing interest and intent (just like SEO). You could look at Google trends, for example, that shows search trends for YouTube:
There is a dedicated suite of tools that tells you more useful information:
VidIQ
This is not exacly a virality score, but it serves the same purpose. VidIQ looks at your search query, looks at the keywords and gives you a score out of 100 based on search volume:
If you have a high score, you’re more likely to get more viewer traffic based on your title. If you get a low score, VidIQ provides related keywords that might make it easier for your video to rank.
Of course, there’s a lot more to YouTube success than just search traffic.
Organic Virality
The way many big creators achieve their growth is not by hitting the pockets of demand in search.
Instead, they pull in huge numbers from pockets of viewers who had no intent to watch their videos at all - they just opened YouTube, saw a thumbnail, read the title, and clicked.
This is the Browse Features traffic source, and THIS traffic source is how most of the BIG YouTubers got there. They consistently made stuff that pulled new people in because their videos were relevant and interesting to a broad swath of people.
That’s why organic and search are at odds. Search tends towards niched-ness and specificity. “Browse Features” tends towards broad-ness and curiosity.
An important step in your analytics-driven strategy is to
It’s a complicated balance, but ultimately you have to learn how to do both at least to some degree.
VidIQ also provides an important piece of data to gauge whether the video is getting a lot of attention NOW - the Views Per Hour measure (VPH).
Seeing the search demand vs supply and whether a video still has viewer interest can be helpful in coming up with your next idea.
But, speaking of VPH…
My Solution - YT Breakout Virality Score
My attempt at building something useful for myself focuses on VPH a LOT.
I believe it solves 2 problems:
Recency
Fair (er) comparison between big and small creators vs views
Remember, we are trying to identify videos worth imitating here - breakout videos. This scoring system takes into account recency because the VPH average takes into account recency, and compares the video to its channel.
The better the video outperforms its own channel’s average, the more likely it is to be worth imitating. By looking at the video’s average VPH vs the Channel’s we can see whether THIS PARTICULAR Mr. Beast video is notable compared to the rest of his work.
Here’s My formula:
Single video’s average VPH / Average Views Per Hour of a Channel (lifetime) x 100
By looking at the average VPH of a channel, I establish what is “normal” for the channel.
Videos with a higher VPH pull up this average. Videos with a sustained higher-than-average VPH pull up the channel average significantly. These are the ones worth imitating - they’re the breakout videos.
Here’s what the numbers mean:
100 and above - A true breakout hit, the video is raising the overall average Views Per Hour benchmark for the channel. The blue line is way above the grey “typical performance” range.
By just looking at a bunch of examples, I figured out that a score of ~30 looks something like this:
29 - 3 - Within the typical performance range - 3 typically corresponds with the bottom of the range. The blue line is somewhere within the typical performance range - the lower the number, the closer the line is to the bottom of said range.
2 and below - Below typical performance range - this video bombed.
Some More Analytics Nerd-ery
An interesting find that came out of this is that a video that has VPH of around 30% of the average for the channel is performing well. Weird, right?
This is because VPH for basically any video looks like this:
There’s a big bump at the start, and then a decay. The decay can be fast, or it can be slow. There can be spikes afterwards (algo is finding new pockets of viewers) but that spike and decay pattern persists for basically all videos.
So, imitating a video that’s getting a nice VPH in the first 24-48 hours is actually not a good strategy - you’re not yet sure if this is a breakout hit or not. It might have high VPH (as happens with big creators) and then flop into 0 views per hour forever after.
When I look for the breakout hits I discount the first 48 hours after publication - and only look at the average VPH after one week.
Reducing the channel and the video to a single number creates a snapshot in time - the virality score of the video changes as views further decay. This takes care of the Recency problem - if you were to calculate the Breakout score for an old video, it would give you the relationship between the average VPH of the channel NOW and the video’s average lifetime VPH NOW.
1of10 doesn’t account for this relevance decay, and since views on the video are cumulative, it’s more prone to give you false positives. (Of course, my thing isn’t a live product so what you get in the newsletter is the snapshot of the first week of the video.)
Limitations of Breakout Scoring
Bottom line is - the Breakout score gives me information about the success of the video relative to the channel that I don’t feel I get from other tools.
No measure of virality will be useful for everyone, and because I don’t have access to YouTube’s raw data (I scrape public data via YouTube API) these are all just ways to guesstimate something about the video.
Another limitation of my process is that I run the sampling manually from a Google Sheet, so I am following only about 100 creators. I could theoretically run the analysis for any video, but I can’t run it on EVERY video. So, I settled for the assumption that the creators I include on the list are producing relevant to my and my client’s content, and worth watching for when they make something worth stealing imitating.
Ultimately, the very idea of trying to “pre-validate” a content idea is a bit iffy…
Problem With Virality Scores
When you’re asking yourself “Will this video get X views” before creating a video, you’re making a bet. Your bet is that:
Demand for this kind of video > the supply in the attention market
Your video is better than the existing supply in the attention market
First, you need to understand if there’s a demand. Which means you probably want to read everything this guy’s written:
Second, you have to understand what “good” is to the audience and give them something they’d rather spend their attention on than what already exists.
This is another bet. You believe you’ve solved the pattern of what the audience wants, and are willing to put it up against the competition.
There are no “sure things” in the attention market, and being successful occasionally is about as good as most creators can hope for. Luckily, that’s all you need - so long as your content actually builds an engaged and interested audience after they discover your channel.
How Successful Creators Start
I’ve seen a lot of new creators rush to ‘use all the best practices’ to try and skip the initial learning period.
I am now convinced it’s not really possible. You have to see enough patterns and dial in your predictive validity by making enough bets to get an animal sense for “what is good”.
I think for most creators, and especially newer ones, success lies not in ‘validating’ their ideas with numbers, but in learning to to make peace with the idea that you are making bets - assumptions.
And then, they do the following:
Figure out who the audience is and what they want
Test a series of hypotheses about the content, message, and format
Concentrate on the quality of the information, not production value (low-production-value videos with good content are FAR more likely to get traction than sleek but unhelpful or useless video)
Once they have a clearer picture of who the audience is, how they talk, and what they need - build a core following
THEN, and only then start expanding by finding similar, related audiences and broadening content to appeal to them.
If you’re new to YouTube - start with search-based narrow content to get initial traction and at the same time find ways to inject your personality, and develop your artistic vision and tone of voice. The first 10 videos will be crap. If you pay professionals, they’ll be crap but with nice graphics.
Use breakout videos for inspiration, but please, for the love of the YouTube Algorithm, stop expecting that “Alex Hormozi titles”, “Iman Gadzhi graphics” or “Ali Abdaal Shorts” will bring you success.
They won’t.
Saying something valuable to an audience who wants to hear it will.
If you like this post, share it with someone who would find it helpful.
1of10 doesn’t account for this relevance decay, and since views on the video are cumulative, it’s more prone to give you false positives. >>
so do you think 1of10 not too useful for beginner that start faceless youtube channel ? my biggest problem right now is to find niches that profitable enough, not too high competition.
thanks
This post is LOOONG