YouTube’s P-Scores, Video Throttling, and More: A Comprehensive Guide

By: Bowblax, Nicholars DeOrio, Optimus, Pescatore

October 29th, 2019

Table of Contents / Intro

  • Part 1: P-Scores (Page 3)
  • Definition of P-Scores
  • Five metrics of P-Scores
  • Ideal content for metrics
  • Explaining more about P-Scores
  • P-Scores compared between two markets
  • Hypothesis
  • Part 2: How to find P-Score/video rating (Page 7)
  • Steps to find P-Score/video rating
  • Part 3: Video ratings + their importance (Page 9)
  • Overview
  • Six categories of video rankings
  • Perceived video rankings importance
  • Part 4: Throttling your videos (Page 11)
  • Code evidence of throttling
  • Two types of throttling
  • What throttling does/is
  • Addressing @TeamYouTube
  • Part 5: More/conclusion (Page 14)
  • More information
  • Conclusion/final word

Intro:

Thank you for taking the time to read this document. We’ve put hours of work into it so please do let us know what you think of it on Twitter! They’re at the end of this document. Also, please note that nothing in this is completely set in stone as we’re currently trying to piece together what we know of this situation.

We do not know everything for certain, but we do provide at least some evidence for what we’ve put in this document. We’ve been forced to speculate on much of what we see here due to YouTube/Google not giving an adequate description or definition.

Our other document with 200 YouTube creators P-Scores: https://tinyurl.com/y54xbz96

Part 1: P-Scores

        P-Scores refer to an internal rating given by Google to channels to rate them based on five metrics according to a video that they released on April 10th, 2019. These metrics are all combined to, “surface among the most popular channels on YouTube.”

The five metrics are popularity, platform, passion, protection, production.

Popularity: Heavily driven by watch time. The more minutes that your videos are being watched, the better that this metric will influence your overall P-Score. Channels that put out longer videos will most likely have better ‘popularity rankings’ than channels that put out extremely short videos.

Passion: Leans more on engagement with an audience, so channels that have more devoted fanbases who engage content more with likes, comments, and in other methods likely will have better rankings in ‘passion.’

Protection: Based on how suitable content is for advertisers. Family friendly channels obviously will do great. Edgy content will not.

Platform: Beneficial to content that is frequently watched on larger screens, primarily TV screens. Seems to target growing audiences.

Production: Focuses on content that uses high-quality editing, camera footage, and more. The better the production value, most likely, the better the score.

Ideally, if you make the following type of content:

  • 10+ minute videos with 45%+ viewer retention per video
  • A highly driven fanbase that shares your content, likes your videos more than the average creator, and comments opinions a lot
  • Is family friendly, little to no swearing, no sexual content, no violence, etc.
  • Content that can be watched on all devices, platforms and in all markets with ease
  • Content that focuses on high production values and good quality

This would make your P-Score higher than the average user.

        Here, we’d like to explain more about P-Scores.

        P-Scores appear to be more than just these five benchmarks being compared against your metrics, they appear to be more like overall scores of an account in general. When you look at a P-Score, it doesn’t give you insight on any strong suits for these five things. It only gives you the one unified number, which is your overall grade. The issue, however, is that creators have no way of determining or identifying the individual values for each of these benchmarks.

        Creators who are doing great with popularity, platform and passion might never have the opportunity to fix their protection scores because they’d never know it’s an issue, for example. You can mix and match between these five to make problems for channels, but they could never fully reach their potential if they can’t identify an issue. Google knows that these are issues but does not alert creators or give them an opportunity to access this data without finding some sort of weird exploit to get to it.

        Another major note for P-Scores is that they appear to be region-based. While obtaining scores for creators from the United States, we noticed that we were getting different scores for these same exact creators from Bowblax, a Canadian creator. He estimated that the average Canadian P-Score value for creators was ‘about 20 points less.’ To test our theory, we used a VPN from the United States to appear as a Canadian user, and we matched the numbers perfectly. P-Scores do appear to be region-based.

P-Scores compared between two markets

Optimus’ P-Score in Canada (771.56)

Optimus’ P-Score in the United States (801.7)

Overall difference: 30.14 points in favor of the United States


Pyrocynicals’ P-Score in the United Kingdom (1004.6)

Pyrocynicals’ P-Score in Canada (839.1)

Overall difference: 165.5 points in favor of the United Kingdom


WatchMojos’ P-Score in Germany (755.8)

WatchMojos’ P-Score in the United States (923.1)

Overall difference: 167.3 points in favor of the United States


CBRs’ P-Score in Germany (856.4)

CBRs’ P-Score in the United States (872.2)

Overall difference: 15.8 points in favor of the United States


        What we’ve been able to hypothesize with this information is that:

  1. Channels seem to do better in their home countries.
  2. Channels that don’t do better in their home countries generally have an extremely large fanbase outside of this location (example: NHL is larger in Canada than in the United States in terms of P-Score)
  3. Our ‘20 point’ guess was off, at least with our limited data that we tested.

(NOTE: We gathered data on 200 channels on our other document.)

Part 2: How to find P-Score/video rating

        We feel that your P-Score is an important metric, therefore, we want you to be able to access it. This is using the current method that we’ve all used to obtain the information that we’ve gathered and what has worked for everyone else. (NOTE: By the time that you’ve read this, YouTube may have already fixed this to where that you cannot obtain your P-Score through this method.) (Further note: YouTube has officially closed entry.)

Step 1: Go to a video that you want to use to obtain a P-Score. If you are doing this for yourself, go to one of your videos. If this is for another creator, select one of their videos. For this tutorial, we’ve selected a random YouTube video from Trending.

BuzzFeedVideo: “$3 Chicken Vs. $62 Chicken - Taiwan”

Step 2: Inspect the element of the page/view the page source. We’ve noticed a much higher success rate with viewing the page source over inspecting the element, so please take this into consideration, however, if for some reason inspect element is all you have, it should suffice. Keyboard shortcuts: Ctrl+U for view page source and Ctrl+Shift+I for inspect element.

WatchMojo: “Top 10 Worst Pay-to-Win Games”

Step 3: Press Ctrl+F to pull up the advanced search feature. This will help you comb through the code of the website and isolate what you’re looking for. Then, paste the following: ContentLabelRating

Step 4: Look for "channelPscore\": to figure out your channel P-Score. Alternatively, look for “":\"DV_[RATING] just before this to figure out your video rating.


Part 3: Video ratings + their importance

Video ratings seem to be extremely important not only to an individual videos’ success, but to overall channel success. If your video is marked as a higher-rated video, it appears to have many negative implications. With this, we’ve just confirmed something that YouTube has been extremely secretive about for a very long time: the rating system for videos on their platform.

There are six known categories of age ratings for videos. For each of these, we’ve attempted to try and link which types of content would most likely fit into them.

  • Rated Y: Super safe content. Good for literally anyone, it seems.
  • Rated G: Some of the safest content on YouTube. Extremely family-friendly, no issues with guidelines, likely will get high CPMs and make your P-Score more relevant due to a boost in the ‘Protection’ category.
  • Rated PG: Safe content on YouTube. Family friendly for the most part, the occasional guidelines issue might arise, will get great CPMs, and will boost P-Scores due to a boost in the ‘Protection’ category.
  • Rated Teen: Teetering the edge of safe and not-safe. Sometimes family friendly, most of the time is deemed ‘not family friendly.’ Will usually get a decent CPM unless guidelines issues limits ads on the video.
  • Rated Mature: Generally deemed not safe. Almost never family friendly, if so, probably by error. Will get lower CPMs if monetized due to limited ad status.
  • Rated X: Never safe. Never monetized, age restricted by default, and deemed problematic altogether.

When you look at how they’ve structured this, you’ll realize that YouTube has been internally judging creators’ videos on this scale for a while now without telling them, once again not giving creators a chance to play fairly. Not too much of a surprise, however, you’ll notice that a lot of  the times, these kinds of definitions aren’t clear enough.

A lot of videos that we’ve looked at are rated ‘teen’ on YouTube. Of course, we don’t know the 100% definition behind these terms so the ones that we’ve come up with are just our own, but if they’re accurate enough, that would mean that a lot of videos on YouTube are teetering on the edge of monetization and being deemed brand safe. When you look through a lot of these videos, as well, you’ll notice that they’ll have code in them that deem them “not family safe”, even if they’re teen, which means that they’re most likely not getting recommended as much. If this is the case, this would mean that if you’re being ranked anything above a PG, you’re risking spots in the algorithm and risking some ad spots to a certain extent. And this is right in YouTube’s code for these videos. YouTube isn’t going to recommend as much content if it isn’t family friendly because it isn’t safe for them to recommend, and it doesn’t make as much money, so this all really does make sense at the end of the day.

So it seems if you’re in the Y or G section of the ratings, you’re good to go.


Part 4: Throttling your videos

Throughout our research, we kept noticing several references throughout nearly every video rated ‘teen’ or above that would be labeled as a ‘throttle.’ To us, these are deeply concerning, highly problematic pieces of code that could potentially be ruining peoples’ channels. Take a look for yourself:

 

Nicholas DeOrio: “The End of The Canceled YouTuber | Zaptie [2]

SomeOrdinaryGamers: “Playing The Controversial Modern Warfare Mission

These are just a few examples of many that we found. Many of them were our own videos, many of them were other peoples’ videos. From this, there appears to be two different main types of ‘throttling’ occurring through YouTube:

  1. Brand safety throttling (advertisements)
  2. Feature video throttling (algorithmic?)

The brand safety throttling appears to be advertisements being restricted from the video in some way, at least by the wording. It will make constant reference to ‘brand safety ads being throttled’ and ‘reason: brand safety, disable google sold reservation ads.’ The odd part is that many of these videos are monetized videos, which suggests that while these videos aren’t fully demonetized, their full potential of ads aren’t being reached. If the code here is to be taken literally, the webpage seems to quite literally be throttling advertisements on videos rated above PG mostly due to brand safety issues. It will disable google sold reservation ads in some spots. It will not fully demonetize, but it will strictly only advertise enough to make the video enough money to seem like it’s doing well. Here is a piece of code that we found peculiar from the SomeOrdinaryGamers’ example above:

 

Also, take a look here:

{\"label\":\"FEATURE_VIDEO_IS_THROTTLED_FOR_OVERLAY\"},{\"label\":\"FEATURE_SESSION_IS_SENSITIVE_VIDEO\"}]

This suggests that while the video is having certain aspects of monetization throttled, the video was also monitored to see what type of content was shown. In this specific context, SomeOrdinaryGamers was playing a ‘controversial’ Call of Duty: Modern Warfare mission which included fictionalized violence in gameplay. This seems to have been noticed, which lead to the ads being removed in some spots, but on top of that, it would also appear that the throttling also carries beyond the advertising.

Feature video throttling seems to suggest that the video is being suppressed through the algorithm in some way due to the gameplay.

        Feature video throttling shows up quite often throughout videos rated teen and above, as well. These are videos that, from our understanding, aren’t considered family friendly very often. They aren’t going to be recommended as often, so would YouTube go as far as to intentionally throttle them? Their code suggests directly through its’ own language that this is the case.

Now, here’s a little something we dug up and wanted to address:

        Technically, they didn’t lie here. The recommendations, search and discovery systems may not be looking at monetization status for your video. It might instead be looking at every single element of the video, finding things that it deems wrong, rating it teen and then potentially throttling it in the algorithm due to the things in the video + throttling the potential advertisements without telling you.

Part 5: More/conclusion

At this point, we’re completely in the dark like usual with how YouTube has developed all of this and implemented this, and also on how they’ll react. We know that this information alone is extremely valuable to the YouTube community, and so figuring this out has been a potential revelation that can benefit creators for a long time to come.

YouTube will potentially fix these exploits to get your P-Score/other peoples’ P-Score very soon, possibly within hours or days. It’s best that you get your information now if you can in case there is no other way to get it. At this point, we’re unsure what this means specifically for the future. We just want this to help everyone else.

If you have read through this and you’ve found this helpful, consider sharing this information with other creators that you know on the platform.

Thank you for your time, and enjoy.

  • Optimus, Pescatore, Bowblax and Nicholas DeOrio

You can find us on Twitter here:

@SubToOptimus

@Bowblax

@JoshPescatore

@Nicholas_DeOrio