The YouTube Formula. Derral Eves

have had both good press and bad, but they did alleviate a lot of YouTube's headaches in those formative years of ad revenue sharing.

I recently sat down with Jim Louderback, CEO of VidCon and former CEO of MCN company Revision3, and talked with him about multichannel networks on my podcast Creative Disruption. He talked about how MCNs affected the early years of YouTube's ad revenue explosion. We discussed the ways they helped, but also the problems they created. “In the end,” said Jim, “a lot of MCNs did not provide the value they offered. They brought on too many creators and brands to manage, and there wasn't enough revenue to go around.”

      Once YouTube had a better handle on operations, they offered their own Partner support rather than losing creators to outside MCNs. In 2011, YouTube acquired Next New Networks, a company that had been managing a lot of YouTube's early creators. YouTube was ready to take back control internally and put that money back into their own pocket. For this reason and others—like fewer creators signing up, and a smaller margin of revenue per view—MCNs have seen a significant decline on YouTube in recent years.

An Evolving, Thriving System

      As you can see, ad revenue sharing completely changed the YouTube ecosystem. YouTube had been a minefield of instability in its fledgling years. They were learning the hard way that their ecosystem was a delicate balancing act—among its copyright holders, its viewers, its Partners, and its advertisers. There has been an “Adpocalypse,” countless issues with the algorithm, FTC COPPA children's privacy issues, Adpocalypse 2.0, and more. YouTube has learned to deal with the problems and tried to make changes to satisfy the masses, but it is a constant effort.

When we come full circle back to the viewer, the first component in the ecosystem, we have to consider how much the viewer has changed. YouTube really wants to have satisfied viewers. Over the years, they have tried to modify their recommendation feature to figure out exactly what each viewer might want to watch. They know that happy viewers will stay around longer, and viewers who stay around longer will produce happy content creators and happy advertisers. And the more the viewers watch, the more money everybody makes. The thousands of changes that have been made to the algorithm over the years have literally paid off, so the better the algorithm gets, the happier everyone will be. Get ready to dive deep into the algorithm in the next chapters.

      YouTube must be figuring out some things, though, because they have seen 31% year‐over‐year growth! In 2020, YouTube announced their revenue for the first time ever. In 2019, they made $15.15 billion, which was nearly double the year before! That is mind‐blowing, in both the amount of money and the percentage of growth. People watch more than five billion YouTube videos a day. Billion. To really grasp how much bigger a billion is than a million, consider this: one million seconds is roughly 11 days, while one billion seconds is 31½ years. Now give a second thought to that $15.15 billion figure for YouTube's 2019 revenue, and gasp. And they are really only just getting started.

      YouTube began as a dating website for a handful of college co‐eds in California in 2005. Now it reaches every corner of the globe on every device. Roughly one‐third of the entire population of the earth is watching YouTube regularly. Again, we are talking in billions. Where it used to be a specific demographic, the YouTube viewer is now everyone.

Creators and businesses can proactively position themselves to win on YouTube by learning about their own role in its ecosystem and the mechanics of a good channel. These are nonnegotiables if you want to succeed on the platform. Don't try to game the system; try to align yourself with YouTube's goals so fewer problems come up and you can focus on good content creation. YouTube has changed exponentially since its inception, and its ecosystem has changed, too. If you want to be a part of that ecosystem, you have to understand how it works and how you can adapt to it, because it will continue to change. You can adapt intelligently by looking at the data YouTube gives you. I'll show you how the algorithm works and how to create and adjust from its data.

3 The YouTube AI: A Deep Learning Machine

      A YouTube creator who recognizes the need to adjust to the data but doesn't have a clue how to do it is like a gardener who wants home‐grown produce but has never planted a seed. Becoming a successful gardener doesn't happen overnight, and neither does becoming a YouTube pro. You have to grab a shovel and dig in. There will be blisters on your metaphoric hands in the beginning, but as you develop your data‐digging muscles, you'll start to unbury a network of underground connections and discover a whole new world of the hows and whys of YouTube and what it takes to produce successful content.

      YouTube's artificial intelligence (AI) is an evolving structure in the digital ecosystem, and it takes work to understand and utilize, because it's malleable. You'll need to be malleable, too, meaning you have to adapt your strategies according to what's currently working. Your best chance at doing this successfully depends on your knowledge of the systems at play.

The AI Evolution

      As contemporary YouTube users, we have grown accustomed to the site dishing up what we like, unprompted, but it hasn't always been this way. Initially, YouTube primarily was a place to find answers to our questions, like how to change a tire, and a place to be entertained, like watching cats play keyboards or laughing at kid videos, like “Charlie bit my finger.” It was built on a simpler system that wasn't good at making recommendations. But YouTube today has a complex machine learning system that has gotten really good at guessing what people want. Let's take a closer look at how its AI has changed over time and why that matters to you.

      About 2011, YouTube started making system changes with one purpose in mind: get people to stay on the platform longer. A YouTube researcher working on this issue found some gaping holes in the framework. For example, a huge portion of YouTube viewers had gone mobile by then, and YouTube didn't have an accurate system for tracking user behavior on mobile devices. Palm to face. There was work to be done.

      Since July 2010, YouTube had been using a program called Leanback that queued up‐next videos that were ready to load after the video being watched was over. There was an initial increase in views, but soon they plateaued. They got the same results from a follow‐up AI program called Sibyl.

YouTube joined forces with Google Brain, Google's machine learning team, whose AI development and tools were leaps ahead of the field. Their goal was to build a system with the Google Brain foundation. Their main objective continued to be viewer longevity. On March 15, 2012, YouTube made the switch from a “View” algorithm that rewarded video view count to a “Watch time” algorithm that rewarded viewer duration. This AI followed the audience everywhere to ensure it found the right video to put in front of them. It had the capability to recommend adjacent videos rather than clone videos (“adjacent” meaning similar but different enough to keep interest). “Clone” videos inevitably pushed viewers off the platform because they were watching basically the same thing on repeat. More importantly, it would queue videos based on how long viewers had watched them instead of how many clicks and views they had gotten.

      YouTube's goal was for users to “watch more and click less,” meaning they didn't want viewers to have to click on a bunch of videos before finding what they wanted. The AI could match them better to content they liked so they could spend more time actually watching.

      This Watch time switch transformed YouTube's viewership—people did stay on the website longer. Misleading “bait‐and‐switch” tactics used by some creators were no longer being rewarded by the AI, because viewers left quickly when the content didn't deliver what the title and thumbnail promised. Viewers did stay to watch videos that delivered what they said they would, and the AI kept track of these videos with longer view duration and recommended them more. Additionally, viewers stayed to watch what the AI recommended next because they were relevant to what they had already shown interest in.

      In other words, viewers were taking this new AI bait: hook, line, and sinker.