Understanding the Cloud Media Workload: Content Distribution

The HPA Tech Retreat is the pre-eminent gathering of industry-leading thinkers and innovation-focused companies engaged in the creation, management, and dissemination of content. At the HPA Tech Retreat, the most compelling topics facing the media and entertainment landscape today are presented, explored and debated.

The impact of Cloud on content distribution is profound. From the role of AI to add insights and improve monetization, to the abstraction of classic bottlenecks into micro-services and APIs — innovations in technology are evolving the very nature of how media can and should be distributed at a rapid pace. Join us to gain state of the art insights into critical considerations and solutions for content distribution from executives across Metal Toad, Veritone, Accedo, and InPlayer, who are all helping the media and entertainment industry navigate this radical transformation.

Featuring:

• Logan Ketchum, Director of Sales & Strategic Partnerships at Veritone
• Bleuenn LeGoffic, Head of Business Development & Innovation at Accedo
• George Meek, CEO at InPlayer
• Tony Rost, CTO at Metal Toad (moderator)

Tony Rost  

Hi, everyone, and welcome to the Hollywood Professional Associations Tech Retreat panel on media distribution in the cloud. My name is Tony Rost. I'm the Chief Technology Officer at Metal Toad, an advanced cloud consultancy specializing in content and media solutions. Joining us are leaders who live and breathe the content distribution in the cloud every single day. Panelists, let's go around the table. Please introduce yourselves and your companies and what you do for them. Let's start with Bleuenn.

Bleuenn LeGoffic 

Hi, everyone, I'm very glad to be here. I'm Bleuenn LeGoffic, Head of Business Development and Innovation at Accedo. Accedo is a video solution provider, creating premium video experiences. We help content providers to grow their video business for the help of our production services.

Tony Rost  

Great thank you, Bleuenn. George.

George Meek  

Hi, good afternoon. My name is George Meek. I'm the CEO of InPlayer. We're the leading provider of monetization services for video online — whether it be live, on-demand, pay-per-view, or subscription — providing all those services for people to make money out of their content.

Tony Rost  

Thank you. And Logan.

Logan Ketchum  

I'm Logan Ketchum, Director of Sales and Strategic Partnerships for Veritone Inc, an AI company, focused and media entertainment, offering different AI services for each part in the content lifecycle.

Tony Rost   

Just when I think I understand where we are heading with content distribution, something comes along to surprise me. Last week, I watched my first movie in a VR headset at my house. Yesterday, I learned that the European Football Championship is partnering up with Tik Tok. So I think it's safe to say that we have a lot more surprises coming in new ways to distribute media, and more importantly, new ways to monetize both new and old media. One thing I want to start with is to talk to the panel here about where we were at ten years ago with monetization generating revenue and distributing to new types of screens, and where we're heading in 2021.

Bleuenn LeGoffic  

I mean, ten years ago — I think people need to remember that in 2000, the internet barely existed, and we are 21 years after that. Everyone needs to realize how fast this has been evolving. Something that is very different from ten years ago to now is the fragmentation of the video ecosystem, in terms of — not only, of course, technology and so on — but also in terms of everyone's fighting for a minute of attention of the user. I think the big shift for big media companies was that we're having the possibility to solve the complexity, the technical complexity, and actually reach the audience on the new media, which was the essence of OTT before. Now the cloud providers have been commoditizing all of this. This is clearly that it's not the question of "Can you get on the platform and monetize it?" it's "Will you sustainably stay and grow?" The big question is there, and I think one key area is that the power now is to the audience. It's not any longer about who's going to have the shoulder to actually bring the content to the audience. It's who's going to be relevant enough to continuously find your audience and stay relevant.

Tony Rost  

In 2020, I was buried in efforts for direct-to-consumer applications. One new challenge that direct to consumer introduces is the idea of dark content and finding new ways to recognize revenue across the portfolio of assets. Logan, a question for you is, what have you observed with studios going direct to consumer and being exposed to dark content?

Logan Ketchum  

I think the challenge becomes a lot of that content is generated for first air — whether it's live news footage or episodic content. So there's no doubt that when it runs first, that content is extremely engaging for the audience. I think the challenge and then the opportunity becomes further down the line. When you have another opportunity — may be on the same platform, maybe on a different platform to engage other viewers or audience members is how can you repurpose that content, or portions of that content, in order to do that effectively and not burden your organization with having to create new content for each platform organically? The challenge becomes, if the metadata associated to that file initially is not robust enough, and you just have the basic metadata of date, time, title, where it was filmed — whatever's on the camera card — it becomes difficult for media companies to repurpose that content moving forward. Content-based metadata, oftentimes -- and from what I'm learning from my customers and partners — is a prerequisite for the content supply workflows, whether those workflows be manual or whether they be programmatic. If tags don't exist -- and by tags, it could be themes, actors, mapping to IPTC or IB taxonomy — to figure out how to drive those content workflows, that content, you know, just sits in the archive. It's not used again. It's not re-monetized if you will. We saw content shortage during the pandemic. Some folks, like Netflix and others, got really creative around ways to capitalize on that. I think moving forward, it will become a requirement for the largest and the smallest content owners and broadcasters to have a strategy on how they can activate their archive, indifferent and new and compelling ways.

Tony Rost   

If our goal is revenue recognition, what are some of the use cases that a more robust machine learning ecosystem could offer in revenue recognition in the coming years?

Bleuenn LeGoffic 

So rather than thinking about personalization — because let's face it, we're very far from having an inventory of ads that will enable personalization — but having contextual ads, which are about positioning the ads at the right moment, making sure that there is not an ad fatigue in that and adding context to what kind of ads should you put at this moment of the content is definitely key.

Tony Rost 

It seems to me that the companies that want to leverage that not only need to shine an AI light on dark content into their histories but also need to unify their metadata with their audience analytics so that you can almost blend the content, content foundations with highly relevant advertising. How are we doing there? Are you seeing a lot of strong moves? Or are we still in the beginning stages of unifying audience analytics and content libraries?

Bleuenn LeGoffic   

We're not any longer in a place where we don't know how. I think that's on the way to be sorted out. I think the major challenge today is really about how do you — once you have identified that — how can you do that? How do you make sure that every component is being attached to this? So how do you make sure that you have a feedback loop between Europeans and all the different components that can be flexible to adjust? Whether it's content distribution in terms of technical items, video quality in terms of GCN, in terms of whatever it is, right? You can imagine. You have a lot of work that is being done around analyzing the data of the client side to make sure that you can optimize your video distribution from a technical standpoint. But, it can also grow into something more, where you can suddenly decide to offer potentially different ads. You could also look into, of course, recommending content, but also start looking at "Okay, what are packages of content that suddenly emerge in your offering?" When you start looking at the usual content packages that providers are putting behind the paywall, then you start realizing that, actually, by looking at the consumption that people have, you start having association of content that you might not have seen before. Suddenly, you can have add-ons that are being created in your offering that are much more relevant to what people are used to watch and might be connected to the next type of content that you could upsell for a package that you hadn't seen before. When we were in a squared box, your monetization strategy should be like this. You're going to run with your sports package, and it's going to run for three to four-five years. Now there's tons of experimentation, working with George in a lot of different OTT services; we can see that now it's very much needed for them to constantly experiment on their packaging and offering. That's only possible if you get the data if you can actually analyze your data and decide on adjusting your offering. I don't know if you agree, George?

George Meek 

I agree. It's funny. It's one of the things we've been working on. It's actually pretty complex. When you look at what we do, because we work with a huge number of different players, — as I'm saying, video players and devices. Then to be able to look at who their audiences are across those devices and those players. And then bring them all together for those individual clients. And then understand the media consumption versus payment versus timing and all of that. We don't have that, and it's something we've been working on looking at with partners again. But to do that, you'd need to then look, "Okay, I need to be integrated with the player 12345. This client has integrated with that. Then you tie all of that together." I have a very simple analogy, again with this is. You know, our clients who succeed with us are digital first. If you bring on an old school sort of broadcast model, they come in and say, "I want to get some content that needs to be actually rather than advertising — I want to charge my users for it." They come from that world. They struggle. They don't get the audiences they think they're going to get. But you know, we've worked with people who have found their whole career on YouTube and create events from the back of that. And they have millions of subscribers who come on board from that. Whereas then you get a big global broadcaster, he thinks he's got a bit of a niche. They might be able to charge for him. He'll probably get 10x less than this one-man band on YouTube. So digital first is important, but the analytics around that, we were working on it. Hopefully, we'll have something that we can provide more data back to the content owners this year.

Tony Rost  

We have an increase in direct to consumer. We have an increase of real-time analytics, passing through feature engineering and strong ML feedback loops. Eventually, you're going to move yourself back upstream and be influencing production. What are your observations in the 2020s on the feedback loop from distribution back into production phenomenon that wasn't necessarily easy to do 20 years ago, but now is more possible? I'd love to hear from the panelists on that feedback loop and then the cycle time from audience back to production and that kind of testing and experimental mindset that introduces in content creation.

George Meek   

I have a quick stab at this. We work with people who, for broadcast equipment, have a stadium with 20 cameras in there. They have a full broadcast -- trucks outside that's uplink via satellite and all of that. Then we work with a guy who has a camera crew of two and try a model of production remote, remote mixing. Then we also work with one of our clients called Synergy Sport that was recently acquired, and they have automated filming and basketball stadiums. This is a funny story. There was a game not long ago. It was a football match, again with automated production. Really with that, I think it's fascinating because any league can monetize their content with zero production costs. The business model around this is they're looking to get the betting rights for the sport. So, therefore, they're given the equipment. So then suddenly I've got free production. But there was a football game not a few months ago, where the automated video camera just followed the referee for the whole game because they mistook his bald head for the football. So there are production issues within that. We see that we have clients who will sell just as many tickets to subscribers with a one-man band production team to the full Sky Sports level offering. It's down to the fact that if someone is a fan of that piece of content, whether it be music, whether it be sport — whether it be entertainment, whether it be leisure, all of those — if they're into it, and they know where they can access it and can access it easily, they're going to pay for it. To be honest with you, some of the content is terribly filmed, but they're fans. So, they will watch it. They know that the more they invest in it, the more the production level is going to go up over time. That's also something we can look at.

Tony Rost   

It's crowdsourcing content creators now. Bleuenn, Logan, what are your observations on the kind of shortening of the feedback loops and the innovation opportunities that provides?

Bleuenn LeGoffic 

I've always been quite keen on seeing more of this. One great example: I think Netflix did that in 2019, where they started to do interactive storytelling. I think HBO also had some series they produced, where you have multiple possible scenarios, depending on who's the audience. How they choose to interact with the storytelling, they will choose that they are attached to it. I find this interesting because, from a business standpoint, it means that you're actually taking a direct insight from the audience, whether it's actually active or passive, into what you should distribute to them. If you look at the e-commerce industry — to be less talking about cats and nice storytelling — I think it's very important that from a business standpoint, we remind ourselves that the e-commerce industry has been exploding because they had been capable of adjusting their digital strategy to what the audience wanted to see to each of the person. If you look at the Amazon homepage, or if you look at the booking.com, they have thousands of experiments running at any given time to just constantly adjust the offering that is being proposed to the end user. Of course, we're not in this capacity to do that when we are in a video service. But imagine suddenly that if we are capable of placing contextual ads, maybe there is some next generation of the video experience to actually put some contextual sort of storytelling. Suddenly, you start having a production of content that the multi-player application of possibilities with AI is huge. 

Tony Rost  

Well, panelists, we're out of time. I would love to keep going with you, but thank you for making the time today to come and talk to us here for the Hollywood Professional Association Tech Retreat panel discussion on media distribution and cloud technology.