State of the Industry - How will streaming evolve in 2021

2020 was certainly an interesting year: COVID forced people across the world to stay indoors, with bars and theaters closing and cancellations of festivals and events resulting in entertainment to be sought elsewhere. Before the summer, it caused spikes in network consumption due to a rise in home working, online gaming and media consumption. These changes have driven large movement in the streaming landscape, with online events becoming the de facto standard, and OTT companies attracting huge numbers of subscribers. On the technical side, there was the official launch of low latency HLS by Apple, an array of new devices entering the market, some interesting moves on the Google Cast front, and a significant increase in adoption of the Open Measurements specification (OMID) on the advertisement front. But what will 2021 bring? Let’s have a look!

A comprehensive Guide to Low Latency (4)


It doesn’t matter if you’re a subscriber service, a free service monetisation through ads, or any other form of entertainment: it’s all about volume. The more viewers you have, the more revenue. It’s plain and simple. As a result, while getting new viewers to your service is important, it’s even more important to keep the accumulated viewers active.

A shrinking pie

In predictions of the past years, the streaming
wars often came back as a way to indicate the
battle for subscribers between OTT services.
Eyeballs increased significantly over the last
year, which of course resulted in a much
larger pie to divide. It does however speak to
reason that with (hopefully) less time being
spent during the second half of 2021, one
will start to question if the subscription of
“Streaming Service X”, which increases the
budget by Y USD/month is still worth it. Shows
will have been binged during the past year
and likely there are other streaming services
promoting their shows, ease of use or
innovative features.

The impact of this is likely not to be underestimated. With the pie shrinking, streaming services will have to battle churn by any means possible. While budgets spent on entertainment shifted from going to bars and shows into the digital world, it is highly
likely they will shift back to the analogue one,
and with it a choice to be made which service to keep. Beers, or Netflix. It will not be a tough choice for some, especially after a year of abstinence. What are they saying again about the Roaring Twenties coming back?


Exclusive content has been one of the key methods for churn prevention over the past years (“I can’t cancel X, the next season of Z is coming out next month!”). With the increase of consumption, there has been a lot more content discovery going on over the past year, with new shows being picked up by large audiences. Banking on the hope people will stick with a service for a handful of shows is likely not a good idea.

We’ve already seen experiments where shows aren't published an entire season at a time. For example with Disney+’s new season of The Mandelorian, forcing viewers to either keep their subscription while the show runs, or miss out on the (virtual) watercooler talk on Baby Yoda. While it’s a great way to prevent people from wandering from one service to the other, the question is if it will be enough.



Disney did not publish the entire season of Star Wars The Mandalorian in the hope to keep their customers in their Disney+ service.


Good content is no longer scarce, and every service has at least some. Large amounts of money have been invested by large services in order to attract new viewers. With dozens of top series and movies becoming available over the past year, it’s no longer the question if you will find something to watch on a specific service. There is definitely something good to watch, likely as good as with the service you’re using today. While for some specific shows it might be a good argument to stick with a service (or subscribe for a few months), I do expect streaming nomads to pop up more often, going from one service to the next depending on that time of year’s “hot” show.


A big trend in order to attract (and retain) viewers is the experience built around the content itself. As mentioned: there is a myriad of great content. In order to differentiate, it’s crucial to go above and beyond. While there are multiple ways, let’s zoom into a few of them.

WAT: Watching Apart Together

Services pioneered watch party capabilities in an attempt to allow viewers to grasp the atmosphere as if they were in the stadium, feeling engaged with the event, and connecting to friends and other fans. The evolutions and innovations which were the only replacement for “the real thing” over the last year might not make it from people capable of actually attending the events. But even then, they do allow you to connect with people globally: watching a movie while talking with your brother living in a different state, or experiencing that world cup game with your friends as if you all were in the stadium while the BBQ is crackling in the background.

One of the things which will be key to deliver these enhanced experiences, is synchronized low latency viewing. It brings the real live experience, ensuring your Twitter app doesn’t light up before you actually see the action happening. The advantage of being in the stadium, or watching through traditional cable, becomes significantly lower. By synchronizing these streams, watching content together has never been easier, even allowing you to connect through any standard messaging or calling application and still being able to watch the same content at the same time.

One technical evolution which will play a key role, is low latency HLS. While the protocol was announced in 2019, it was published only in 2020. At this point however, player support is on the rise: Apple implemented it in iOS 14 mid September, THEOplayer added support in the last Q3 release, and (even though developed at “the speed of open source”), players such as hls.js are adding it as well, with a release to be expected in the next months. On a server side, most vendors seem to be aiming for a 21Q1 release, which would allow for mass adoption. And mass adoption should go fast in 2021 for all live streaming needs.


Viewer experience often makes people think about VR. While I definitely expect (and already see) a rise in requests around 360 and VR streaming, the platform which will really win it in 2021 is… the TV. Seriously. Over the past years, we saw a big shift in streaming from a desktop/laptop only experience to the mobile one. That trend however is continuing towards the big screen. Smart TVs are more common. With COVID, people have spent a lot of time at home, so why not watch on that big screen which is really made for this: watching content!

Already at the start of 2020, Conviva published in its reports a growth of 239% for smart TV apps. A significant number when compared to the 63% average. Based on the press releases of streaming services, such as Apple TV+ announcing availability on Samsung and LG smart TVs it is clear that smart TV is a new focus area for OTT platforms.

Consumption on smart TVs is huge, and the ease of use of an application on the device, rather than messing with cables and external devices, is quite significant. Given these trends, it speaks to reason this evolution will continue and accelerate well into 2021, and smart TVs (especially the big vendors such as Samsung and LG) will become a must-have platform in the support matrix of any OTT service.


An important aspect of building an experience is by making it interactive.
In this, and many other enhancements, metadata is key. In its most basic
form, metadata is used to draw overlays on top of sports matches to showcase
the score, add a news ticker feed to the bottom of the news program, add in a
channel logo, … Often, these things are simply burned into the feed. You can’t
burn in everything though: it will become increasingly hard when you have
a multitude of players on the screen, and might want to highlight the viewer’s
favorite player, show stats for specific players, or simply what the odds are in
case you would want to make a bet right now. Services started to experiment
with this over the last year, for example Verizon combined the power of 5G with multi camera viewing and interactive overlays. This isn’t fiction anymore, but it is starting to enter the realm of possibilities.

While the game show app HQ Trivia had to throw in the towel last year, it clearly showed the power and potential of interactive video. With sport betting being legalised in the US (and expected to be legalised for most states by end of 2021), there are big market opportunities emerging in not just showing metadata, but also allowing viewers to action on it. In parallel, product placement and merchandising are growing ever more popular, so why not provide viewers to buy the same T-shirt their favourite character is wearing, directly from within their streaming app?

All these opportunities have triggered people to start looking more for the solution to metadata’s biggest problem: gathering and inserting this data into a media feed. While we still seem to be pretty far from this being standardised, the massive generation of metadata and ongoing projects will surely bring more cool and interesting use cases in the months and years to come.


Surely video quality is important, I hear you ask. Certainly. The arms race in regards to streaming quality is an ever continuing feat. While most large OTT services at this point provide 4k content with HDR and often multichannel audio with Dolby Atmos, small and medium sized services often are stuck on 1080p or even 720p content. With the move to the big screen, it is to be expected these resolutions will increase further. For the larger OTT vendors, we expect the horizon will shift towards 8k content, but with a limited footprint of devices capable of displaying it in full glory, adoption will likely not be massive.

A metric which however retains its place in the spotlight as the most important quality metric to achieve, is startup time. It’s about instant gratification. People are used to getting access to the content when they want. And they want it now. It comes as  no surprise viewers abandon content if it doesn’t load fast enough. With high volumes of quality content and alternative entertainment available, a focus on startup time is an investment which will pay off.

There are of course other (obvious) quality metrics which remain important, such as error rates and rebuffing rates. All arguments which make sense to reduce startup time definitely make sense for these metrics as well. It’s all about keeping the viewer’s attention: when playback stops, risk of churn tends to increase, and nobody likes to watch spinning wheels while waiting for content to load. We expect more and more focus on these crucial metrics as streaming services start feeling ever increasing competition.


Bringing in new features, supporting more platforms, building experiences, improving quality, … It all costs money. And loads of it. While you can keep investing on new capabilities either trying to outrun the competition or simply to catch up, budgets are not infinite. So what do we expect services to do to keep costs in check?


Most streaming services today are generating multiple sets of streams. For services using DRM, it often boils down to a set of HLS qualities with Fairplay to reach the Apple ecosystem on iPhone, iPad, Safari and Apple TV and a set using MPEG-DASH with Widevine and PlayReady to reach the Android, Chrome, Edge, … ecosystems.

With more platforms being targeted, this set is often (forced to) increase. SSAI doesn’t always function properly on all platforms, neither do all encoding profiles, resolutions, … or even features such as multiple audio. Let’s not even get started on more advanced capabilities such as low latency. To get compatibility with old platforms, in some cases even Smooth Streaming sets start popping up. The amount of different sets can grow rapidly. It’s often easier to set up a new set than to work around it. Things have to move fast after all.

Every set is often reencoded, packaged, encrypted and stored separately, with different configurations all around. While not obvious for every step, they all cost money, even down to the different configurations for your DRM server. On top of this cost, the user is often paying for this in QoE, with higher startup times as CDNs have more misses due to the need to cache different sets rather than a single set.

There is however some white smoke on the horizon. With CBCS (the encryption mode which has been adopted by Widevine and Playready, allowing to encrypt once) becoming more popular and interoperability between MPEG-DASH and HLS stabilising with CMAF, reusing your encodes across multiple (if not all) platforms is no longer a pipe dream. Leveraging these improvements, the simple generation of an additional manifest or playlist can reduce encoding, encryption and storage cost quite significantly. In some cases, even the additional manifest is not even needed. Don’t underestimate the impact here. It means you only need to do ad insertion once. Hell, it even works for low latency streaming…

While one could argue that this cannot be done for older devices, most legacy devices do allow for this. As long as you are using a Bring-Your-OwnPlayer approach, the options are there. Where the ABR set might be slightly bigger to cover the different form factors, it still allows for a significant reduction of compute needed, but also a huge simplification of your pipeline. While every engineer loves to build a complex system to steer all of this, it does come at a cost in development, but also for maintenance. If you’re not paying your own employees to do it, you’re probably paying someone else to do it in the end.

With costs becoming big, and the technology being matured, we expect more and more streaming services will start migrating towards a stack where only HLS or MPEG-DASH is being used, with the go-between of sharing the actual segment data. The shift might go bumpy as I do expect people will take the easy way out when targeting new platforms, the trend already seems to be starting and will most definitely continue in 2021 and beyond.


CDNs have become significantly cheaper over the past years. As volume for services grew, they were able to get better and better rates. Thanks to the advantages of scale, CDNs could in turn reduce costs further, allowing for lower prices.  According to Dan Rayburn, there is a reduction in price of about 30% every year. The question is however how long this can continue. With CDNs operating at extremely small margins, I fear we might be reaching the end of how far they can stretch. Despite this, CDN often remains one of the biggest operating costs for any streaming service. As audiences grow, this only increases.

There has been a lot said and done about the improvements of media codecs, and how they reduce your delivery cost. This of course is true. It’s actually not just delivery cost, it’s also storage. There is another side of that medal though: encoding costs often grow. On top of that, different platforms support different codecs, or might not even support one of the fancy “new” codecs. This will again add complexity to identify and provide sufficient flavours so every platform can benefit.

There is a big promise with AV1 becoming more available. While market penetration is still very low, and it will likely take still a number of years before device support is steady (making it more my 2025 prediction), more and more platforms support HEVC these days. This includes even your average smart TV, Roku or Fire TV, … Yes, even the “old” ones. Market penetration for HEVC looks good. Except of course where it doesn’t: the browser world remains the biggest issue with Chrome nor Firefox having support (but then again browser usage for streaming is far from what it used to be).

As a result, even for a codec which can be used nearly everywhere, you will always need a fallback. With existing AVC (h.264) streams available, these can however easily be used as a fallback (or one could leverage AV1 or VP9 as a fallback for the browser world). The reduction in cost in delivery (and storage if you can replace a number of sets with a better codec completely) should allow you to offset the additional cost of encoding. It’s an option which is clearly making it to the table with more and more providers, especially as audiences grow. Multi-codec support is definitely expected to be a must-have in 2021.


While I might be biased given I work for a player company, I must say I see little to no value for most streaming services to build their own player. Protocols keep changing. Features need to be added. Requirements expand and use cases keep popping up. The time spent on building, maintaining and evolving all of this is high. It makes much more sense to “share” this cost with multiple streaming services as customers of an already existing player.

Leveraging a solid base allows you to focus on what matters most. Those features which really make a difference for your audience. Working on the app experience. Adding interactive layers. Avoiding churn, and attracting new viewers. While they exist, it’s very rare to see a streaming service having a competitive difference out of the way how they support HLS. While it pains me to say as a player developer: players have become a commodity. Developing all of this yourself is quite crazy.

That does of course mean there aren’t any drawbacks when it comes to picking a player. Making the wrong decision can have a big negative impact. A player which actively follows (or even better leads) the trends and evolutions is of course way better than one lagging behind or is developed slowly. The same can be said when looking at QoE and rock solid user experiences: all of the hygiene factors have to be there. If they outshine the others that’s great, and this might be a must have for your use case. But you won’t develop that on your own without some serious investments.

Beyond the shier cost and effort needed to build and maintain this yourself, there is another aspect which is often overlooked: streaming has become complex. There are a plethora of different options throughout a streaming stack, and things have to work together. All of these usually come together in the client side, and often in the player itself. As a result, having a good partner on the player end of things is very important. (While one could argue that this is a reason to do this yourself, the question is at what price). We see this often when issues pop up with customers: it’s often unclear where the cause really is. From a player perspective, you can often root cause more easily, or even work around specific issues.

Long story short, we do expect the focus on the features which matter most to increase ROI. Optimising the player strategy can be a big gain, and we already see a lot of companies looking at options. Reducing the amount of work spent on base features or integrations with components, and working on what really matters for the viewer base.


So what do we expect? An exciting year! Loads of new (crazy) innovative use cases, more interactivity and the strengthening of user experience, and all of it while reducing costs. At THEO we still aim to enable streaming services to easily offer exceptional video experiences across any device. Are you looking to provide some new user experiences, or reduce your costs? Talk with us, our experts are happy to help you.

Any questions left? Contact our THEO experts.

Subscribe by email