Live Video : The Less Discussed Streaming Wars
The Rookie VC #8
Hi ! Today I wanted to talk about how video streaming is breaking out of broadcasting, what this means for the infrastructure supporting it and future applications that we could see emerge.
I felt like the topic of creating, producing and broadcasting content to the world and the war waged by incumbents took up a lot of the media and commentator attention, while infrastructure enablers and future applications were a lesser discussed subject.
Let’s dive right in.
The COVID-19 crisis has shone light on how critical live video and the infrastructure supporting it was to keep track of our world, work, buy and sell things. Broadcasting was first to benefit, but we tend to forget that it’s only a subset of the video industry as a whole.
Before we start diving in the macro-trends, I just wanted to make a quick distinction between the two families of use-cases (although you’ll see the latter is getting more and more bi or multi-directional):
Collaborative live video streaming, with Zoom as a poster child and orientated towards the use cases that involve bi-directional interaction via video, as opposed to ..
Unidirectional live video streaming, with Periscope, TikTok as poster childs, orientated towards the new livecast formats, which i’ll cover later on
Macro-Trend #1 - The number of use cases outside of broadcasting is already skyrocketing
While Zoom and Microsoft Teams have become the poster child of live video for collaboration and meetings across the board, the space has been largely “unbundled” and verticalized as shown by this brilliant mapping from the Tech Nexus team :
As JJ and the team explain in their primer, the rise of this verticalization is driven by 3 main reasons :
Casual social connections. It’s now okay to connect in any social context via video, and the interactions are getting better and better, as founders are building with a verticalized product focus
An enabler for virtual services. As I previously wrote, live video favors pure platform models in which creators are empowered by a tool to scale a service business. Passion drives engagement in communities, and video is one of the best vectors for the emotion and communication that cements the Passion Economy
Remote collaboration. It’s very commonplace to say, and anybody lucky enough to have a remote-compatible job during SIP knows how practical things are with live video. Under this driver, purpose-built software has sprawled across pretty much any specific use case for collaboration at work
Now a particularly interesting thesis would be the emergence of a no-code layer to democratize customization and unleash the creativity stemming from the mind-boggling variety of use cases there will be for live video.
We can see how compelling the SaaS case is to unbundle “usage-neutral” suites and incumbents, but I think there’s many other emerging trends in live video.
Macro-Trend #2 - The proliferation of video-specific APIs
Video streaming components for apps and site used to be a pain to embed and implement, but many API-based tools have captured the value there is in offering a frictionless experience to virtually any application builder. These are base layer enablers that remove the friction there is in running video chat anywhere.
Yet another great mapping by the Tech Nexus team shows how large the customer base reaches :
There’s also other types of feature-related APIs built for “nice-to-have” use cases such as video AR (DeepAR, AR gear, etc.), audio improvement for video calls (Krisp), notes and captioning (Assembly.ai, Deepgram), insights (Kairos).
After quickly brushing on the primary live video space software opportunity as an unbundling existing live video software, I want to touch on my favorite and most salient use-cases as an innovation driver for other spaces outside of work and productivity.
Live streaming has been normalized thanks to eSports and video games: Twitch streamers have made it normal for the average person to spend hours looking at another person’s live feed. Twitch has emerged as a clear winner for this specific use case, and users have then appropriated the platform for various other uses, some completely unrelated to gaming. In my opinion, this shows that (i) gaming is the birthplace of many innovations that can move entire industries (ii) Twitch is ripe for unbundling and verticalization in livecasting, just like the massive growth that Zoom experienced also accelerated the unbundling of multi-directional video & work-related use-cases
Live commerce
Until recently, cable TV live teleshopping had gradually made selling via video the lamest thing on earth. Though incredibly popular in the golden age of TV, this idea failed to adapt to a fast-pitched world and was trumped by asynchronous video : influencers and YouTube unboxing had effectively replaced the good old “Téléachat”. They were proposing a more authentic experience, delivered by people consumers thought they could trust (at least more than an old TV anchor).
In our beloved Western world, live streaming on popular platforms (Instagram Live, Facebook Live) is still an ancillary, “nice-to-have” bonus content for many creators making a living out of ad and affiliate revenue from “asynchronous” video. Instead, innovation in the space picked up from the East. In China, Taobao Live was first to launch in 2016, followed by Douyin (Tiktok) and Kuaishou in 2018.
To really get a grasp of how big a trend this is in China, consider this : livestreaming-related e-commerce sales in China have eaten up 9% of the total, adding up to a whopping $61bn in value as of 2019, and with this specific market seeing a +71% uptick versus the prior year.
In some cases, the business model is quite different from regular influencer marketing as we know it: instead of broadcasting from their personal account where they locate the audience, creators often act like hosts instead, with a lot of the following flocking to corporate accounts. For instance, L’Oreal’s Tiktok account counts 120k+ followers as we speak.
Let’s stop for a minute and examine why exactly this new channel has been a success so far :
Product discovery and the paradox of choice. Let’s take a simple example to intuitively explain the paradox of choice: you are browsing Asos on your phone and after 15 minutes, your vision starts to blur and no clear choice for your next summer outfit has emerged. Sales-oriented livestreaming makes it easy for retailers to narrow down marketing and promotion to a few select choices. Mitigating this paradox is key for retailers, who can capitalize on the virality of livestreaming to reduce marketing costs, especially when they are advertising an aggressive sale
Consistent engagement and trust. Asynchronous content creators on YouTube already leverage the trust their viewers and community place in them to push product recommendations and make a (sometimes hefty) living out of it. But livestreaming goes one rung up the ladder by enabling two-way conversations, with creators cherry-picking questions whose responses are most compelling to impulse-buy products
Good old cash discounts and urgency. Particularly when the sale is platform or brand-run, it comes with exclusivity - which in turns drives repeated viewership and purchasing. Some viewers even wait for their favorite product to be “on sale live” before buying it, which in my opinion tells a lot about the catalyst role livestreaming plays in impulse-buying
It’s still pretty unclear as to how this trend will come to Europe, and how dependent it will be on Chinese-led platforms, which can of course be another source of controversy. I am skeptical we will see a dominant European player emerge anytime soon on the old continent or in the US. Here’s why :
Tiktok penetration. Bytedance’s app has achieved impressive virality and penetration so far. In the US, the app gathers 65m DAUs in the population over 18 years old. Roughly one in five American, all ages. In Europe, it’s estimated that the app was downloaded at least 80m times, but we’re not clear yet about daily usage. In this context, any competing product would have to offer a significantly better shopping experience, and grow at a crazy pace. Another way to think about this, though, could be to offer private group sales on higher-end luxury products, hence tapping a niche market.
Awkward attempts by western incumbents. In early 2018, Amazon launched Amazon Live, but made in my opinion several key mistakes along the way. First, they launched the service without having a community of influencers and brands ready to go live with them. The Seattle giant forgot that the engagement and bonding was between a known figure and the viewer, not some unknown representative. Second, the videos are packed with information at the expense of bidirectional interaction with the viewer. The third issue was consistency : Taobao’s top grossing streamers stream more than 300 days a year, with an average session time of 8 hours. Amazon’s attempt was very sporadic, but as US social media incumbents are copying super-app roadmaps (WhatsApp evolving to become a Wechat copycat for example), we could expect more thoughtful attempts from the West in the future. Why not partner for a white-label, pure platform company like Shopify to make it happen ? This would likely fit in better in the culture and the product’s intent
Surveillance
In old democracies like France, a large chunk of the population tends to perceive increased surveillance as an authoritarian foray into everyday life. More recently, the death of George Floyd and the globalization of the BLM movement gave way to many voices standing up against racial bias in video surveillance AI. But beyond the widespread distrust, innovation is carrying on at a fast pace.
What’s happening in that space that really moves the needle ?
Edge-based AI. Today most surveillance cameras still send over data to a distant server to be analyzed by computer vision algorithms or by humans. In the future, the data efficiency of these devices could be dramatically increased using edge computing. Edge networks could efficiently analyse and store locally large sums of data on-device, then send pre-analyzed, shortened outputs and insights to central control, in turn reducing the associated server costs.
Privacy protection. There is a growing divide between the partisans of video surveillance and monitoring as part of a larger smart city play where video is just a source of data integrated in a broader system, and the people wary of the authoritarian temptations that large-scale surveillance enables. To help implement the privacy laws such as GPDR, CCTV manufacturers and systems designers will have to switch to a “Privacy by design” mindset and build appropriate software to anonymize and securely store the mountains of data produced.
Redundant systems. The magnitude of cyber-attacks is on the rise, and it’s pretty probable we’ll see a large-scale attack on video systems soon. To counter this, hardware engineers are building chips called “system-on-chips” that can be easily isolated from the connected components of cameras, to enable redundancy in case of large-scale attacks.
Healthcare
The COVID-19 crisis has made evident the use of live video in healthcare had a bright future coming. But I think there is many other use cases that are interesting to point out :
Education. Instant live distribution of medical interventions, consultations and surgery is making large-scale education possible and affordable to the masses, which could prove useful for the medical community in sharing cutting-edge new techniques.
Camera-aided remote surgeries. This new type of surgery is not just a fancy invention for show. It helps doctors reduce the degree of invasivity in procedures and increase recovery for patients, and enables the assisted piloting of robotic arms that achieve far better stealth than humans, restoring the focus and accuracy of surgeons.
Integrated Internet of Medical Things. Live streaming could prove to be a useful enabler when connected with other sensors in an integrated ecosystem. With incumbents like Google relaunching their headset and glass initiatives, video feeds can be a good addition to monitoring capacities, when coupled with diagnostic data from other tools
To enable live streaming to be breaking out of the entertainment and gaming space where it gained initial traction, key technological enablers come into play at both the infrastructure and software layers :
5g and widespread satellite-based internet. Both technologies have the power to significantly reduce the friction associated with streaming and interacting with streamers in a variety of contexts. But it also gives way to new formats for live streaming. For instance, vastly increased speeds and lower latencies means we can envision holographic displays to be more widespread in media broadcasting, that we can decongest networks in crowded places such as sporting venues, enable a more widespread adoption of live AR and VR feeds of real-life events, etc. 5G internet has a big caveat: although more economical and energy-saving than the previous generation of cellular networks, the density it demands to be functioning well at that stage makes it difficult to scale worldwide. With the advent of cheap, private cargo spearheaded by players such as Space X’s Starlink broadband network, satellite internet is another route. Just to draw a quick comparison, comprehensive 5G coverage in the US would cost $150bn in infrastructure investment, while Starlink could be completed for as little as $10bn
Edge computing for streaming tech. Today, the major pain experienced in live streaming is the cost of running an application on a centralized server hosted by a few large cloud providers. Edge computing can help declutter traditional networks by providing users with storage and computational resources necessary to live streaming closer to them, and will act as yet another enabler of low-latency and speed. It can play a big role in reducing cost and nocive emissions linked to our gargantuan consumption of data, 80% of which has already been eaten up by video streams.
Evolving low-latency protocols. Over the past ten years, the streaming industry has been hard at work driving down latency. Several cutting-edge protocols will serve as the tools to deliver low-latency streaming at scale. These include Web Real-Time Communications (WebRTC), Secure Reliable Transport (SRT), the Common Media Application Format (CMAF), and Apple’s Low-Latency HTTP Live Streaming (LL-HLS) extension.All of these protocols are available today, and wide-scale adoption promises to follow. In the years ahead, companies will put them to the test for a variety of use cases. Here’s a synthesis of the pros and cons for each of these new protocols :
If you liked this issue, don’t forget to like, subscribe and follow me at @LarocheUlysse on Twitter ! Catch you there :)
Ressource list :
Unbundling Zoom, JJ Oslund
Amazon is Alibaba’s Live Streaming Without the Good Bits, Forbes, Lauren Hallanan
E-Commerce in 2019, The Year of The Lifestreamer, Jing Daily
The Rise of Tiktok and Understanding its Parent Company Bytedance, Turner Novak
Why SpaceX is Making Starlink, Real Engineering
Edge Computing For Live Media and Video Streaming, Kashif Bilal, Aiman Erbad
Live Commerce Has Risen In The East, Will It Settle In The West?, Contagious