Ads will solve the Live streaming equation
The revenue of broadcasters comes from two baskets: subscription and advertising. While “free to air” broadcasters only rely on ads, pay-tv broadcasters usually use a mix of ad-supported and subscription-based model.
It’s now a no-brainer that streaming is the de-facto model for watching Video On Demand, and will be the de-facto model for watching live in the next couple of years. But the rise of so many different streaming platforms comes with an increase of the overall price for the end-user. You now have to subscribe to multiple services (Netflix, Disney+, …) to have access to your favorite content, instead of relying on your favorite cable operator to aggregate it. Advertising will be the key to balance this price increase and make it sustainable for the end users.
For instance, the recently announced Comcast/NBC Universal platform Peacock will use a mix of AVOD (Ad-supported VOD) and SVOD (Subscription-supported VOD): you can either pay $4.99 per month and watch ads, or pay $9.99 for the commercial-free version.
What is true for VOD will be true for Live, on a much larger scale: live streaming is hardly profitable today because of the unicast nature of streaming. The money spent on live streaming increases with the number of viewers, and ads are still difficult to monetize because of technical limitations that prevent the broadcasters from having the appropriate level of control on ad insertion and monetization. Over-The-Top (OTT) will continue growing only if the financial equation makes sense; content personalization and targeted ads are the only way to make it happen.
Unicast is not a technical threat, it’s a business opportunity
The unicast nature of streaming is often depicted as a threat for the Internet. “Unicast” means that every single viewer will have its dedicated transport layer to watch its stream. On the opposite, traditional systems (satellite, cable, terrestrial) use a broadcast scheme, where the transport layer is shared amongst viewers.
Unicast video accounts for 80% of the overall Internet traffic, and this will only get worse in the future. There have been several initiatives to address that:
P2P streaming will relieve the CDN (both on the egress and on the interconnection), but comes with some limitations that prevent it from being massively adopted
Multicast ABR will relieve the core and the access network, but can be extremely difficult to put in place and operate.
The OpenConnect initiative (Netflix) is a great tool to relieve the Telco CDN, but can’t scale as the number of streaming services grows. Moreover, it can’t really be used for Live.
OpenCaching, led by the Streaming Video Alliance, is another option but will require a long and difficult joint effort amongst CDN providers and content providers, both on the technical side and on the business side.
On top of offering a more reliable viewing experience, all these initiatives lower the cost of streaming; unfortunately, they don’t succeed in making it economically sustainable. Lowering the expenses is important, but streaming, above all, offers an amazing and unique opportunity to increase revenues. With unicast, content providers now have as many potential channels as the number of viewers.
Ad insertion techniques and limitations
The commercial agreement between the broadcaster/content owner and the advertiser is based on the number of “views” (“impressions”); having a reliable way of acknowledging that a commercial has been watched is the only way to get money from the advertiser. In other words, the broadcaster can’t afford losing control on how the ads are received, played, and reported by the end user device.
Several studies show that targeted ads are well accepted: the end users prefer viewing ads that are relevant for them. They can be priced up to 10 times higher than non-targeted ads. Let’s see how ads can be targeted with current technologies.
Client side ad-insertion
Client-side ad insertion (CSAI) has been used in broadcast schemes (when authorized by the legal operator) to personalize the ad on the receiver side. During breaks, the device will poll commercial from an ad-server and insert those ads in replacement of the broadcast feed. The replacement is done based on SCTE markers inside the stream. This technique comes with several drawbacks:
You have to make sure that all the devices used to have access your content properly support those SCTE markers,
You also have to make sure that the streams that you receive from you own post-production site or from your affiliates will have the appropriate markers!
You have to make sure that these devices have all the required updates when the standard evolves,
Ads are usually not prepared (encoded, packaged, etc) in the same way as the live feed itself, leading to a discontinuous Quality Of Experience for the end user: the resolution may not be same, the bitrate may not be same, ...
Even worse, getting ads on a remote server may take time and lead to a risk of delay (the famous spinning wheel !) in displaying the content (both when switching to the ad and from the ad). This can be catastrophic as viewers can’t accept to miss a single part of their favorite sport event.
As one can guess, the fact that there are specific markers in the stream for the player to insert the ad … makes it extremely easy to implement ad-blocking capabilities.
The broadcaster has limited control on the ad playback and may not be able to properly monetize the ad slots.
For all these reasons, CSAI is now losing traction, especially in the streaming space, were other techniques can be used. It’s still predominant in the broadcast world where there are limited options to offer a personalized ad.
Server side ad-insertion
Streaming offers much more flexibility than broadcast. The same logical bricks are used for server-side ad insertion (SSAI), but they are combined in a slightly different way, as the ad replacement is done upstream, hence the naming “server-side”. This solution is sometimes referred to as “playlist manipulation”, as “someone” will take the live stream, replace specific segments of the manifest by the ad asset, so that the player receives a continuous stream, empty of SCTE markers, with limited possibility of discriminating between the live stream and the ads.
The replacement can be done “before” (usually in the origin, or right after it), “in” , or “after” the CDN (leveraging edge functions). It’s worth noting that the player, in the vast majority of the use cases, will still have to download the ad from a different location than the live stream. Server-side ad insertion provides a better level of control on the ad slots, but the replacement is usually still done outside of the broadcaster network: there is no absolute guarantee that the ad will be correctly replaced, played, and reported by the player. Server-side ad-insertion also comes with a number of drawbacks:
SSAI can still be complex to put in place, although this is, by essence, compatible with all the devices. You still have a complex workflow to put in place to make sure the SCTE markers are correctly set all along your production chain, including affiliates or third party feeds you may aggregate.
The generated playlists are not cacheable (at least the personalized portions), as they are meant to be unique per user; this can cause a bottleneck for the ad replacement server.
Stream manipulation is done at the playlist level, giving no guarantee that the commercial characteristic will match the live stream in terms of resolution, bitrate*. The spinning wheel is less likely to happen, but is not guaranteed to disappear …
SSAI generates playlists with EXT-X-DISCONTINUITY (in HLS) or makes use of multi-period (in DASH), which may not be equally supported by all the devices/players.
Long story short, SSAI brings several benefits overs CSAI and is now predominant to offer targeted ads in the streaming space. Nevertheless, SSAI may not be able to provide the appropriate level of control on ads for the broadcaster, and, in return, may not give the expected ROI to the latter when negotiating with the advertiser. The complexity of the workflow to put in place is also a hurdle as it is likely to prevent quick evolutions, in a world where new services can be launched by the competition is less than a month ...
The Streaming ad-insertion Opportunity
To summarize, broadcasters and content owners need to make sure that:
The commercials do not create a discontinuity of experience for their users (otherwise they are likely to leave the stream: people are unforgiving)
They have a perfect control on the ad reporting (so they can correctly sell their placement opportunities)
Neither CSAI nor SSAI provide the right level of control, as broadcasters rely on ‘someone else, somewhere else’ to insert the ad.
The Quortex workflow can be leveraged to rearrange the ad-insertion workflow and process the ads in your own network, ensuring a perfect control and an optimal monetization of the commercials.
By essence, our solution processes everything on the fly (that’s even our marketing mojo!), and provides a per-user (or per group-of-user) quality of experience. We have the unique ability to switch from sources to sources seamlessly, going from/to Live to/from Ads, and this is done on the fly, as users start to pull their streams. The ads will then follow the exact same workflow as the live stream, ensuring a perfectly smooth and continuous experience for the end user. On the reporting side, things are way simpler: you have the absolute guarantee that the ad is viewed as it is part of the stream itself. This is what we call Just-In-Time Playout Side Ad Insertion (abbreviated PSAI !).
* Some solutions will transcode on the fly the commercial, now that VAST supports embedding a link to the mezzanine ad. But this solution is not compatible with modern solutions that adjust the rendition ladder to the network and audience dynamics.