Investments in broadcast technology have always been geared to commercial reality and evolving viewer consumption patterns, with feedback between these. That is true more than ever today with the severe economic headwinds facing most sectors, ensuring that innovations must be affordable and likely either to cut costs or boost revenues. This is evident in the growing application of AI in broadcasting, which means primarily machine learning (ML), for optimization across the workflow and also for helping improve quality and experiences to attract or maintain viewers.

It also applies to targeted advertising, which will be a major theme for 2023 as more brands and broadcasters embrace it to improve efficiency and conversion rates of ad distribution. The trend towards ad targeting has been driven also by evolving audience behavior, especially the rising consumption of ad supported VoD (AvoD) content in developed economies.

AvoD has long been more popular than Subscription VoD (SVoD) in developing countries such as India and Malaysia where many consumers are unwilling to pay more than a small fee at best for viewing, and as a result ARPUs (Average Revenue Per User) are typically very low, often around $2 to $5 a month. More recently AVoD viewing has grown substantially in the USA, driven by major services such as Viacom’s Pluto, Fox Corporation’s Tubi TV, The Roku Channel, Comcast’s Peacock with an AvoD supported version, and Amazon’s IMDb TV. This has been followed by a lesser uptake in European countries, with a common driver being fragmentation of the existing SVoD scene, leading consumers to seek out lower cost options for viewing at least most of the content they want.

Combined with continuing growth in video streaming to some extent at the expense of traditional linear TV viewing, the AVoD boom is stimulating targeted advertising as the most effective route to monetization. The most visible notice of this trend came in April 2022 when Netflix became the last of the major subscription VoD streamers to capitulate with the announcement it would launch ad supported versions of its packages in various countries.

Such a launch came in the UK in October 2022 with an AvoD package costing £4.99 a month, compared with £6.99 for the cheapest basic ad free offering. This came after Netflix had lost over a million SVoD subscribers globally over the three months April to July 2022. So having for years insisted it would never carry ads, because it was following a different model from say Facebook and Google, Netflix was forced to bow to commercial reality, as other video services had already done.

This comes at a time ironically when there has been some kickback against ad targeting in the purely online digital world, as both Apple and Google have made it more difficult to target individuals in order to meet tightening rules and expectations over privacy. To some extent this developing dichotomy between online and TV over targeting reflects conflicts in the minds and attitudes of consumers, who on the one hand constantly say they prefer to receive ads that are relevant to them, but at the same time resent being bombarded with ads at all when engrossed in their viewing.

Brands, broadcasters and content producers therefore have a common interest in squaring this circle. As in a growing number of other parts of the workflow, ML is being brought in to help target ads effectively and sensitively, not just for delivery but also for creating the content itself. Admittedly both these are at early stages of development and implementation, but the efficacy is already becoming well established.

On the targeting side, ML can mitigate to some extent the impact of privacy constraints, by identifying behavior of individual consumers that correlates with demographic traits useful to advertisers, such as gender, age, and preferences for certain goods or services. Location has also become valuable in the mobile age, given the potential for targeting ads on that basis, say for shops and restaurants in the locality. That can be combined with ML to target ads effectively.

Time is another factor that can be added to the analysis to exploit knowledge for example of a viewer’s changing circumstance, such as leaving a job or moving to a different area. ML can help out by enabling deductions to be made for targeting purposes even when the consumer’s identity is unknown. It can also help ensure ads are targeted sparingly to avoid annoying the recipient, and ensuring they are as relevant as possible. This is not an exact science and mistakes will be made, but there is no doubt there is great scope for both improving viewer satisfaction and increasing ad conversion rates, whether that goes as far as making a purchase, or just showing interest.

These are early days though and Netflix itself has confined targeting so far just to basic geolocation and content category, with as yet no third-party segmentation or first party retargeting.

First party data is derived from a service provider’s own subscribers only, so is quite comprehensive but usually limited in scale. The retargeting refers to use of this data say for targeting ads. Second party data is then derived from partners or other sources that provide it with consent and is again accurate with the advantage here of larger scale, although also the form most likely to be curtailed by privacy constraints. Third party data is often more anonymized and summarized, shared between many companies, but has the virtue of being readily available.

ML can also help with development of the ad itself, more so than for the longer form entertainment, documentary or other content surrounding it. The idea is to converge around ad types, genres or categories that are already known to appeal to viewers. This can be more effective than traditional A/B testing which can be cumbersome and not guaranteed to derive the best results. The two can work in tandem with ML helping direct content production verified at some stages by A/B testing.

Such developments play into the growing value of streaming content, which is more conducive for ad targeting than traditional linear, noting that while it still accounts for only 8% of overall viewing it now generates 30% of revenue.

Another trend driven by streaming through a combination of emerging technology and evolving viewer behavior is highlights or clips of events just after they are shown live. This applies particularly to live sports and the growing value of highlights packages has just been underlined by the 2022 FIFA World Cup.

While the live stream still accounts for nearly all of the viewing, especially of major events, consumption of near live highlights packages or clips is rising fast, including at the event itself where fans increasingly review actions such as goals, they either missed or did not see in as much detail as they would like. Highlights or clips can show the actions from different camera angles and also allow simulation through a combination of AI and Augmented Reality (AR) to allow “what if” scenarios to be played, of potential value in sports training as well as for entertainment. Highlights packages can also be offered by broadcasters or sports leagues themselves at low prices or free to entice new, often young, viewers, hoping to convert some into longer lasting fans.

The ability to simulate action is also extending into new forms of sports coverage that combine the real with the virtual, in a sense converging physical sports and esports. A grand example of that is the Drone Racing League (DRL), where top drone pilots fly, attracting large audiences that are then also drawn into simulations. DRL then combines fast competitive IRL (In Real Life) racing and virtual simulation, while doffing its hat to the metaverse.

DRL also exemplifies the trend towards making content available in all the streaming places people visit, including YouTube, Facebook, and the more recently emerging social media platforms like TikTok. These are also becoming vital destinations for traditional content producers and broadcasters, as again we have seen in events like the FIFA World Cup where rights holders would make highlights packages and clips available free.

This leads to another evolving trend that is highly relevant for major broadcasters and content producers or rights holders, including major sports events and leagues, which is exploiting synergy between different services and platforms. On this front Amazon is a pace setter, noting its cross fertilization between its various outlets under the Prime brand, certainly eCommerce and video.

Amazon CEO Jeff Bezos boasted as much when he commented, “Every time we win a Golden Globe we sell more shoes.” Few broadcasters or rights holders are in ecommerce of course but they can exploit the same principle, especially as they fan out across multiple streaming and social media platforms. The primacy of video is increasing and the lines between long and short form content are blurring, with scope for exploiting synergies for gaining eyeballs and deriving extra revenues through ad targeting. The key point is that streaming offers in various ways the potential to atone for revenues lost, whether through declining linear viewing, or where applicable shrinking income from licence fees.