Ad tech revolution: how AI, dynamic ad insertion and personalised streaming ads are transforming digital advertising

Between advancements in automation and hyper-personalisation, advertising technology, or ad tech, is currently in the middle of a makeover. Shoppable ads, dynamic insertion, split screens and real-time overlays are just a few of the ways in which advertisers are capitalising on our collective digital consumption.

The state of ad tech today is “marked by industry consolidation and rapid technological shifts,” begins Mike Shaw, director of EMEA ad sales at Roku. He believes artificial intelligence is arguably the biggest disruptive force of all, poised to streamline workflows, solve problems, increase opportunities for personalisation and aid in data collection and analysis.

Meanwhile, Jean-Louis Lods, VP of media and monetisation at Ateme, touts the significance of ad- tech evolutions like server-guided ad insertion (SGAI), which can ‘improve underlying costs’, reduce interruptions (if executed well) and increase monetisation opportunities, particularly during live events. He has also witnessed a shift towards ‘standards-based approaches that allow broadcasters and platforms to scale monetisation across devices and formats’, which is essential to making the most of our modern viewing habits.

Adaptive ads, at scale

Compared to what appeared on traditional, linear television, today’s ads look very different. Streaming has become the dominant method of consuming video content, whether it’s a TV series, a sporting event or even a live news segment. While the streaming services themselves – Netflix, Amazon Prime and Disney+, for example – were initially ad-free, they now offer subscriptions with ads, though at a lower cost. This has opened the door for interactive advertisements that make use of both large and small devices.

Similarly, there is an abundance of FAST (free ad-supported streaming television) channels, such as those under Roku’s umbrella. Dynamic ad insertion (DAI) is one of the biggest innovations in this arena, allowing “specific ads to be delivered at precise moments,” adds Shaw. “Its complexity comes from timing and making sure there is smooth playback without interrupting the viewer experience.”

According to Shaw, “DAI involves stitching personalised ads into high-volume live streams without latency, making the process high risk and challenging. Any issues can lead to a poor user experience and under-monetisation of premium assets.”

Typically handled on the server side, Lods explains that “DAI places ads directly into the content stream before it reaches the user’s device, so the viewer experiences one seamless video rather than a jarring transition.” He goes on to mention: “The biggest hurdles are preserving video quality and a seamless user experience, managing latency for live streams, scaling to large audiences and also ensuring consistent behaviour over a fragmented device ecosystem.”

While DAI can be high risk, it can also be high reward. “DAI has given streaming platforms broadcast-quality ad delivery, with the key difference being personalisation,” explains Lods. “Two households watching the same show can see different ads, which makes the load feel far less repetitive than linear TV.”

Shaw adds that ‘streaming ad loads are still significantly lighter than linear TV’ – despite adopting many of the same strategies – and ‘TV-style sponsorships and first-in-break video are included’.

“The next challenge,” argues Lods, “is to leverage the flexibility of the user experience.” He suggests that deploying less distracting or all-encompassing ads, such as ‘overlay and side-by-side ad presentations’, will keep the consumer focused on the video’s content – but not so engaged that monetisation opportunities are missed entirely.

Dynamic ad insertion is not just exclusive to video. Radio stations and podcast providers, such as SiriusXM, can also place ads based on genre, the listener’s location and even the weather where they are – rather than sticking them with an embedded, permanent ad that becomes irrelevant after the first listen.

Spontaneous creation and automation

Artificial intelligence has certainly made its mark on the entertainment industries, and now it’s coming for ad tech, too. “AI is both solving traditional ad-tech problems and changing media company workflows, leading many people to reassess their existing partnerships and advertising stacks,” shares Shaw. “A new wave of AI companies, some highly specialised, others more general, are entering spaces that ad-tech companies once dominated, integrating focused AI capabilities into established platforms.”

Besides catalysing mergers and acquisitions and generally reshuffling the corporate landscape, AI is expanding what ads can do – and how quickly they can take shape. In particular, large language models (LLMs) and generative AI are leading the charge in letting advertisers make live adjustments. “Imagine the ability to generate copy, adjust visual layouts or even synthesise personalised voiceover in real time based on audience signals,” Lods suggests. “The ability to produce thousands of creative variants at low cost changes the economics of personalised advertising on a fundamental level.”

Artificial intelligence, while useful in efficient ad creation, also touches video content analysis, ‘dramatically improving’ this process, according to Lods. “Contextual targeting – targeting based on what content surrounds an ad rather than who the viewer is – has become a much more viable, privacy-safe alternative to audience-based targeting, in part because LLMs understand nuance, sentiment and brand safety risk in ways that earlier keyword-based systems were simply not able to,”
he explains.

Lods is right to mention privacy. Local regulations and the availability of information pose barriers to genuine personalisation, thus limiting an ad’s effectiveness. While companies might argue that compliance hinders business, consumers do – presently, at least – have the right to keep some of their information (demographics, viewing habits and the like) to themselves.

Failings and future opportunities

In the simplest terms, ad tech exists to turn audience engagement into earnings, but sometimes that return on investment is tricky to quantify. To account for the dynamism of digital environments, Shaw says: “Advertisers are moving from traditional reach-and-frequency models and longitudinal brand studies to real-time, user-level measurements. They can now link exposures to actions – like website visits, purchases or app interactions – using granular data.

“Advanced measurement is the next frontier,” he adds. “As the industry moves away from panel-based audience tracking to hyper-personalised, campaign-centric metrics, there is a big opportunity to bridge gaps and improve insights for advertisers while enhancing the viewer experience.”

For now, faulty measurement tactics and other key issues, such as fragmentation, remain. “Inconsistent standards across publishers can lead to overexposure,” Shaw explains. “Viewers may see the same ad ten to 40 times, even within the same show.” If the goal is engagement, that’s a sure-fire way to fail.

“While ad tech can effectively emulate the traditional TV viewing experience, there are many areas where it can improve,” adds Lods, who points out similar shortcomings. “Challenges remain when trying to maintain viewer experience at scale, especially when ad loads or latency are poorly managed.”

For Lods, “the future lies in lower-latency ad insertion for live content, improved standards such as SGAI, as well as shoppable and interactive ad formats.” But, in the long term, he sees ad tech moving towards real-time, personalised creative assembly, or put in simpler terms: “ads built dynamically for the individual in the moment. This is where generative AI and DAI infrastructure are heading.”

This article appeared in our NAB 2026 issue