By Shirish Nadkarni
Latency limit: 150 milliseconds. Yes, milliseconds … this is the time limit permitted in the definition of low latency for keeping the end-to-end time from ad request to rendering. Longer delays induce yawning and boredom among even the least impatient of TV viewers and will significantly reduce click-through rates and viewability.
In modern AdTech, low latency, combined with targeted advertising, is critical in the delivery of personalised advertisements, ensuring high engagement without disrupting the user experience.
In real-time advertising, speed is not a technical preference; it is a revenue determinant. Each millisecond of bid request-ad delivery affects the win rates, user experience and campaign return-on-investment (ROI).
With the increased competition of programmatic ecosystems, both advertisers and platforms are re-evaluating their systems in terms of how they are processing data, making decisions, and acting at scale, in real-time.
AWS Media Services are a set of managed, cloud-based tools — primarily powered by AWS Elemental — designed to transport, prepare, process, and deliver secure high-quality live and on-demand video to any device. These services enable scalable video workflows, including encoding, packaging, and monetisation, reducing the need for traditional hardware.
AWS’s Define Media tool has been using machine learning (ML) for the past two years at least to optimise ad placement with low latency. Their architecture leverages AWS services like Lambda, EKS and S3 to process 100,000 predictions per minute.
The system uses embeddings and a two-step neural network approach to make fast decisions on ad placement based on context, user location, and interaction data. This innovative solution allows Define Media to deliver highly targeted advertisements efficiently at scale.
In the first tiers of contemporary ad platforms, Custom AdTech Software Development is highly important in defining the efficiency on how real time decisions will be processed, particularly in instances where off-the-shelf applications impose processing delays that are exponential under heavy traffic.
There is cumulative latency in AdTech. It assembles based on bid requests, data lookups, targeting logic, creative selection and ad rendering. The slightest system inefficiencies in any of the steps may lead to loss of auctions or inefficient placements.
Due to the evolution of demand-side and supply-side platforms, the tolerance of delay is decreasing. Advertisers now anticipate that systems can analyse user context, campaign requirements and performance indicators on a near real time basis, leaving no time to spare for wasteful computing.
“That is why a lot of platforms resort to consultation services of specialised AdTech developers in order to restructure their implementation layers instead of using incremental optimisations,” said Elodie Levrel, Corporate Marketing & Communication Executive with Broadpeak, a company that provides a top-tier server-side ad insertion solution for live, cloud digital video recorder (DVR), and video-on-demand (VoD), designed to eliminate re-buffering.
Several companies and platforms specialise in low-latency server-side ad insertion (SSAI), focusing on eliminating buffering and providing uninterrupted ad experiences for live and VoD content. However, the latest buzz surrounds server-guided ad insertion (SGAI).
“A big topic in dynamic ad insertion at the moment is SGAI, which is an advancement SSAI built on industry standards,” said Paul Davies, Head of Marketing & Partnerships, with Yospace, a company that was the first to showcase SGAI at low latency using the new MPEG-DASH Events standard at IBC 2025.
“We are the acknowledged world leaders in real-time ad stitching and are able to achieve sub-five-second glass-to-glass latency, with ad insertion responsible for less than one second of that latency.”
SGAI has generated a lot of interest especially among sports broadcasters who have an opportunity to offer low latency with advanced advertising for the first time. New AdTech standards are paving the way for innovative ad formats for Connected TV (CTV), such as L-shaped banners and side-by-side ads, where live production examples can be expected to be in 2026.
“SGAI and new ad formats will continue to be big topics at the forthcoming NAB Show at Las Vegas in April,” Davies added. “We will be demonstrating the latest advancements, of course. Low latency with targeted advertising will continue to gain traction towards live production use cases, which will transform the sports viewing experience and monetisation.
“Another advancement that we will showcase is the latest tools to improve ad measurement and viewability data. This topic isn't discussed much but is absolutely critical to advertisers and will be key in persuading them to spend more on CTV.”
Other key players in low-latency SSAI and SGAI include Phenix Real Time Solutions, which provides an SSAI technology that, in partnership with Amagi, delivers end-to-end latency in under half a second (sub-second). Amagi itself is well known in the broadcast industry for its cloud-based broadcasting, offering low-latency, frame-accurate ads for live and linear free ad-supported streaming television (FAST) channels.
Harmonic offers the VOS ad solution for FAST platforms, providing cloud-native, high-performance, low-latency ad insertion, while Velocix offers a Video Personalisation Platform (VPP) with a microservices-based design that allows for ad placement transactions within microseconds.
Broadcasters are also looking closely at Synamedia, which offers a cloud Software-as-a-Service (SaaS) SSAI solution designed to replace or insert ads into live/VoD content without causing latency; Castlabs, which enables SGAI for over-the-top (OTT) and live IPTV through PRESTOplay to improve ad personalisation; and Aniview, which provides high-volume ad stitching for various platforms, including Samsung, LG, TCL and others.
To achieve both speed and relevance, platforms use several advanced techniques, like edge computing and caching, pre-fetching and local storage, SGAI, real-time bidding optimisation and ML models. Edge computing, for example, moves data closer to the user, as systems like content delivery networks (CDNs) minimise the physical distance data travels, significantly reducing load times.
“Ads that load quickly are more likely to be seen and clicked, whereas delayed ads often result in higher bounce rates,” said Broadpeak’s Levrel. “There is also reduced user churn; in streaming or web browsing, fast-loading ads prevent annoying delays, improving the overall user experience and reducing the likelihood of users installing ad blockers. There is thus higher revenue for publishers.”
The emerging new monetisation technologies have provided one huge advancement — scale. In a single year, 2025, Yospace recorded 8.5 billion ads stitched per month, which was up from 6 billion in 2024.
“2026 will see us surpass 10 billion ads per month for the first time,” said Davies. “And all with one-to-one addressability. Multiple technologies on both the streaming side and AdTech side are required to make that happen.
“The most crucial part is the strength of the relationship between the two sides, which is dependent on the know-how of the dynamic ad insertion team to improve performance.”
