Low latency has moved from a technical requirement to a commercial differentiator for live sports operators. With the FIFA World Cup underway, we examine where the benchmarks stand today, the trade-offs involved, and what Media over QUIC could change next.
Managing this gap is one of the most technically interesting challenges in streaming, and one of the most commercially significant. Indeed, highlighting low latency has become a marketing tactic for live sports, Sky UK advertising a “Real Time” mode for its own consumer hardware ahead of the World Cup (the trade-off is that live pause and rewind become unavailable).
In short, low latency is no longer a secondary concern for operators, it is a competitive differentiator.
The business case for low-latency streaming in live sports rests on three connected factors.
The most direct of these is churn. A viewer who perceives that one service is meaningfully closer to real-time than another is given one additional reason to use the service. And one who perceives their service is slower has one more excuse to leave. Interestingly, Sky’s initiative is not even tied into its own channels, but rather the free-to-air coverage being offered by UK broadcasters. The goal is to tie consumers into its overall ecosystem.
Quality of experience is closely related to this. Latency is one of the key factors that helps shape whether watching live sports feels like a genuinely live experience or just a very quick replay. For viewers who engage with second-screen content, social media commentary, or sports betting during a match (and increasingly, data says that is most of them), any observable delay creates a frustration that tends to degrade the experience as a whole.
Sports betting is the third driver, though it requires specific functionality to enable in-play wagering, not just a low latency broadcast. It is, however, a use case that is expanding rapidly, as well as being the one that places the most demanding latency requirements on the delivery chain.
Standard OTT delivery typically operates with a glass-to-glass delay of 15 to 20 seconds. This is the gap between the broadcast signal and what a streaming viewer on the same network sees at the same moment and is generated by all the additional processes involved in delivering streaming video. These are encoding, packaging, and player buffering, with the latter contributing by far the greatest number of seconds to the process.
Real-world data shows that the problem may be even worse. The annual Super Bowl is often used as a benchmark for live sports performance, and this year delivered broadcast feeds that were 19 seconds behind the action versus 48 to 62 seconds for various streaming platforms. That indicates a gap of at least 29 seconds.
Currently, with low-latency configurations enacted, this gap can be reduced to approximately five seconds on a stable network. This is not a theoretical figure; it is an achievable and operationally sustainable target under normal conditions that has been delivered by several broadcasters worldwide.
Going below five seconds is technically possible, but the stability trade-offs become more pronounced. The problem is that with so many of the processes in the chain fixed and immobile (encoding, packaging, etc), the place where you can exert most pressure is in the buffer.
That means that every meaningful reduction in latency is also a reduction in the buffer that protects against network problems. A stream configured for five seconds of glass-to-glass latency has less margin to absorb a brief network fluctuation than one configured for fifteen. And it should not really need to be pointed out that rebuffering events are to be avoided at all costs. If consumers dislike those when watching video-on-demand content, they positively hate them during live sports.
This trade-off of latency versus continuity is particularly difficult to manage on mobile. Cellular networks and Wi-Fi connections both contain more variability than a fixed broadband connection: signal strength changes as a user moves, cell load fluctuates as others connect and disconnect, and Wi-Fi interference comes and goes. The consequence is that the latency target appropriate for a set-top box on a stable home network may be wildly off target when talking about mobile delivery of the same event.
What is needed therefore is an approach that differentiates latency targets by delivery context. This allows the player to manage latency more conservatively on mobile and accept a slightly longer delay in exchange for greater continuity, and can be reversed when supplying video via broadband. This produces a better overall viewer experience than applying a single aggressive latency target across all device types.
Further change is possible. The OTT protocols that dominate live streaming today (DASH, HLS, CMAF) are built on TCP-based transport. TCP is reliable and well-understood, but its congestion control mechanisms introduce a firm floor on achievable latency that becomes increasingly difficult to push below.
The emerging response to this is Media over QUIC, or MOQ. MOQ moves live media delivery onto a different transport foundation which is based on broadcast-style delivery rather than the request-response model of HTTP streaming. In principle, MOQ enables glass-to-glass latency of below one second. This will close the gap with traditional broadcast delivery to a degree that DASH and HLS are simply not engineered to match.
MOQ is currently at the active standardization stage and is a subject of significant industry attention. It is not yet widely deployed at production scale, and full device ecosystem support will also take time to develop. That means that the existing delivery technologies will remain the dominant delivery formats for live sports for the foreseeable future.
However, MOQ represents a meaningful shift in what is technically achievable. As the standard matures and device support grows, it is highly likely to become the preferred protocol for high-value live events where the marginal cost of the more complex infrastructure is justified by the premium nature of the content.
The question of whether OTT can achieve parity with traditional broadcast latency is, for practical purposes, already close to being answered. The gap between a live IPTV signal and a well-configured OTT stream is currently measurable in seconds rather than tens of seconds. With the introduction of MOQ, the difference could be reduced to approximately one second. For most viewers this will be imperceptible.
There are still issues, of course, though these tend to be adjacent to the pipeline rather than a direct component. Server-side ad insertion needs to be managed carefully as it can add significant latency, as can the granting of DRM licenses, especially at high volume events.
But the distinction between broadcast and streaming, at least when it comes to talking about latency, is narrowing fast. By the next World Cup, co-hosted in 2030 by Morocco, Portugal, and Spain, it may no longer be a meaningful distinction.