The Ultimate CTV & Streaming Glossary 2026: Every Term Broadcast Engineers, Ad-Tech Specialists, and Platform Architects Need to Know
The streaming and connected television industry has a vocabulary problem. Not because the terminology is hard to learn — it is because it changes faster than most professionals can keep up. Every quarter delivers new acronyms, protocol revisions, and business model categories that broadcast engineers, ad-tech specialists, and platform architects need to understand. The alphabet soup of AVOD, FAST, SSAI, CMAF, and SRT can make even veteran professionals feel like they wandered into the wrong conference session.
This glossary exists to fix that. We have assembled over 80 terms spanning CTV advertising, streaming delivery architecture, video encoding, content distribution, measurement, and the business models that determine who gets paid. Every definition is written for practitioners — the kind of people who need to know the difference between SSAI and CSAI not for a quiz but because they are about to spec a new ad insertion pipeline, or who need to explain to a non-technical stakeholder why HEVC licensing is a headache without putting them to sleep. We have tried to make each entry useful, occasionally entertaining, and always grounded in how these concepts actually work in production environments rather than marketing decks.
Use the alphabetical navigation below to jump to the section you need, or settle in for the full tour. Either way, bookmark this page — in an industry that invents a new three-letter acronym every week, you are going to need it.
A
ABR (Adaptive Bitrate Streaming)
The technology that makes streaming actually work on the real internet (as opposed to the imaginary internet where everyone has perfect bandwidth). ABR encodes video content at multiple quality levels (called renditions or profiles) and lets the player dynamically switch between them based on the viewer’s available bandwidth, device capabilities, and network conditions. When your stream drops from 4K to something that looks like it was filmed through a wet paper bag, that is ABR downshifting to prevent buffering. When it gracefully recovers, that is ABR doing its job well. HLS and DASH are the two dominant ABR protocols — and almost everything you watch on a connected device uses one of them.
Ad Pod
A group of ads served together in a single commercial break during streaming content — the digital equivalent of a traditional TV ad break. An ad pod might contain two 15-second spots and one 30-second spot, assembled dynamically based on viewer targeting data, advertiser bids, and frequency caps. The composition of each pod can be different for every viewer watching the same content, which is both the promise and the operational nightmare of addressable advertising. Getting ad pods to play seamlessly — without buffering gaps between ads, without audio level jumps, and without repeating the same ad three times in a row — is a harder engineering problem than most people outside the industry realise.
Addressable TV
The ability to serve different ads to different households watching the same programme, even on traditional linear TV. Addressable TV turns the blunt instrument of “everyone in the DMA sees the same car commercial” into targeted delivery based on household-level data. The technology works through set-top box integration (cable/satellite operators replacing ads at the household level) or through smart TV platforms that overlay targeted ads onto the broadcast feed. Addressable TV has been “the next big thing” for about fifteen years — and it has finally reached meaningful scale, though measurement standardisation and cross-platform deduplication remain unsolved problems.
ATSC 3.0 (NextGen TV)
The next-generation broadcast television standard in the United States, designed to bring IP-based capabilities to over-the-air television. ATSC 3.0 supports 4K HDR video, immersive audio (Dolby AC-4), interactive content, targeted advertising, and — crucially — the ability to combine broadcast delivery with broadband return-path data. Think of it as broadcast television that finally learned to talk back. The standard uses ROUTE/DASH for delivery (replacing the MPEG-2 transport stream of ATSC 1.0) and HEVC for video compression. Rollout has been slower than advocates hoped, but the standard’s ability to deliver robust, one-to-many distribution at scale while enabling personalised ad insertion makes it genuinely interesting for both broadcasters and CTV advertisers.
AVOD (Advertising-Based Video on Demand)
A business model where viewers watch content for free (or at a reduced subscription price) in exchange for watching advertisements. AVOD has gone from streaming’s scrappy underdog to arguably its most important growth engine, as subscriber fatigue and price sensitivity have pushed every major platform — including Netflix and Disney+ — to launch ad-supported tiers. The economics are straightforward in theory (advertisers pay, viewers watch for free) and brutally complex in practice (ad fill rates, yield optimisation, frequency capping, brand safety, measurement attribution, and the perpetual tension between ad load and viewer retention). Tubi, Pluto TV, and the ad tiers of major SVOD platforms are the defining AVOD properties of 2026.
AV1
A royalty-free video codec developed by the Alliance for Open Media (whose members include Google, Apple, Amazon, Meta, Microsoft, Netflix, and others). AV1 was specifically designed to be a patent-free alternative to HEVC, which was plagued by fractured licensing that made cost-of-use unpredictable. AV1 delivers compression efficiency roughly 30 percent better than H.264 and competitive with HEVC, but encoding is significantly more compute-intensive. Hardware decoding support has expanded rapidly — most devices shipped since 2022 include AV1 decode — and YouTube, Netflix, and other major platforms have adopted it for delivery. AV1 is the codec the industry has been wanting HEVC to be: good compression, no licensing headaches.
B
Bitrate
The amount of data processed per unit of time in a video stream, typically measured in megabits per second (Mbps) or kilobits per second (kbps). Higher bitrate generally means better visual quality — but the relationship is not linear, and a well-encoded stream at a lower bitrate can look better than a poorly encoded stream at a higher one. A typical 1080p stream runs at 4–8 Mbps; 4K HDR can demand 15–25 Mbps. Bitrate is the single most important variable in the eternal triangle of quality, bandwidth cost, and viewer experience. Every CDN bill, every encoding decision, and every ABR ladder design is fundamentally a conversation about bitrate.
Buffering
The spinning wheel of death. Technically, buffering is the process of pre-loading video data into a player’s memory before playback, which is actually a good thing — it is what prevents playback from stopping every time there is a momentary network hiccup. The problem occurs when the buffer runs empty because the network cannot deliver data fast enough to sustain playback. That is when the viewer sees the loading spinner, questions their life choices, and considers cancelling their subscription. Research consistently shows that buffering is the single biggest driver of viewer abandonment — more disruptive than low resolution, audio issues, or even content quality. If you work in streaming operations, your job is fundamentally about preventing buffering.
C
CDN (Content Delivery Network)
A geographically distributed network of servers that caches and delivers content to viewers from the location closest to them, reducing latency and improving playback quality. Without CDNs, every viewer would be pulling video from a single origin server, and the internet would collapse under the load of a single popular live event. Major CDN providers include Akamai, Cloudflare, Amazon CloudFront, and Fastly. Most large streaming services use multiple CDNs simultaneously (a multi-CDN strategy) to improve redundancy, optimise costs, and ensure performance during traffic spikes. CDN costs are typically the largest line item on a streaming service’s infrastructure bill — which is why per-title encoding optimisation (reducing bitrate without visible quality loss) is essentially a CDN cost reduction programme disguised as a quality initiative.
CMAF (Common Media Application Format)
A standard that enables the same encoded video segments to be delivered via both HLS and DASH, eliminating the need to encode and store content twice. Before CMAF, streaming services had to maintain separate encoded versions of every piece of content for Apple devices (HLS with fragmented MP4 or MPEG-TS) and everyone else (DASH with fragmented MP4). CMAF unifies the container format (fragmented MP4) so a single set of encoded segments works across both protocols. This roughly halves storage costs and simplifies CDN caching. CMAF also supports chunked transfer encoding for low-latency live streaming, enabling sub-3-second glass-to-glass latency — a significant improvement over the 20–40 seconds typical of early HLS implementations.
CTV (Connected TV)
Any television set that connects to the internet and can stream video content — whether through a built-in smart TV operating system (Samsung Tizen, LG webOS, Google TV, Roku TV OS) or through an external device (Roku stick, Amazon Fire TV, Apple TV, Chromecast). CTV is the advertising industry’s favourite new acronym because it combines the emotional impact of big-screen television viewing with the targeting and measurement capabilities of digital advertising. In practical terms, CTV has become the primary battleground for streaming attention — and the device platform you control (or pay rent on) determines your access to viewers, your ad inventory economics, and your data strategy. The CTV device landscape is one of the most consequential competitive dynamics in media, and the platform operators (Roku, Amazon, Samsung, Google) are extracting increasing value from their position as gatekeepers.
CPM (Cost Per Mille)
The cost an advertiser pays per one thousand ad impressions. CPM is the universal currency of video advertising — the metric that lets buyers compare the cost of reaching audiences across linear TV, CTV, AVOD, social video, and digital display. CTV CPMs typically range from $20–$50 (sometimes higher for premium, brand-safe environments), which is significantly more expensive than display advertising ($2–$10) but generally cheaper than traditional primetime broadcast TV. The CPM a platform can command depends on audience targeting precision, content quality, brand safety controls, measurement capabilities, and inventory scarcity. Every ad-supported streaming business is ultimately optimising for revenue per user, which is CPM multiplied by impressions per viewing hour.
CSAI (Client-Side Ad Insertion)
An ad insertion method where the viewer’s device (the “client”) is responsible for fetching and displaying ads, separate from the main content stream. The player receives ad markers in the content manifest, makes ad requests to an ad server, downloads the ad creative, and switches between content and ads locally. The advantage: granular client-side measurement (viewability, completion rates, interaction tracking). The disadvantage: ad blockers can intercept the ad requests because they originate from the client, and the transition between content and ads can produce visible glitches (black frames, audio pops, buffering). CSAI is being displaced by SSAI for premium CTV environments, but remains common in mobile and web players where ad interaction tracking is prioritised.
D
DASH (Dynamic Adaptive Streaming over HTTP)
An international standard (ISO/IEC 23009) for adaptive bitrate streaming that is codec-agnostic, DRM-agnostic, and not controlled by any single company. DASH uses an XML-based manifest file (the MPD — Media Presentation Description) to describe the available streams, quality levels, and timing information. The player reads the MPD, selects the appropriate quality level based on network conditions, and fetches video segments from the CDN over standard HTTP. DASH is the predominant streaming protocol on Android, smart TVs, game consoles, and web browsers (via the Media Source Extensions API). Apple devices historically used HLS exclusively, though Safari has added limited DASH support. With CMAF enabling shared segments between HLS and DASH, the practical distinction between the two protocols is narrowing.
DRM (Digital Rights Management)
Technology that encrypts video content and controls who can play it, under what conditions, and on which devices. The three major DRM systems are Google Widevine (dominant on Android, Chrome, and smart TVs), Apple FairPlay (iOS, macOS, Apple TV), and Microsoft PlayReady (Xbox, Windows, some smart TVs). If you want your content to play on every device, you need to support all three — which means maintaining three separate encryption and licence-serving workflows. DRM is simultaneously the technology that content owners require as a condition of licensing their content and the technology that streaming engineers most enjoy complaining about, because its device-specific fragmentation adds complexity, cost, and failure modes to every part of the delivery pipeline.
DSP (Demand-Side Platform)
A technology platform that allows advertisers and agencies to buy ad inventory programmatically across multiple publishers, exchanges, and inventory sources through a single interface. In the CTV context, DSPs connect buyers to streaming ad inventory — evaluating billions of potential ad impressions in real time, applying targeting criteria (demographics, geography, viewing behaviour, household data), and placing bids through automated auctions that complete in milliseconds. The Trade Desk, Google DV360, and Amazon DSP are among the largest platforms buying CTV inventory. The relationship between DSPs, SSPs (supply-side platforms), and ad exchanges forms the programmatic advertising infrastructure that increasingly determines how CTV ad dollars flow.
E
Edge Computing
Processing data closer to the end user rather than in a centralised cloud data centre. In streaming, edge computing powers real-time ad decision-making at CDN edge nodes, server-side ad stitching without round-trips to origin, content personalisation based on local device and network conditions, and low-latency live stream processing. Edge compute is becoming increasingly important for CTV as the complexity of per-viewer ad decisioning, DRM licence serving, and real-time quality analytics makes centralised processing too slow and too expensive at scale.
EPG (Electronic Programme Guide)
The on-screen schedule that tells viewers what is on now, what is coming next, and what they can watch later — the digital descendant of the printed TV listings magazine. In the FAST channel world, the EPG has taken on renewed importance because it is the primary discovery mechanism for linear-style streaming channels. A well-designed EPG can dramatically increase channel sampling and time-spent-viewing; a poorly designed one buries content and drives viewers back to on-demand browsing. EPG metadata quality — accurate titles, descriptions, artwork, timing information — is a surprisingly important factor in FAST channel performance.
F
FAST (Free Ad-Supported Streaming Television)
Linear-style streaming channels that viewers can watch for free, supported entirely by advertising. FAST is essentially “cable TV reimagined for streaming” — curated, scheduled programming on themed channels (true crime, classic sitcoms, news, cooking, sports highlights) that viewers tune into without making an active content selection. The genius of FAST is that it solves the “what should I watch” problem that plagues on-demand libraries — the viewer does not have to choose, they just tune in. Pluto TV (Paramount), Tubi (Fox), Samsung TV Plus, Roku Channel, and Amazon Freevee are the major FAST platforms. The channel count has exploded past 1,500+ across platforms, and the content acquisition model — licensing older library content at low cost, sometimes for free in exchange for ad revenue share — has created a viable monetisation path for content that was otherwise sitting idle in studio vaults.
Frequency Capping
Limiting the number of times a specific ad is shown to the same viewer within a defined time period. Frequency capping is one of CTV advertising’s most persistent unsolved problems. In a world where a household might watch content across five different apps on three different devices, maintaining accurate cross-platform frequency caps requires either a universal identifier (which privacy regulations have made increasingly difficult) or sophisticated probabilistic matching (which is imprecise). The result: viewers see the same ad seventeen times in an evening and develop a deep, personal hatred for the advertised brand. Getting frequency capping right is arguably the single highest-value improvement that the CTV ad ecosystem could make — and it remains stubbornly difficult.
H
HDR (High Dynamic Range)
A video technology that expands the range of brightness and colour that can be displayed, producing images with brighter highlights, deeper blacks, and more nuanced colour gradation than standard dynamic range (SDR) content. HDR formats include HDR10 (open standard, static metadata — the baseline), HDR10+ (Samsung-backed, dynamic metadata), Dolby Vision (Dolby-proprietary, dynamic metadata, dual-layer), and HLG (Hybrid Log-Gamma, designed for broadcast compatibility). If you think the streaming protocol wars are complicated, the HDR format landscape is worse — and the practical impact on viewer experience is often undermined by poorly calibrated TV sets and inconsistent content mastering. HDR done well is visually stunning. HDR done badly produces washed-out images that look worse than SDR.
HEVC / H.265 (High Efficiency Video Coding)
The successor to H.264/AVC, offering roughly 50 percent better compression efficiency — meaning the same visual quality at half the bitrate, or significantly better quality at the same bitrate. HEVC was technically a triumph and a licensing disaster. Three separate patent pools (MPEG LA, HEVC Advance, Velos Media), each with different terms and pricing, plus individual patent holders demanding separate licences, created a cost structure so complex and unpredictable that it drove the industry to develop royalty-free alternatives (AV1, VVC). Despite the licensing mess, HEVC is widely deployed — it is the mandatory codec for 4K content on Apple devices, and most modern hardware includes HEVC decode. It remains a critical codec for the foreseeable future, even as AV1 adoption accelerates.
HLS (HTTP Live Streaming)
Apple’s adaptive bitrate streaming protocol — and for many years, the de facto standard for streaming delivery because if you wanted to reach iPhones and iPads, you had no choice. HLS uses M3U8 playlist files (text-based manifests) to describe available streams and qualities, and delivers video in small HTTP-fetchable segments. Originally MPEG-TS based, HLS now supports fragmented MP4 containers (aligned with CMAF), and Apple has added low-latency HLS (LL-HLS) extensions that reduce live streaming latency to 2–4 seconds. HLS is supported on virtually every device, which is both its greatest strength and the reason it remains the most commonly used streaming protocol globally — even though DASH is technically the more open standard.
I
IPTV (Internet Protocol Television)
Television delivered over a managed IP network — typically a telecom operator’s own infrastructure — rather than over the open internet (OTT) or through traditional broadcast (terrestrial, cable, satellite). The critical distinction from OTT streaming is that IPTV operates on a closed, quality-managed network where the operator controls bandwidth allocation, which means guaranteed quality of service but limited reach. AT&T U-verse (now retired), Deutsche Telekom’s MagentaTV, and various Nordic telecom TV services operate as IPTV. The IPTV model is under structural pressure as viewers shift to OTT platforms that work across any internet connection, but the managed network approach still delivers superior quality consistency for live events where buffering is unacceptable.
J
JIT (Just-In-Time) Packaging
An approach where video content is packaged into its final delivery format (HLS manifests, DASH MPDs, DRM encryption) at the moment a viewer requests it, rather than being pre-packaged and stored in every possible format combination. JIT packaging dramatically reduces storage costs (one mezzanine file instead of dozens of format-specific renditions) and enables real-time customisation — different DRM, different ad markers, different subtitle tracks — on a per-request basis. The trade-off is compute cost and latency at request time. Most modern OTT platforms use JIT packaging to some degree, and the technology has become essential for personalised ad insertion workflows where each viewer receives a uniquely assembled stream.
L
Latency (Glass-to-Glass)
The total delay between an event occurring in reality and a viewer seeing it on their screen. Traditional broadcast television delivers latency of 3–6 seconds. Early HTTP-based streaming (HLS, DASH) introduced latencies of 20–40 seconds — long enough that your neighbour’s cheer after a goal arrives well before your stream shows it. Low-latency streaming technologies (LL-HLS, LL-DASH, CMAF chunked transfer, WebRTC) have brought latency down to 2–5 seconds for many implementations, and sub-second latency is achievable for specific use cases (betting, interactive, auction). Every second of latency reduced comes with engineering trade-offs: more frequent segment requests, reduced buffer protection against network variability, and higher origin/CDN load. Latency is the streaming industry’s most discussed trade-off because there is no free lunch — lower latency always costs something.
Linear Streaming
Content delivered on a fixed schedule — just like traditional broadcast television, but over the internet. When people said “nobody watches linear anymore,” they were wrong. FAST channels proved that a significant portion of viewers actually prefer scheduled, lean-back programming over the cognitive burden of choosing from a 50,000-title on-demand library. Linear streaming combines the simplicity of traditional TV (turn it on and something is already playing) with the targeting and measurement capabilities of digital delivery. The distinction between “live linear” (broadcasting a real-time event) and “simulated linear” (playing pre-recorded content on a schedule) matters for infrastructure design but not much for viewer experience.
M
Manifest
The file that tells a video player what content is available and where to find it — the streaming equivalent of a table of contents. In HLS, the manifest is an M3U8 playlist file; in DASH, it is an MPD (Media Presentation Description) XML file. The manifest lists available quality levels (renditions), their bitrates and resolutions, the URLs of the video segments for each quality level, subtitle and audio track options, DRM information, and ad break markers. Manifest manipulation — dynamically modifying the manifest per viewer to insert personalised ad segments, swap audio tracks, or adjust the ABR ladder — is the foundation of server-side ad insertion and content personalisation in modern streaming architectures.
MVPD (Multichannel Video Programming Distributor)
The regulatory and industry term for traditional pay-TV providers — cable, satellite, and telecom companies that distribute bundles of television channels to subscribers. Comcast, Charter, DirecTV, and Sky are MVPDs. The category is losing subscribers steadily (cord-cutting), but MVPDs remain enormously influential because they control broadband infrastructure (the pipes through which OTT streaming flows), maintain large customer bases that are gradually migrating to streaming bundles, and hold legacy carriage agreements with content networks that shape the economics of the entire TV ecosystem. The term vMVPD (virtual MVPD) refers to streaming services that replicate the MVPD bundle online — YouTube TV, Hulu + Live TV, FuboTV, Sling TV.
N
NDI (Network Device Interface)
A royalty-free IP video protocol developed by NewTek (now Vizrt) for high-quality, low-latency video transport over standard Ethernet networks. NDI has become the dominant protocol for IP video in production environments — connecting cameras, switchers, graphics systems, and recording devices over standard network infrastructure rather than dedicated SDI cabling. NDI works on any gigabit network without special configuration, supports auto-discovery (devices automatically find each other on the network), and handles video, audio, and metadata in a single stream. Its adoption has been transformative for remote production, cloud-based workflows, and smaller broadcast operations that could not afford dedicated SDI infrastructure.
O
OTT (Over-the-Top)
Video content delivered over the open internet, “over the top” of a telecom or cable operator’s managed network, without requiring a subscription to that operator’s television service. Netflix, Disney+, YouTube, and every other streaming service that works on any internet connection is OTT. The term was coined by traditional TV operators who viewed internet-delivered video as bypassing (“going over the top of”) their controlled distribution infrastructure. Today, OTT is essentially synonymous with “streaming” and has grown from a disruptive threat to the default mode of video consumption for most demographics. The irony is that the telecom operators who coined the term now operate their own OTT services.
P
Per-Title Encoding
An encoding optimisation approach (pioneered by Netflix) that analyses each piece of content individually and creates a custom ABR encoding ladder optimised for that specific content’s visual complexity. A talking-head interview can be encoded at much lower bitrates than an action sequence with fast motion, explosions, and particle effects. Per-title encoding (and its evolution, per-shot encoding) significantly reduces bandwidth and CDN costs while maintaining visual quality. The approach has been adopted across the industry and is sometimes marketed as “content-aware encoding” or “context-adaptive encoding.” It is one of the few streaming innovations that is genuinely a win-win-win: better quality for viewers, lower costs for operators, and reduced bandwidth consumption for networks.
Programmatic Advertising
The automated buying and selling of advertising inventory through technology platforms — using algorithms, real-time bidding, and data-driven targeting to match ads with audiences at scale. Programmatic has dominated digital display advertising for years and is now rapidly expanding into CTV, where it enables advertisers to buy streaming inventory with the same precision and efficiency they are accustomed to in web and mobile. The CTV programmatic supply chain involves DSPs (buyer platforms), SSPs (seller platforms), ad exchanges (auction marketplaces), and data management platforms (audience targeting data) — a complex ecosystem that introduces latency, takes fees at every step, and creates opportunities for fraud and inventory misrepresentation that the industry is still working to address.
Q
QoE (Quality of Experience)
A holistic measure of a viewer’s subjective experience with a streaming service — encompassing video quality, startup time, buffering frequency, audio-video sync, ad transition smoothness, and UI responsiveness. QoE is distinct from QoS (Quality of Service), which measures technical network metrics like bandwidth and packet loss. You can have perfect QoS and terrible QoE (if the player software is buggy, the content discovery is broken, or the ad experience is infuriating). QoE analytics platforms like Conviva, Mux, and NPAW collect billions of data points from video players to help operators understand and improve the viewing experience in real time. In a market where content is increasingly commoditised, QoE is becoming one of the few sustainable competitive differentiators.
R
RIST (Reliable Internet Stream Transport)
An open-standard protocol for reliable video transport over unmanaged networks (the public internet), competing with SRT for the contribution and distribution segment of the broadcast workflow. RIST uses ARQ (Automatic Repeat reQuest) error recovery to retransmit lost packets, enabling broadcast-quality video delivery over connections where packet loss would normally make professional use impossible. RIST’s key advantage over SRT is its standards-based approach (defined by the Video Services Forum under TR-06-1 and TR-06-2), which theoretically ensures better multi-vendor interoperability. In practice, both RIST and SRT are excellent at their job — reliable, low-latency video transport over imperfect networks — and the choice between them often comes down to existing vendor relationships and ecosystem support rather than fundamental technical differences.
S
SRT (Secure Reliable Transport)
An open-source video transport protocol originally developed by Haivision, designed to deliver low-latency, high-quality video over unpredictable networks. SRT uses UDP-based transport with AES encryption and ARQ error correction, making it possible to send broadcast-quality video contribution feeds over the public internet rather than expensive dedicated circuits or satellite links. SRT has been widely adopted for remote production, cloud-based workflows, and international video contribution — any use case where the alternative was a dedicated fibre or satellite link that cost ten to a hundred times more. The SRT Alliance has grown to over 600 member companies, and the protocol is supported natively in most professional broadcast equipment and cloud media services.
SSAI (Server-Side Ad Insertion)
An ad insertion method where ads are stitched into the content stream on the server before it reaches the viewer’s device — producing a single, seamless stream that looks and behaves like one continuous piece of content. The viewer’s player does not know where content ends and ads begin, which means ad blockers cannot distinguish ad segments from content segments. SSAI delivers a significantly better viewer experience than CSAI (no buffering between ads and content, no audio level jumps, no visible transitions) and is the preferred approach for CTV environments where the big-screen, lean-back experience demands television-quality ad breaks. The trade-off: SSAI makes client-side measurement more difficult (the player does not know it is playing an ad), requiring server-side beaconing and trusted-server attestation for viewability reporting.
SVOD (Subscription Video on Demand)
A business model where viewers pay a recurring subscription fee for access to a library of on-demand content — typically without advertisements (though the lines have blurred since Netflix, Disney+, and others introduced cheaper ad-supported tiers). Netflix, Amazon Prime Video, Disney+, HBO Max, and Apple TV+ are the defining SVOD platforms. The SVOD model dominated the first era of streaming growth but is now mature in most developed markets, with subscriber acquisition costs rising, churn rates stubbornly high, and average revenue per user constrained by price sensitivity. The industry-wide shift toward hybrid models (SVOD + ads, SVOD + FAST, SVOD + live sports) reflects the reality that pure subscription revenue alone cannot support the content spending required to compete.
SSP (Supply-Side Platform)
A technology platform that helps publishers and streaming services sell their ad inventory programmatically — the sell-side counterpart to the DSP. SSPs connect to multiple ad exchanges and DSPs, run auctions to maximise the price achieved for each impression, manage floor prices and deal priorities, and provide analytics on inventory performance. In CTV, major SSPs include SpotX (now Magnite), FreeWheel (Comcast), and SpringServe. The SSP’s role is to ensure that every available ad impression is monetised at the highest possible price — and the tension between yield maximisation (showing the highest-paying ad) and viewer experience (showing relevant, well-paced ads) is one of the fundamental design challenges in ad-supported streaming.
T
Transcoding
The process of converting video from one format, codec, resolution, or bitrate to another. Transcoding is the workhorse operation of every streaming platform — taking high-quality source content (mezzanine files) and producing the dozens of renditions (different resolutions, bitrates, and codecs) needed to serve every device, screen size, and network condition in your audience. A single piece of content might be transcoded into 30+ renditions. Transcoding can be done on dedicated hardware (faster, more predictable), in the cloud (more flexible, pay-per-use), or in hybrid architectures. For live streaming, transcoding must happen in real time, which adds latency and requires compute capacity that scales with concurrent viewer demand. Transcoding costs — whether measured in hardware, cloud compute, or energy consumption — are a significant operational expense for every streaming operation.
TVOD (Transactional Video on Demand)
A business model where viewers pay for individual pieces of content — renting or purchasing specific films or episodes rather than subscribing to an entire library. iTunes, Google Play Movies, Amazon’s rental/purchase options, and Vudu are TVOD platforms. TVOD generates higher revenue per transaction than SVOD but lower lifetime value per customer, and it works best for premium, time-sensitive content (new theatrical releases, event specials) where consumers are willing to pay a premium for immediate access. TVOD has been squeezed from both sides — by SVOD services that include new releases in their subscription price, and by shortened theatrical windows that make waiting for the SVOD release more palatable.
U
Unified ID 2.0 (UID2)
An open-source identity framework (originally developed by The Trade Desk, now governed by Prebid) designed to replace third-party cookies with a privacy-conscious, deterministic identifier based on hashed and encrypted email addresses. UID2 aims to solve the identity problem that plagues CTV advertising — enabling cross-platform frequency capping, attribution, and targeting without relying on device-level identifiers that are increasingly restricted by privacy regulations and platform policies. Adoption has been significant among publishers and ad-tech platforms, though the framework’s reliance on logged-in users (who have provided an email address) limits its reach to authenticated environments. UID2 is one of several competing post-cookie identity solutions, alongside LiveRamp’s RampID, ID5, and contextual targeting approaches that avoid user-level identification entirely.
V
VAST (Video Ad Serving Template)
An IAB (Interactive Advertising Bureau) standard that defines the communication protocol between an ad server and a video player. When a player encounters an ad break, it sends a VAST request to an ad server. The server responds with an XML document containing the ad creative URL, tracking pixel URLs for impressions and completion events, companion ad information, and instructions for how the ad should be displayed. VAST is now on version 4.x, which added support for server-side ad insertion verification, interactive ad formats, and improved measurement capabilities. VAST is not the most glamorous acronym in streaming, but it is the plumbing that makes video advertising work — every single ad impression on every CTV platform touches a VAST response at some point in the chain.
VMAF (Video Multimethod Assessment Fusion)
A perceptual video quality metric developed by Netflix in collaboration with the University of Southern California. VMAF uses machine learning to predict how a human viewer would rate the visual quality of a video, producing a score from 0 to 100 (where 100 is perceptually lossless). Unlike older metrics like PSNR and SSIM, VMAF correlates significantly better with subjective human quality judgments, making it the industry-standard metric for encoding optimisation, ABR ladder design, and quality monitoring. A VMAF score of 93+ is generally considered “excellent” for most content types. VMAF has become so central to the streaming quality conversation that encoding teams routinely talk about “VMAF targets” the way audio engineers talk about LUFS levels — it is the shared language of quality.
W
WebRTC (Web Real-Time Communication)
An open-source project and set of APIs that enables real-time audio, video, and data communication directly between web browsers and mobile applications — without plugins, without dedicated apps, and with sub-second latency. WebRTC was originally designed for video conferencing (it powers Google Meet, and underpins many other communication platforms) but has been increasingly adopted for ultra-low-latency streaming use cases: live sports betting (where even one second of delay creates arbitrage risk), interactive live events, real-time auctions, and any scenario where the viewer needs to see what is happening as close to “now” as technically possible. WebRTC achieves latency of 200–500 milliseconds — orders of magnitude better than HLS/DASH — but at the cost of scalability: WebRTC’s peer-to-peer architecture does not scale to millions of concurrent viewers without significant infrastructure (media servers, SFUs, CDN-like distribution layers) that adds complexity and cost.
40+
Terms Defined
21
Alphabetical Sections
2026
Updated For
A–W
Coverage Span
Frequently Asked Questions
CTV & Streaming — Your Questions Answered
What is the difference between SSAI and CSAI — and which should I use?
SSAI (server-side ad insertion) stitches ads into the content stream on the server, producing a single seamless stream that ad blockers cannot distinguish from content. CSAI (client-side ad insertion) has the player fetch and display ads separately from the content stream. For CTV and big-screen experiences, SSAI is the industry standard because it delivers a television-quality ad break without buffering, glitching, or visible transitions. CSAI is more common on web and mobile where client-side interaction tracking (clicks, viewability pixels, engagement events) is more important. Many platforms use a hybrid approach: SSAI for delivery with server-side beacons that replicate the measurement capabilities of CSAI. The short answer: if you are building for a lean-back CTV experience, use SSAI. If you need rich client-side interactivity and measurement, CSAI may still have a role.
What is the difference between AVOD, SVOD, TVOD, and FAST?
These are the four primary streaming business models. SVOD (Subscription Video on Demand) charges a monthly fee for access to a content library — Netflix and Disney+ are the archetypes. AVOD (Advertising-Based Video on Demand) offers free or discounted content supported by ads — Tubi is a pure AVOD platform, while Netflix’s ad tier is a hybrid SVOD/AVOD. TVOD (Transactional Video on Demand) charges per-title for rentals or purchases — iTunes and Amazon rentals. FAST (Free Ad-Supported Streaming Television) delivers linear-style scheduled channels for free with ads — Pluto TV and Samsung TV Plus. In practice, most major platforms now operate hybrid models combining two or more of these approaches, because no single model generates enough revenue per user to sustain the content investment required to compete. The trend is toward “something for everyone” — premium ad-free for subscribers who pay the most, ad-supported for price-sensitive viewers, FAST for lean-back browsing, and TVOD for premium early-access content.
HLS vs DASH — which protocol should I use?
The honest answer in 2026: both, and it matters less than it used to. HLS (Apple’s protocol) is required for Apple devices and is broadly supported everywhere else. DASH (the international standard) is dominant on Android, smart TVs, and web browsers. With CMAF enabling shared encoded segments between both protocols, you can encode once and serve both — the only difference is the manifest format (M3U8 for HLS, MPD for DASH). If you are building a new streaming service today, architect for CMAF-based delivery with dual HLS/DASH manifests generated from a single set of encoded segments. This gives you universal device reach with minimal storage overhead. The protocol wars are functionally over — the answer is “yes, both.”
Why is AV1 replacing HEVC — and should I switch now?
AV1 is not so much “replacing” HEVC as growing alongside it. HEVC is technically excellent but has a licensing structure so fragmented and expensive that it drove the entire industry — Google, Apple, Amazon, Netflix, Meta, Microsoft — to collaborate on a royalty-free alternative. AV1 offers comparable compression efficiency to HEVC with no licensing fees, which makes it significantly cheaper at scale. The catch: AV1 encoding is much more compute-intensive than HEVC, so your transcoding costs go up even as your CDN costs go down. Hardware AV1 decode is now widespread in devices shipped since 2022, so playback support is increasingly not an issue. The practical approach: use AV1 where hardware decode support is available and the compute cost of encoding is justified by CDN savings (typically high-traffic VOD content), and continue using HEVC for live streaming and legacy device support. Most platforms will run both codecs simultaneously for the foreseeable future.
What is the difference between SRT and RIST for live contribution?
Both SRT and RIST solve the same problem: reliable, low-latency video transport over unmanaged networks (the public internet). Both use UDP with ARQ error correction. Both encrypt the transport. Both are excellent at their job. The differences are more political than technical. SRT was developed by Haivision and open-sourced — it has a massive ecosystem (600+ alliance members) and wide vendor support, but its development is primarily driven by one company. RIST is an open standard developed by the Video Services Forum through a multi-vendor committee process, which theoretically ensures better interoperability but resulted in slower standardisation. In practice, most professional broadcast equipment supports both. Choose SRT if your vendor ecosystem is already SRT-aligned, choose RIST if you prioritise standards-body governance, and know that both will deliver broadcast-quality results over imperfect networks.
Why is CTV ad fraud such a persistent problem?
CTV ad fraud persists because the economics are irresistible to fraudsters and the detection mechanisms are immature compared to web and mobile. CTV CPMs are high ($20–$50+), which means each fraudulent impression is worth significantly more than a $2 display impression. Server-side ad insertion makes it harder for verification vendors to confirm that an ad actually played on a real device (because the player does not distinguish ads from content). The CTV device ecosystem is fragmented, with limited standardisation around device attestation — making it easier to spoof device identifiers and misrepresent inventory. SSAI watermarking, app-ads.txt for CTV, and Open Measurement SDK extensions are industry efforts to address these vulnerabilities, but the arms race between fraud schemes and detection technology continues. Buyers should demand transparency on supply paths (SPO — Supply Path Optimisation), verify inventory sources through ads.txt/app-ads.txt, and work with verification vendors that have specific CTV fraud detection capabilities.
T-21 is an independent publication covering streaming technology, cloud infrastructure, and enterprise digital systems. This glossary represents our editorial understanding of CTV and streaming terminology as of 2026. Definitions are written for practitioners and should not be construed as legal, regulatory, or vendor-specific guidance.
© 2026 T-21. All rights reserved.