Connected TV advertising.

Connected TV advertising is television bought the way the rest of digital is bought: each impression is placed programmatically, on a streaming app running on a TV screen, against a household you can describe and cap. It goes by more than one name. Streaming TV advertising, OTT, and CTV all point at the same shift, which is that the television stopped being a broadcast endpoint and became an addressable device.

The buying is the easy half. Any DSP can put a spot into a streaming break. What separates a disciplined CTV practice is everything around the spot: which app and which publisher actually served it, whether the path from budget to that publisher had two hops or seven, whether the household had already seen the creative on another device, and whether any of it can be shown in the log files afterwards. That is the standard this page describes, and it is the reason proving where an ad ran is harder than running it.

Four supply families make up most of the streaming market, and they behave differently enough that a plan should name which ones it is buying. Ad-supported streaming services are the premium tier: the ad-funded plans of the major subscription apps, where full-episode television meets a controlled ad load. FAST channels, free ad-supported streaming TV, are linear-style channels inside the big aggregation apps; they carry lighter content costs, heavier ad loads, and prices to match. Live sports moved to streaming with tune-in attached, which brought appointment audiences and game-day dayparting back to an addressable screen. Broadcaster and network apps carry the same content as their linear feeds with digital ad decisioning behind the break.

The same spot lands very differently across those four, which is why we treat supply selection as a planning decision, not an optimization detail. Named publishers are the default path, the app-level placement shows up in reporting, and when a price difference matters we can walk you back to the publisher and the path that produced it.

A television is not a person. Treating it like one is how frequency gets wasted, so CTV targeting starts from household identity: the graph that ties the living-room screen to the phones, laptops, and tablets behind the same door. Audiences built from a client’s first-party data, from modeled cohorts, or from contextual signals resolve to households, and frequency is capped there rather than per app or per device. The fifth impression on the same family is counted as the fifth.

ACR data, the content recognition running on smart TVs, adds what the household actually watches, including linear, which is what makes incremental-reach planning against pay TV possible. Geography works down to the DMA and the zip, so regional brands and multi-market footprints buy the same premium screens with local control. And because the biggest screen in the house is shared, co-viewing is modeled in: reach is adjusted by household composition instead of pretending one impression met one viewer.

CTV measurement has a reputation for black boxes, and the fix is unglamorous: log-level data as the standard, so every impression carries its app, publisher, time, and price into a file we can hand over. Completion and audibility are read from the same logs. ACR panels de-duplicate streaming exposure against linear campaigns, so a household reached in both places is counted once, and incremental reach over the linear buy is a number rather than an assertion. Content-level visibility caught up in July 2026, when show-level reporting reached programmatic buyers across seven major streaming publishers; our show-level reporting analysis covers which supply paths carry the signal and the setup work involved.

Outcomes follow the same discipline as every other channel we run: screening is built to catch invalid traffic before the bid, brand studies read lift where the spend data cannot, and the cross-channel effect of TV rolls up through marketing mix modeling alongside everything else on the plan. The wider context, including where streaming’s share of TV time now sits, is collected in our programmatic statistics reference.

A CTV engagement starts with the outcome, not an IO template. The brief names what television is supposed to do, and the supply list follows from it: which of the four inventory families, which named publishers, which dayparts. Screening and suitability rules are set before anything bids. In flight, pacing and frequency read at the household level, and the optimization notes say what changed and why. At the end, reporting reconciles to the logs, reads against the outcome the brief named, and runs under your agency’s name throughout, which is the agency-enabled model in practice.