CTV Advertising Statistics

Nielsen's monthly Gauge report is the reference point for US television time because it measures the same panel across broadcast, cable and streaming. Its milestone reading — streaming taking a larger share of total television use than broadcast and cable put together — matters less as a headline than as a structural fact: the majority of incremental attention is now on screens that are addressable.

Two caveats travel with the figure. It measures share of television use, not households, and it includes free ad-supported channels and YouTube on the TV set, which many buyers plan separately from premium streaming. Read the platform breakdown rather than the headline before you split a budget on it.

Buyable CTV impressions come from four different places, and they behave differently on price, targeting and transparency. Publisher direct-sold inventory on a premium service is the closest to old television. Programmatic inventory through private marketplaces sits in the middle. Free ad-supported channels supply volume cheaply. And the television operating system itself sells the home screen, which is the only placement nobody else can offer.

The expansion of ad-supported tiers on subscription services added large volumes of premium inventory in a short period, which is the main reason CPMs in the category softened while total spending kept rising. A falling CPM alongside rising spend is not a contradiction; it is supply arriving faster than demand.

The one thing television did well — deduplicated reach and frequency across a whole campaign — is the thing CTV still does worst. Each platform can report the frequency it delivered on its own inventory, but nobody sees the full picture without a common identifier or an independent panel, so the household that saw the same commercial eleven times is a routine outcome rather than an edge case.

In practice buyers narrow the problem instead of solving it: fewer platforms per flight, frequency caps set deliberately low, and post-campaign reconciliation against a single measurement partner. Any statistic quoting campaign-level reach across many platforms deserves a question about how duplication was handled.

Most disagreements between two CTV figures come from scope rather than accuracy. Ad spend can include or exclude YouTube. Reach can be devices, households or people. Viewing can be minutes or sessions. And a platform's own installed base counts sets shipped, not sets switched on this month.

Before quoting a number in a plan, write down three things next to it: who measured it, what population it covers, and what period it describes. If any of the three is missing from the source, use the figure to describe a direction and not to size a commitment.