CTV advertising for car dealers: what streaming buys, what it costs, and how to read the report

Streaming took 49% of all television viewing in July 2026, broadcast 19.5%, cable 18.7%, and YouTube on its own reached 14.2% of TV. The IAB forecasts connected TV ad spend up 15.6% this year while linear television falls 1.5%. The average dealership still records one budget line called TV, $61,556 and 10.5% of spend, with no split between the two. CTV advertising for car dealers is the question of what happens when that line is divided.

VulcanAX does not sell connected TV, streaming or any paid media. It handles what happens after the screen, which is where this piece ends. First the numbers, from the sources, then what the platforms admit it costs, then which figures on the report are real.

CTV advertising for car dealers in current numbers

What connected TV is, in a dealership's terms

Connected TV is the television spot bought like digital media. The same 15- or 30-second creative that ran on local broadcast runs inside ad-supported streaming services and on YouTube on the television set, bought by the thousand impressions, aimed at households in the store's market by geography and audience data, and reported with digital measurement rather than a ratings estimate. OTT is the broader term for video delivered over the internet to any screen, including a phone; CTV is the subset that lands on the living-room television, which is the screen the broadcast budget was always trying to reach.

The audience shift is not in dispute. Nielsen's July 2026 Gauge has streaming at 49% of all TV viewing against 19.5% broadcast and 18.7% cable, and YouTube alone at 14.2%, ahead of every streaming service. The IAB's September forecast has connected TV spend up 15.6% and linear TV down 1.5%. And the automakers who used to buy the national air cover under a dealer's local spots are cutting it: iSpot puts category spending on national linear TV down 16.7% in the first half of 2026. The dealer budget, meanwhile, still reports one line called TV. Dividing it is the decision this piece is about.

What it costs, as far as anyone will say in public

Streaming platforms and demand-side platforms do not publish rate cards for dealers, and this page will not invent one. Two published points bracket the market. Paramount's self-serve Ads Manager states that most entry-level CTV campaigns start between $25 and $65 CPM, that competitive 2026 CPMs run $20 to $40 for standard inventory and $40 to $60 for premium targeted placements, and that its own campaigns start at $7 CPM, with a completion rate around 96% on Paramount+. Those are one platform's statements about itself and its competitors, and they are quoted as such.

The arithmetic a store can do from them: at a $30 CPM, $5,000 a month buys about 167,000 impressions. At a frequency of four per household, that is roughly 42,000 households, which is a meaningful share of one metro market and a rounding error across a region. The budget benchmark shows where a store's TV line sits against NADA's $61,556; a dealership splitting that line should expect the streaming half to buy fewer, better-targeted households than the broadcast half bought impressions, and to be judged on a different report.

How to read a CTV report: three layers

Every connected TV report has three kinds of number in it, and they are not equally real.

What the report counts. Impressions served, households reached, frequency per household, and video completion rate. These are measured. Completion runs above 90% on most streaming inventory because the ad cannot be skipped; it is a property of the format, not a result of the campaign, and a store that is shown a 96% completion rate as evidence of performance has been shown the shape of the inventory.

What the report infers. Website visits and leads matched to exposed households through IP addresses and device graphs, and lot visits matched through the same mobile location data that geofencing runs on. Simpli.fi's description of its addressable product covers CTV alongside mobile and video: devices inside a fence, followed for up to 30 days, measured through conversion zones and a geo conversion lift metric. These are matches. A household that saw the spot and later visited the site is a correlation, and the match says nothing about why.

What the report cannot show. Whether those households would have visited anyway. Cox Automotive's 2025 study has buyers spending 2 hours 55 minutes visiting other dealers and 41% using a search engine; the households a CTV campaign reaches in a car-buying market were already going to search and already going to visit lots. Only an unexposed holdout, a random share of the same households withheld from the campaign, separates the visits the ad produced from the visits that were coming. The geofencing piece covers the same three questions for the phone; the paid search piece covers the brand-term version of the same test. It is one problem in three inventories.

The CTV KPIs worth holding

Households reached in the market, rather than impressions, because impressions divide by frequency and the vendor sets the frequency. Frequency per household, because the same spot eight times to one house is the most common way a streaming budget disappears. Cost per reached household, which makes the streaming buy comparable to the broadcast buy and to the search line. Completion rate, recorded and then ignored. And incremental visits or leads against a holdout, the only figure that answers the budget question, and the one to insist on before renewal.

How a dealership starts, in rising order of commitment

YouTube on television screens. The largest single streaming service at 14.2% of TV viewing, and the cheapest to test: bought in the same Google Ads account that runs the store's search campaigns, with TV-screen placement, household geography and no minimum. The trade-off is the environment, since the spot may run before a video the store did not choose. For a first test of whether streaming moves anything, it is the lowest-risk path.

A platform's self-serve tool. Paramount's Ads Manager and its equivalents sell one service's inventory at published entry pricing with a modest minimum. One service, one audience, a clean report.

A demand-side platform through an agency. Reach across services with audience data and cross-device matching, at higher minimums and with the reseller's margin on top of the platform's CPM. This is where most dealer CTV is bought, and where the six questions from the geofencing piece apply unchanged: which platform, what CPM by format, what minimum, where the data comes from, how wide the conversion zone, and whether they will run a holdout.

Before any of the three, ask the manufacturer about co-op. OEM co-op programs have covered television for decades and their rules on streaming vary by brand; a dealership that can run the same spot on connected TV with co-op reimbursement has a different arithmetic from one that cannot.

What happens after the screen

A streaming ad reaches a household on the couch. What the household does next is the part of the process the ad cannot control and the report cannot see clearly: 41% of buyers use a search engine, 75% use third-party sites, and 12% overall and 17% of new-vehicle buyers ask an AI site or read an AI overview. The spot creates or reinforces a name; the search decides whether the store attached to that name appears.

That is the connected TV version of the point every piece in this series ends on. The channel rents a moment of attention. Whether the store is there when the attention turns into a question is decided by what the store has published, and that is owned rather than rented. The trends piece places CTV among the other eight shifts of the year; this one is about the line item.

What VulcanAX does and does not touch here

VulcanAX does not sell connected TV, streaming, programmatic or any paid media, and will say so rather than sell it badly. Nothing here argues for or against splitting the TV line; a holdout result is the only thing that settles that, and the six questions are how to get one.

What VulcanAX handles is the step after the screen: whether the store is named when the viewer picks up a phone and asks a search engine or an AI assistant which dealer to visit. That is the dealership search work, and it is the only part of this subject VulcanAX sells.

FAQ