Dealership Marketing

When Do Social and CTV Earn a Place in a Dealership’s Budget?

Social and connected TV earn budget when each has a job it can be held to and a test that shows whether sold units moved. How a dealer group sets the job, cleans the feed, and runs the holdout.

B
Brian Fidler
November 20, 2026·9 min read

In Nielsen’s August Gauge, streaming took 49.6% of US TV watch-time. Cable held 18.9%. Broadcast held 17.5%. YouTube alone accounted for 15.8%, close to broadcast’s entire share. One month of data, not a trend line, but a clear snapshot: most of the time Americans spend in front of a TV now goes somewhere a broadcast buy doesn’t reach.

So the money follows the audience to social and connected TV (CTV), the ads delivered through streaming apps on the living-room screen. Social and CTV earn budget when each has a specific job it can be held to, and a test that can show whether sold units moved. Not clicks. Not the platform’s own conversion report. A job and a test.

For a group with three to 15 rooftops, ad budget is gross you already earned, spent to earn more. A broadcast schedule that nobody can tie to units sold is a cost without evidence. A CTV schedule that nobody can tie to units sold is the same cost with better targeting slides. Moving the money does not fix the problem if the problem was never the channel.

I learned this moving money, not reading about it. I spent three and a half years running marketing for La Mesa RV, an 11-store RV dealer group. We moved $500K a month out of broadcast and into digital. Not because broadcast failed. We could never tell whether it failed. Digital could be measured, and broadcast showed no evidence it worked. Absence of evidence was the whole case. I have never worked at an auto dealership, and the method is what transfers: give a channel a job, then test whether it did the job.

Where did the broadcast audience go?

To streaming, and the Nielsen snapshot earlier puts numbers on it. Broadcast at 17.5% of watch-time means a local news buy now reaches a minority of the time people spend in front of a TV. Households did not stop watching television. More of it now arrives through streaming apps than through an antenna or a cable box.

CTV is the industry’s answer: the same living-room screen, bought like digital. Many CTV platforms sell impressions by household, with frequency caps and geographic lines you draw yourself. On paper, television with the targeting of a search campaign.

Nielsen is adjusting its methodology and will publish September and October together, so resist anyone who turns a single Gauge into a slope. The share split alone is enough to force the question: if most viewing happens outside broadcast, why keep buying broadcast weight the old way?

Why didn’t moving the money fix the measurement?

Because broadcast’s measurement problem moved to streaming with the audience. A CTV impression is a view without a click. Nobody taps a sedan on their living-room screen and drives it home. The chain from impression to showroom visit to sold unit runs through weeks of shopping, a spouse, a trade appraisal, and a Saturday. The platforms bridge that gap with modeled attribution, and modeled attribution means the platform scores its own work.

Social has the same structure. The in-platform conversion report counts events the platform defines, matched by methods the platform controls, scored by the party that gets paid when the number looks good. None of this makes the reports fraudulent. It makes them testimony from an interested witness.

Google calls incrementality experiments “the way to measure the causal impact of ads,” and its Conversion Lift product measures the increase in conversions “that are caused by the presence of the ad.” In my view, that’s the question standard conversion reports don’t answer. Google also notes the product “isn’t available for all Google Ads accounts.” The reliable measurement is the gated one. The flattering measurement is the default on every dashboard your vendor screenshots into the monthly recap.

When we moved $500K a month at La Mesa RV, we moved it toward channels where the measurement was possible, not toward channels where the dashboard was prettier. The distinction decides everything downstream. A general manager (GM) who cannot tell the difference between a measured channel and a self-reported channel will keep funding whichever vendor presents last.

What job can social do for a dealership?

Put specific inventory in front of in-market shoppers. Not brand films. Not community-page engagement. A 2022 F-150 with the trim, the price, and the store, served to a household showing shopping signals, clicking through to that exact vehicle detail page.

Size the job realistically. Cox Automotive’s 2025 Car Buyer Journey study, covering 2,344 recent buyers, found 26% visited social media while shopping, against 75% who visited third-party sites. Social is a real surface in the journey, and a minority one. Fund it like a minority surface with a specific job, not like the main road to the showroom.

The machinery exists for this job. Meta’s automotive inventory ads “use cross-device intent signals to automatically promote relevant vehicles from your inventory.” The operative word is inventory. The ads are only as good as the feed behind them. A feed with stale units, missing photos, placeholder pricing, or vehicles that sold last Tuesday will promote cars you cannot deliver, and a shopper who clicks through to a dead listing has little reason to come back. Before a dollar goes to inventory ads, someone has to own the feed: every unit current, priced, photographed, mapped to a live detail page.

Generating form fills for the business development center (BDC) to call is not the job. Form fills are a byproduct. The job is getting the right metal in front of the right household, and the test is whether units in that market moved.

What job can CTV do for a dealership?

Reach and frequency inside your primary market area (PMA), the geography your store is responsible for and pulls most of its buyers from. The job broadcast used to do: keep the store’s name, offer, and location in front of households before they are in-market, so your store is on the short list when the lease ends or the transmission goes.

CTV can draw that geography tighter than a broadcast signal ever could. Platforms that sell by geography let you buy the ZIP codes in your PMA instead of the whole designated market area, so a five-rooftop group stops paying to reach households three stores over, or two hours past the last exit it sells to. Frequency caps limit how often one household sees the ad, instead of bombarding whoever happens to watch the most TV.

What CTV cannot do is prove itself through a click, because there is no click. Hold it to the same standard broadcast failed: not “did the campaign deliver impressions” but “did the markets that saw it sell more units than the markets that didn’t.”

How do you hold either channel to account?

With a lift test or a holdout. Both answer the question that matters, whether sales moved because of the ads, by comparing a market that got the ads to a matched market that did not. Google’s Conversion Lift applies the same logic inside its own system, and where your account qualifies, use it. For CTV and for anything Lift does not cover, build the holdout yourself.

Here is a hypothetical, labeled as one. Say your group runs six rooftops. Pick two stores with steady month-over-month units and similar PMAs. Run CTV in those two markets at real weight for 90 days. Hold out two matched stores, with similar unit volume, inventory mix, and co-op calendar, and run nothing new there. At the end, compare sold units in the test markets against the holdouts and against each market’s own prior-period baseline. If the test stores outsold their baseline and the holdouts did not, the channel earned the next quarter. If the four stores moved together, the CTV spend bought impressions and nothing else, and you found out for the price of one test instead of one year.

The design matters more than the math. Keep everything else still during the window: no new incentive stacked on the test stores, no price repositioning in one market and not the other. A dirty test is worse than no test, because it produces a confident wrong answer.

And define the conversion as sold units in the market, pulled from your dealer management system (DMS), not leads, not clicks, not the platform’s modeled conversions. Units and gross are the books you already trust. The test should settle into those books, not into a vendor’s dashboard.

When should neither channel get a dollar?

Two conditions, and either one is disqualifying.

First, when you cannot tie sold units to a market. If your DMS and your reporting cannot tell you how many units each rooftop sold into which ZIP codes, month over month, no lift test is possible, because the outcome side of the test does not exist. Fix the reporting first. Every ad dollar spent before then is spent on faith, whichever channel takes it.

Second, when the feed is broken. For social inventory ads, a feed with wrong prices, sold units still live, or photo-less listings will burn budget proving the wrong thing. The campaign fails and you cannot tell whether the channel failed or the feed did. Repair the feed, confirm every unit resolves to a live, accurate detail page, then test.

Neither fix costs media dollars. Both are less expensive than a quarter of spend that produces an argument instead of an answer. The discipline runs one direction: measurement readiness first, budget second. A channel that cannot be tested is not a bet. It’s a habit.

Who owns the test?

The work in this post is not a media plan. It is an operating discipline: name the job each channel is hired for, confirm the DMS can report units by market, audit the inventory feed until it tells the truth, design the holdout, keep the test clean for 90 days, and read the result in units and gross rather than in a platform’s self-graded report. A dealer principal can run every piece of it in-house. What a fractional chief marketing officer (CMO) owns is making sure it happens in that order, holding the vendors to the test they would rather not take, and cutting the spend the test cannot defend, before the next buying season, not after it.

Frequently Asked Questions

Is broadcast dead for dealer groups?

No, and Nielsen’s numbers don’t say so. Broadcast still held 17.5% of US TV watch-time in August 2026, and one month is not a trend. The case against a broadcast schedule is its measurement, not its audience size: if it can’t show evidence it moved units, it competes for budget against channels that can.

Why not trust the conversion numbers Meta and Google report?

Because the platform selling the ads also scores them, using matching and modeling it controls. The reports are useful directional data, not proof of cause. Google’s own Conversion Lift measures the increase in conversions caused by the presence of the ad. In my view, standard conversion reports don’t answer that question.

How long should a CTV or social test run?

Long enough to produce a readable difference in sold units. Ninety days is a reasonable starting window for a holdout design, because shorter tests drown in the normal month-to-month noise of unit sales. Keep incentives, pricing and other media constant across test and holdout stores for the whole window.

What has to be true before spending anything on social inventory ads?

A clean feed and market-level sales reporting. Meta’s automotive inventory ads promote vehicles from your inventory, so stale units, wrong prices or dead detail pages waste the spend before targeting ever matters. And without sold units tied to each market in your dealer management system, you can’t test whether the ads worked.

Where does your dealership’s marketing stand on AI?

Answer 19 multiple-choice questions about your team, data and tools, and see your score on screen as soon as you finish.

Get my score

About 5 minutes · Instant result

Rather talk it through? Book a strategy call

Related Articles