The conventional wisdom on connected TV is that it belongs in your awareness budget, not your conversion stack. A Southwest dealer cohort of 11 to 20 rooftops just ran a week that challenges that assumption directly. With CTV taking 20% of the total spend, the second-largest allocation behind Google Search, the blended cost per lead for the week landed between $50 and $60. That is not a brand result. That is a direct-response result, delivered by a channel the industry still argues is too soft to measure.
The data covers one calendar week: September 11 through September 18, 2026. The cohort spent approximately $52K across eight sub-channels, generated roughly 2.7 million impressions and close to 69,000 clicks, and closed around 918 conversions. The average cost per click across the entire stack held at approximately $1. Eight channels. One budget. A CPL that most single-channel Google Search campaigns would consider competitive.
What Does a Properly Distributed Dealer Budget Actually Look Like?
Most dealer budgets cluster. Google Search takes 50 to 70 percent because that is where intent is highest and the attribution story is cleanest. Everything else fights over the remainder, and the remainder is usually an afterthought.
In this cohort, Google Search took 35% of the total allocation. That is significant not because 35% is low, but because six other channels each held meaningful, deliberate slices alongside it. CTV took 20%. Meta Leads took 10%. Google Demand Gen took 10%. Meta AIA took 10%. Microsoft Search, Google PMax, and Meta Traffic each held 5%.
A budget that looks like this was built from a single reasoning pass: given this market, this inventory mix, and this week, what is the optimal distribution across every available channel? Single-channel thinking produces the 70% Google stack. Cross-channel reasoning produces the distribution above.
The practical consequence is that no single channel is carrying the whole load. Google Search is the anchor, but it is not doing all the work. The remaining channels are real contributors, and the blended CPL reflects that the stack, as a whole, is performing.
Why Did CTV Take the Second-Largest Allocation?
CTV at 20% of a direct-response dealer budget is a deliberate decision, not an experiment. It means the cohort's budget engine made a call that streaming inventory earned that allocation over the alternatives.
The argument for CTV in a dealer stack is not about awareness. It is about reach at a price that paid search cannot match in a competitive automotive market. A qualified household that sees a model-specific video ad in their living room on Thursday night does not behave identically to a household that sees a search ad on Friday morning, but both are part of the same purchase decision. CTV does not replace the search click. It raises the probability that the search happens at all.
The broader point is attribution. The industry's resistance to CTV in direct-response budgets is partly structural: measure CTV only by last-touch conversions and it looks expensive. Measure the stack at the blended level, as this data does, and you see the actual cost per conversion the full channel mix produces together. This cohort's blended CPL fell in the $50 to $60 range with CTV holding 20% of the budget. That is the number that matters.
The channel also matters for the Southwest market specifically. Streaming penetration in the Sun Belt is high, and automotive purchase intent in the region has historically been strong enough that OEM incentive programs price Southwest markets separately from the rest of the country. A channel that reaches that audience at scale, with model-specific video creative, earns its budget share.
What Did Google Search Actually Do With 35% of the Budget?
At 35% of a ~$52K weekly spend, Google Search was responsible for roughly $18K of deployed capital. The blended cost per click across the entire stack held at approximately $1. In the current automotive search auction, where competitive brand and model terms regularly clear $8 to $15 per click, a $1 blended CPC is only possible if the mix of search traffic includes high-volume, lower-cost terms alongside the expensive head terms.
That is exactly what a well-structured search campaign does. The channel mix shifts when inventory conditions change, and the keyword strategy should shift with it. A campaign that runs broad match on high-cost brand terms with no structure is spending $10 a click on traffic that converts at the same rate as a $1 click from a long-tail model term. The blended CPC in this data suggests the search campaigns were built with that discipline in place.
The cohort generated approximately 2.7 million impressions across all channels. With Google Search at 35% of budget, a significant share of those impressions arrived at moments of active intent: someone typing a model name, a trim query, or a financing question into a search bar. Those impressions are expensive to buy but valuable in a way that upper-funnel inventory cannot replicate. The stack earns its efficiency precisely because it is buying different types of attention at different prices.
How Does a $1 Average CPC Survive in the Current Automotive Market?
It survives because the $1 figure is a blended average, not a paid search benchmark. When you run eight channels together and weight the clicks by channel volume, the search clicks at $8 to $12 blend with display, CTV, social, and Microsoft impressions where click costs are structurally lower.
Microsoft Search, at 5% of this cohort's budget, is the clearest example. Measurement differences across platforms mean that Microsoft's automotive CPCs typically run 20 to 40% below equivalent Google terms, with an audience skewing older, higher-income, and more likely to be in the final stages of a purchase decision. Five percent of this budget on Microsoft is not a rounding error. It is a deliberate allocation to a channel where the same dollar buys more qualified traffic.
Meta AIA at 10% similarly operates at a different cost structure from paid search. Dynamic catalog ads that pull from the dealer's live inventory and serve in-market households a specific vehicle at a specific price do not compete in the same auction as search. The cohort generated approximately 69,000 clicks at a blended $1 average, which means the stack was generating real traffic at a cost that search alone could not sustain.
The stack's efficiency is a portfolio effect. The expensive channels (Google Search head terms) are justified by the conversion quality they produce. The lower-cost channels (Meta AIA, Microsoft, Demand Gen) extend reach and volume. CTV provides the context that makes both work better. You cannot isolate one piece and ask whether it was worth it at its channel-specific CPA. The question is whether the portfolio, run together, produced the blended result. In this case it produced approximately 918 conversions at a CPL between $50 and $60.
What Does the Rest of the Channel Stack Contribute?
Google Demand Gen at 10% is running a different job from Search. Demand Gen campaigns serve across YouTube, Gmail, and Google's Discover surface, reaching people who are not actively searching but are the right audience by interest and behavior. In a vehicle purchase cycle that often stretches 60 to 90 days, touching the same household on display and video before they reach the search stage changes what they search for and who they search for.
Google PMax at 5% handles the machine-managed layer: inventory ads that appear across Shopping, Search, Display, and YouTube simultaneously. The risk with Performance Max is attribution opacity, not performance. PMax at 5% means it is in the stack but not dominating it. That is the right posture for a cohort that wants PMax's cross-surface reach without giving it enough budget to cannibalize the structured intent capture that Search does better.
Meta Traffic at 5% is the lightest allocation in the stack. Traffic campaigns on Meta serve clicks to the dealer's site rather than leads, and at 5% of budget they are retargeting work: reminding in-market households who visited a vehicle detail page that the unit is still available. First-party audience depth is what separates a useful retargeting campaign from a generic one. At this allocation level, the value is in the audience match, not the volume.
Meta Leads at 10% operates at the opposite end of the funnel. Instant Forms that capture contact information and vehicle interest directly in the Meta feed, without requiring the user to load a website, carry a different cost profile than a website-originated lead. The lead quality question is real, but so is the cost advantage, and for a cohort running a $50-$60 blended CPL, the Meta Leads component is doing its part.
How AUTONOMi Runs a Stack This Wide Without Fragmenting the Budget
Eight channels in a single week requires a single budget decision, not eight. The failure mode of multi-channel advertising is not running too many channels. It is running them independently, each with its own budget that was set once and never revisited against the others. When Google Search overdelivers in the first four days of a week and Meta Leads underdelivers, a siloed budget structure does nothing. The money that Google Search would have absorbed after hitting its efficiency ceiling sits idle, and the channels that could have used it stay capped.
AEGIS makes one daily allocation decision across every sub-channel it manages, including Google Search, PMax, Demand Gen, Meta's full portfolio (model-showcase traffic, vehicle catalog, and lead-capture campaigns), TikTok AIA, and Microsoft Search. That is the architecture described in §8 of the platform's capabilities: a single cross-channel reasoning pass, not piecemeal per-channel adjustments, run against the dealer's live inventory and the market conditions that day. When a model sells out, the ad groups for that model rebuild automatically. When an OEM offer changes, the affected campaigns update. The budget does not stay pointed at inventory that no longer exists.
CTV, as an add-on channel available through the Connected TV and Streaming Layer module, sits inside that same budget engine. It is not managed separately by a streaming vendor who sends a monthly report and a bill. It competes for budget allocation in the same reasoning pass as every other channel. That is why it held 20% of this cohort's budget with a result that justified the allocation. The system looked at the alternatives and decided streaming inventory earned it.
AXIOM, the governance layer that gates every AEGIS action, enforces what the budget engine cannot override: the geo constraint, the spend ceiling, the compliance review on every piece of creative before it serves. The eight-channel stack in this cohort did not produce an uncontrolled result. It produced a governed result, where every dollar was allocated by reasoning and every ad was reviewed before it ran.
Who the Next Dealer to Run This Stack Is
The Southwest cohort is not a special case. It is what multi-channel dealer advertising looks like when the channels are orchestrated from one budget engine rather than siloed by vendor. Any franchise dealer with a CTV module active and an inventory mix that supports eight channels can run the same structure. The question is not whether the channels exist. They do. The question is whether your budget engine is making one decision or eight.
The dealers who will own the Southwest market in the next purchase cycle are not the ones with the highest Google Search budget. They are the ones with the tightest channel integration: search capturing the intent that CTV raised, social retargeting the households that search missed, Microsoft catching the buyer who is three days from a decision. That stack produced a $50-$60 CPL across 11 to 20 rooftops in one week. The next cohort to run it does not have to guess what the result looks like. If you want to see how this structure performs in your market, start a 30-day pilot with AUTONOMi and let the budget engine build the distribution from your actual inventory and your actual market.
