There is a version of this that sounds like a targeting problem. It is not. A shopper searches "certified 2023 RAV4 XLE near me," Google matches that query to a specific VIN on a specific lot, and the listing that appears — photo, price, mileage, dealer name — either fires or it doesn't.
Vehicle Listing Ads work differently from standard text ads by design: instead of matching a shopper's query to keywords a dealer chose in advance, Google's automated targeting matches the query directly against live inventory data in the Merchant Center feed — surfacing the specific vehicle, price, and photo for that shopper's search rather than a generic ad. Google describes this as helping 'get in front of auto shoppers online at the right moment, with the right listings and information to move them closer to a purchase,' with more qualified leads as a stated benefit of showing vehicle details before the click. That structural difference — inventory-to-intent matching with no keyword layer in between — is what distinguishes the format from text ads, rather than a fixed performance ranking between the two.
That advantage is real. Almost nobody is fully capturing it.Google Autocomplete is currently showing "vehicle listing ads google" and "vehicle listing ads vlas" trending at the same time — a dealer and marketer audience actively trying to understand a format that has existed for years. That gap between search volume and search sophistication is the tell. Awareness has outrun execution. Dealers know VLAs exist. Most have no idea why theirs underperform.
What Are Vehicle Listing Ads and How Do They Work on Google?
Vehicle Listing Ads run on a structured data feed, not a keyword list.
Google Vehicle Ads run on a feed-driven mechanism: dealers submit a structured vehicle inventory feed through Google Merchant Center, and Google's own feed specification confirms the core attributes it ingests — VIN, price, condition (new/used), make, model, trim, and year, with mileage required for used vehicles. When a shopper's search query matches inventory in that feed, Google assembles the ad listing directly from those attributes rather than from a static, pre-built creative — which is why feed data quality (accurate pricing, current condition, complete trim detail) directly determines whether a vehicle is eligible to show and how the listing renders.
There is no ad copy to write. There is no headline to test. The feed is the creative and the targeting logic at once.This is a fundamentally different mechanic than a Search campaign, where a human (or AEGIS) writes responsive search ad copy against keyword themes. VLAs remove that layer entirely. The output a shopper sees is a direct function of what's in the feed at the moment they search. If the feed is accurate and current, the ad works. If it isn't, no bid adjustment fixes it — you're optimizing the wrong variable. Feed ownership is a separate, serious problem from the one this piece is about: even a dealer who fully controls their Merchant Center account can still be running VLAs wrong at the structural level, which is the failure mode most of the VLA conversation skips.
Why Are Most Dealers' VLA Campaigns Underperforming?
Three failure patterns show up again and again in dealer accounts, and they compound rather than existing independently.
The first is absence. A meaningful share of franchise dealers simply aren't running VLAs — the campaign was never built, or it was built once during a website migration and never revisited. Google's own vehicle ads policies state that incorrect or inaccurate data in a vehicle feed can trigger product disapprovals or Limited status, either of which prevents the affected listings from showing at all.✓ Jul 11 A dealer who set up VLAs once and never checked back doesn't see an error message in a place they're looking — they see a campaign that's technically live and functionally empty.
The second is a stale or malformed feed. This is closely related to the disapproval mechanic above but shows up even without a hard policy flag: a feed updated once a day, or once every few days, is representing a lot that no longer exists. A vehicle sells Tuesday morning; the feed reflects it as available through Wednesday's refresh cycle. Every impression served against that VIN in the interim is a wasted auction and, worse, a shopper sent to a dead vehicle detail page.
The third — and the one most dealer conversations skip entirely — is structural. VLA and Performance Max spend get lumped into a single black box with no visibility into which unit is actually pulling. That's the failure this piece is built around.
Why Do Dealers Keep Losing Visibility Into Which Unit Is Actually Performing?
Vehicle Listing Ads can serve through a standalone Shopping-style campaign or through Performance Max, which activates the Shopping surface alongside Search, Display, YouTube, and Discover in a single opaque bidding pool. Most agencies default to the second option because it's less setup work — one campaign instead of several. The dealer gets a blended conversion number and a blended cost-per-lead. What they don't get is an answer to the only question that matters: did that lead come from a VIN-matched VLA listing, or from a Display impression that happened to get attributed to the same campaign?
Performance Max's reporting model doesn't separate spend by surface, which means a dealer running VLAs inside PMax has handed away the one diagnostic that would tell them whether their feed problem is a feed problem or a bidding problem. When performance is soft, the agency's answer is usually a bid adjustment. The actual issue might be that 30 VINs disappeared from the feed last week and the campaign is quietly bidding against a smaller lot than the dealer thinks they have. Nobody can tell the difference from a blended dashboard.
Why Does Separating New From Used Inventory Change VLA Performance?
New and used vehicles are different products with different buyer psychology, different margin structures, and — critically for VLA performance — different feed volatility. Used inventory turns over faster, carries wider price variance vehicle-to-vehicle, and is far more likely to have incomplete data (a used unit still in reconditioning often has no photo yet, which is a required feed attribute). New inventory is more stable but subject to OEM incentive changes that shift the price and offer terms shown in the feed on a different cadence entirely.
A single campaign structure serving both conditions inherits the volatility of the worse-behaved half of the feed. When used-vehicle attribute gaps trigger disapprovals, a co-mingled campaign doesn't just lose the affected used VINs — it can drag down the account-level feed health signal that affects how aggressively Google serves the new-vehicle listings sitting in the same campaign. Separating new and used into distinct campaign topologies isolates that risk: a reconditioning backlog on the used side stays contained to the used campaign, and the new-vehicle campaign — usually the higher-margin, OEM-incentive-driven side of the lot — keeps serving cleanly regardless of what's happening upstream on trade-ins.
This is also where the platform's own automation actually gets a chance to work. Google's Vehicle Ads formats are built to let the platform's bidding and serving automation optimize within a defined inventory set — but that automation optimizes within whatever boundary the campaign structure gives it. Feed a single campaign both conditions and the automation is optimizing across two products with different economics as if they were one. Separate them, and the automation is finally solving the problem it was built to solve.
Is Feed Quality Really the Root Cause, Not Bid Strategy?
Every diagnostic path in this article leads back to the same place. A dealer with a clean, current, correctly structured feed and a modest bid strategy will outperform a dealer with an aggressive bid strategy sitting on top of a feed with gaps, staleness, or co-mingled conditions. That ordering is not intuitive to anyone trained to think about paid search as a bidding discipline, because for a decade that's what it was. VLAs invert the model. There's no bid lever sophisticated enough to compensate for a feed that's advertising yesterday's lot.
This is the same structural pattern showing up across how automotive inventory data actually gets sourced for advertising generally — the freshest, most authoritative record of what's actually on the lot is rarely the system marketing vendors default to building against. VLAs just make the consequence of that gap visible faster and more expensively than any other ad format, because the feed isn't supporting information next to the ad. The feed is the ad.
How AUTONOMi Solves This
AEGIS manages vehicle feed creation, updates, and diagnostics directly against Google Merchant Center✓ Jul 11, built on top of AUTONOMi's own inventory capture layer rather than a static export handed off once at setup. AEGIS re-scrapes each dealer's live inventory on a sub-daily cadence, diffs it VIN-by-VIN against what was previously known — arrivals, sales, price moves — and rebuilds only the affected campaign structures rather than the whole account✓ Jul 11, which is the mechanism that keeps a feed from drifting into the staleness problem described above. A VIN that sells doesn't wait for a batch cycle to disappear from what's being advertised.
On the structural side, AEGIS evaluates the dealer's spend split across new, used, and certified pre-owned inventory on every inventory refresh, weighing current stock mix, OEM incentive strength, and feed-age pressure as a reasoned judgment rather than a fixed formula✓ Jul 11 — the same distinction this article argues matters for campaign topology. That evaluation runs alongside AEGIS's broader daily budget allocation across the full Google surface set (Search, Performance Max, Demand Gen) plus Meta, TikTok, and Microsoft, made as a single reasoning pass rather than piecemeal per-channel decisions, which is what closes the visibility gap a co-mingled PMax/VLA structure creates: the dealer can see what each surface and each condition segment is actually doing, not a blended total.
The Merchant Center account, like every ad account and analytics property AEGIS operates, remains dealer-owned, with AEGIS holding delegated access the dealer can revoke at any time✓ Jul 11 — feed diagnostics aren't a black box sitting inside a platform the dealer can't inspect.
Where Does This Leave a Dealer Deciding What to Fix First?
The fix order matters. A dealer who jumps straight to campaign restructuring without addressing feed completeness is rebuilding a clean topology on top of dirty data — the new/used split will still inherit whatever gaps exist in either segment's underlying feed. The fix order is feed health first, structural separation second, bid strategy a distant third. Most agencies run that order backward, because bid strategy is the easiest thing to bill hours against and the hardest thing for a dealer to independently verify.
The dealers who get this right over the next year won't be the ones who found a better bidding algorithm. They'll be the ones who stopped treating VLA performance as a campaign problem and started treating it as what it actually is — an inventory data problem with an advertising symptom. If you want to see what your own feed health and campaign structure actually look like before deciding what to fix, you can connect your inventory feed through AUTONOMi and see the diagnostic AEGIS runs against it.
