Car dealer PPC: what paid search costs a store, what it buys, and what it cannot move

Search ads are the largest single line in the average dealership's advertising budget: $123,698 a store in 2025, 21.1% of the total, ahead of the listing sites and ahead of the website itself. Car dealer PPC is also the line most stores can describe least well, because the report arrives from the vendor who spends it.

VulcanAX does not run PPC. It handles the organic result and the page the paid click lands on, and it has no stake in whether a store's search budget goes up or down. That is the reason to read this rather than a reason to discount it: what follows is the arithmetic, from the published sources, with the parts that are usually left out of the monthly deck put back in.

Car dealer PPC in current numbers

What a dealership is buying, campaign by campaign

PPC for car dealers is five purchases wearing one invoice, and they behave differently enough that a blended report is close to useless.

Brand terms. The store's own name and its variants. Cheapest clicks, highest conversion, lowest cost per lead, and the campaign every agency leads the deck with. It intercepts buyers who had already decided to visit the store, which is why the next section exists.

Non-brand inventory terms. A model or body style plus a city or a "near me." This is the demand the store is actually renting: a buyer who has not chosen a dealer yet. It is where the click prices in the benchmark table are set and where the conversion rate is decided by the landing page.

Conquest terms. A competitor's name or a rival model. Expensive, low quality score, low conversion, and legitimate in the specific case where a store wants to be the second option a decided buyer sees.

Fixed ops terms. Service, repair, tires, oil change. The clicks cost twice what sales clicks do ($5.31 against $2.41) and convert at half again the rate (11.94% against 8.13%), because the searcher has a car that is broken today. The fixed ops marketing post covers why most stores under-buy this pool.

Google Vehicle Ads. The newest type and the one that changes the dependency. A vehicle ad shows a specific unit with its photo and price, drawn from a Merchant Center feed of the store's inventory, and sends the click to that unit's own page. It is eligible to dealers, retailers, aggregators and OEMs with a physical location in the listing state, and not to private sellers or brokers. Google's policy requires the price in the feed and in the page's structured data to match the price shown on the landing page exactly. A store with a late feed or a VDP whose price is rendered by a script it does not control is running a campaign against its own disapprovals.

Brand terms: the campaign that measures best and buys least

The largest controlled experiment on brand-keyword advertising was run at eBay and published in Econometrica in 2015. When eBay switched off its brand-term ads in test markets, the searchers who had typed "ebay" clicked the organic result instead. The ads showed no measurable short-term benefit. On non-brand terms, the study found new and infrequent users were influenced by the ads while frequent users, who accounted for most of the spend, would have bought anyway, so the average return was negative.

A dealership is not eBay, and there are two honest reasons to keep a brand campaign. A competitor bidding on the store's name has to be answered, and a Toyota store in a metro with four other Toyota stores is not the only result for its own brand term. But the study explains why the brand campaign always reports the best numbers. It is measuring buyers who had already decided, then taking credit for the decision.

The test is cheap: pause brand terms in one geography for four weeks and measure total store traffic, paid plus organic plus direct, against a matched control. Stores that run it usually find some of the brand spend was real and most of it was rent paid on demand the store already owned. The paid and organic comparison covers the attribution mechanics in more depth.

Why the Toyota store pays $121 a lead and the Honda store $26

The benchmark table's widest spread is not in click prices. Toyota dealers pay $2.27 a click and Honda dealers $1.82, a difference of 45 cents. The spread is in conversion: 1.81% against 7.05%. The Toyota store gets a comparable click and converts it a quarter as often, and its cost per lead is nearly five times higher as a result.

Some of that is the buyer. Toyota's demand runs to a small number of models with long waitlists in some markets, and a searcher checking availability is not filling out a form. Some of it is the page. A conversion rate is measured on the landing page, not in the ad account, and the two things that move it, whether the page loads with the unit's real price and photos and whether the lead form is reachable without fighting a widget, are website decisions rather than campaign decisions.

This is the part of a PPC program that is not a PPC problem. An agency can lower a cost per click by a few cents with better bidding. It cannot fix a 1.81% conversion rate from inside Google Ads, because the rate is being set on a page it does not control. When a store's cost per lead sits above its brand benchmark, the first place to look is the VDP, not the bid strategy.

What paid search cannot move

Two things, and both are worth saying plainly because the monthly report tends to imply otherwise.

It cannot move the organic result. Running ads does not raise a store's ranking; Google's own documentation states that advertising with Google has no effect on a site's presence in its search results and that it costs nothing to appear organically. The two results in the diagram above are produced by separate systems that share a landing page and nothing else. A store that reads a rising organic line as a return on ad spend is misattributing, and a store that expects the ads to carry organic once they stop is going to be surprised on the first of the month.

It cannot outlast the budget. Paid demand is rented by definition: the position exists while the campaign runs and disappears the day it stops. That is the product rather than a criticism of it. But it means the $123,698 line leaves nothing behind, where the SEO and website line next to it in the NADA table ($114,318) is the only advertising line that produces an asset the store still owns in year two. The channel-by-channel sort of owned against rented demand is the longer version of that point, and the budget benchmark shows where a store's own split sits against the NADA average.

Google Vehicle Ads and the feed the whole store depends on

Vehicle ads changed something about dealer PPC that the industry has been slow to say out loud: the ad now depends on the same data the rest of the store depends on. The feed that populates a vehicle ad is the feed that populates the website's inventory pages, the third-party listings and, through the structured data on each VDP, what a search engine or an AI assistant can read about the unit.

Google's own policy makes the dependency explicit. The price in the data source and the structured data must exactly match the price on the landing page. A store whose website vendor renders price through a script, or whose feed updates on a lag, is not just running a slower organic program. It is generating vehicle ad disapprovals and paying for clicks to pages that contradict the ad.

Which means the highest-return work on a vehicle ads program is often not in the ad account at all. It is feed hygiene, VDP indexation and structured data that carries the real year, trim, mileage and price. That is VDP and SRP work, and it improves the paid and organic result at the same time, because they are reading the same page.

How to read a car dealer PPC report

Four cuts of data the platform already holds, and a store should ask for each one by name.

Cost per lead by campaign type, brand separated from non-brand. A blended cost per lead is a brand campaign propping up everything else. The non-brand figure is the price of rented demand; the brand figure is mostly the price of measuring demand the store already had.

Impression share on non-brand inventory terms. This is the share of the market's actual shopping searches the store appeared for. It is the closest thing in the account to a market-share number, and it is rarely in the deck because it is rarely flattering.

Lead-to-sale by campaign, matched in the CRM. A lead count is not a sales count. The lead generation post covers why leads from different query classes close at different rates; the PPC report should show the close rate for each campaign, not the total.

An incrementality test at least once a year. A geographic holdout on brand terms, and if the store is brave, on non-brand. It is the only measurement that answers the question the report is supposed to answer, which is what the store bought that it would not otherwise have had.

What VulcanAX does and does not touch here

VulcanAX does not run PPC, paid social or media buying, and will say so rather than sell it badly. Nothing in this piece is an argument to cut the search line or to move it; those are decisions for the store and its ad agency, made against the store's own close rates.

What VulcanAX handles is the two things a PPC program depends on and cannot fix from inside the ad account: the page the paid click lands on, and the organic result that shows up beside the paid one whether or not the campaign is running. Indexed VDPs with structured data that matches the feed raise the paid conversion rate and the organic result at once. That is the shared dependency in the diagram, and it is the reason a firm that does not sell ads has an opinion about them.

FAQ